Digital twinborn simulation method combined with target recognition technology

Abstract

The invention discloses a digital twinborn simulation method combined with a target recognition technology. The method comprises the following steps: building a digital twinborn model of a physical workshop; establishing a target detection and recognition system, and dynamically extracting the unsteady state information of the physical workshop production line by adopting a target recognition algorithm; carrying out noise removal processing on the extracted information by using a quartile method; establishing a mapping relationship between the physical space and the virtual space; efficient transmission of data is realized by means of a port communication mechanism; constructing a data processing mode according to the operation logic of the physical workshop production line, and applying the data processing mode to the virtual model; and carrying out real-time simulation on the production of the physical workshop in combination with the rollback processing on the time. According to the invention, the target identification technology is successfully integrated into the digital twinborn model, the defect that the existing model lacks to effectively describe and respond to non-linear and sudden events in a physical scene is overcome, and the problem that the mirror images of the physical scene and the virtual scene are insufficient is overcome.

Description

Technical Field

[0001] The present invention relates to the field of simulation processing technology, and in particular to a digital twin simulation method combined with target recognition technology. Background Art

[0002] The global manufacturing industry is shifting from traditional automated manufacturing to intelligent manufacturing. With the integration of knowledge and algorithm training, more work is completed through intelligent technology. Its core technology is intelligent automation based on information-physical fusion systems, characterized by highly digitalized, networked, intelligent, and integrated production. The key technologies of intelligent automation include digital twins, which are based on the mapping of physical objects in digital space. Through the training of algorithms with collected actual data, self-learning obtains optimization solutions and realizes decision control. Digital twins make full use of data such as physical models, sensor updates, and operation history, integrate multi-disciplinary, multi-physical, multi-scale, and multi-probability simulation processes, complete mapping in virtual space, and thus reflect the full life cycle of the corresponding physical equipment. The simulation process of the simulation technology currently used is basically independent of the physical world, and lacks models that describe and reflect information such as production line behavior, rules, and constraints in physical workshops. Although the theoretical knowledge and technical research on applying digital twin technology to production workshop modeling has made significant progress through extensive research by experts and scholars, especially in the construction of simulation system framework and the exploration of all-factor information abstraction, the existing models still have the shortcomings of being unable to effectively characterize and respond to nonlinearities and emergencies in physical scenarios. There is still a lack of in-depth research and practical verification in this regard. Summary of the invention

[0003] In response to the technical problems mentioned in the background technology, the present invention proposes a method for integrating target recognition technology into a digital twin system. Through target detection technology, physical system information is extracted in real time and mapped to the virtual space, dynamically driving the update of the twin model, thereby enhancing the virtual system's simulation and response capabilities to sudden events in the physical system.

[0004] To achieve the above object, the present invention provides the following technical solution: a digital twin simulation method combined with target recognition technology, comprising the following steps:

[0005] S1. Build a twin model of the physical workshop;

[0006] S2. Build a target detection and recognition system to extract non-steady-state information from the physical workshop;

[0007] S3. Build a noise elimination system to eliminate false detection data generated by the target detection and recognition system;

[0008] S4, establish a mapping relationship between physical space and virtual space;

[0009] S5, through the port communication mechanism, transmits the valid data processed by S3 to the virtual space after being mapped by S4;

[0010] S6. Build data processing logic in the virtual space to drive the twin model, and combine it with time rollback processing to achieve real-time simulation of the physical workshop.

[0011] Furthermore, S1 includes:

[0012] The static data of the workshop production line is measured through physical measuring instruments. Based on the measured data, the physical workshop is three-dimensionally modeled, and then a twin model of the physical workshop is constructed.

[0013] Furthermore, S2 includes:

[0014] S2.1, design target detection algorithm;

[0015] S2.2. The image information collected from the workshop is trained through an algorithm to obtain a weight file. Combined with the weight file, the algorithm is used to track the workshop operation image in real time. While tracking the workshop image, the position parameters and quantity information target data are output.

[0016] Furthermore, S3 includes:

[0017] S3.1. Use the interquartile range method to obtain the fluctuation range of the normal value of the data, and its expression is:

[0018] IQR=Q3-Q1

[0019]

[0020] Where IQR represents the interquartile range; Q1 and Q3 represent the 25% and 75% quantiles in the data, respectively; B L and B U are the lower and upper limits of normal values, respectively;

[0021] S3.2. By using the fluctuation range of normal values, the data obtained by the target detection and recognition system that is beyond the normal range is eliminated.

[0022] Furthermore, the interval mapping method is adopted in S4, and the expression is:

[0023]

[0024] V virtual =V vL +N detect ·(V vU -V vL )

[0025] Where VdL and V dU Respectively represent the upper and lower bounds of the coordinate values ​​of the target object in the detection image; V d Indicates the real-time coordinate value of the target object in the detection image; N detect Indicates V d Normalized value; V vL and V vU Respectively represent the upper and lower bounds of the coordinate values ​​of the target object in the virtual space, V virtual Indicates the real-time coordinate value of the target object in the virtual space.

[0026] Furthermore, the time backoff processing expression in S6 is:

[0027] T response =T image_process +T transport +T data_process

[0028] T v =T n -T response

[0029] Where, T response It indicates the response time from when the camera reads the physical image information to when the virtual system reproduces the image information; T image_process T represents the processing time of the target detection and recognition system for the information read by the camera; transport Indicates the transit time of data transmitted through the port; T data_process Indicates the logical processing time of data; T n Indicates the current time; T v Represents the running time of the workshop production line in the virtual space.

[0030] Beneficial Effects

[0031] Compared with the prior art, the present invention extracts non-steady-state data of the workshop through the target recognition method and optimizes the digital twin system. The present invention first constructs a digital twin model of the physical workshop; then builds a target detection and recognition system to extract the non-steady-state information of the physical workshop production line in real time and dynamically; then uses the interquartile range method to de-noise the extracted data; then establishes a mapping relationship between the physical space and the virtual space; realizes efficient data transmission through the port communication mechanism; designs a data processing flow according to the operating logic of the physical workshop production line, and applies it to the virtual model; finally, with the cooperation of time rewind processing, the physical workshop production line is simulated in real time. The present invention integrates target recognition technology into the digital twin model, and develops a simulation model that can respond to the irregular dynamic changes of the physical scene in real time, overcoming the shortcomings of the existing model that cannot effectively portray and respond to nonlinear and emergencies in the physical scene, and solves the problem of incomplete mirroring of physical and virtual scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the process structure of the present invention;

[0033] Figure 2 This is an example diagram of the quartile method noise suppression; Figure 2 (a) is the effect diagram before noise reduction. Figure 2 (b) Effect diagram after noise reduction;

[0034] Figure 3 This is a diagram of the operation mechanism of the digital twin system. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that the embodiments in this application and the features in the embodiments can be combined with each other arbitrarily without conflict.

[0036] Example

[0037] In this embodiment, a digital twin simulation of the traditional Chinese medicine workshop is carried out, and a Figure 3 The digital twin operation system shown in the figure has the following specific steps: Figure 1 As shown:

[0038] S1. Use physical measuring instruments to measure the static data of the workshop production line, including equipment size, equipment space layout, etc.; use 3D modeling software such as SolidWorks and 3Dmax to perform 3D modeling of the physical workshop based on the measured data; finally, give the model material and texture, and import it into Unity to build a twin model of the physical workshop;

[0039] S2. Build a target detection and recognition system to extract non-steady-state information of the physical workshop, including the location information and quantity information of equipment and products;

[0040] S2.1. Design target detection algorithm. In this embodiment, EFMS_YOLOv8 algorithm is used to improve detection speed while maintaining high accuracy. At the same time, it is compared with other two algorithms. The results are shown in Table 1.

[0041] S2.2. The screen information taken by the workshop is trained through the algorithm to obtain the weight file; then, combined with the weight file, the algorithm is used to track the workshop operation screen in real time; while tracking the workshop screen, the target data such as position parameters and quantity information are output.

[0042] Table 1 Detection results of different algorithms for traditional Chinese medicine workshops

[0043]

[0044] As can be seen from Table 1, compared with other algorithms, EFMS_YOLO v 8 has advantages in both detection accuracy and FPS indicators, indicating that this algorithm is more suitable for promoting the accuracy and real-time performance of the twin model.

[0045] S3. Build a noise reduction system to remove a small amount of false positive data generated by the target detection algorithm through the interquartile range method, such as Figure 3 The before and after comparison of noise suppression shown;

[0046] S3.1. After the target detection and recognition system processes the operation screen information of the physical workshop for a period of time, the interquartile range method is used to obtain the fluctuation range of the normal value of the data, and its expression is:

[0047] IQR=Q3-Q1

[0048]

[0049] Where IQR represents the interquartile range; Q1 and Q3 represent the 25% and 75% quantiles in the data, respectively; B L and B U are the lower and upper bounds of normal values ​​respectively.

[0050] S3.2. By using the fluctuation range of normal values, the data obtained by the target detection and recognition system that is beyond the normal range is eliminated.

[0051] S4. Establishing the mapping relationship between physical space and virtual space, including interval mapping method and direct mapping method;

[0052] The interval mapping expression is:

[0053]

[0054] V virtual =V vL +N detect ·(V vU -V vL )

[0055] Where V dL and V dU Respectively represent the upper and lower bounds of the coordinate values ​​of the target object in the detection image; V d Indicates the real-time coordinate value of the target object in the detection image; N detect Indicates V d Normalized value; V vL and V vU Respectively represent the upper and lower bounds of the coordinate values ​​of the target object in the virtual space, V virtual Indicates the real-time coordinate value of the target object in the virtual space.

[0056] S5, through the port communication mechanism, transmits the valid data processed by S3 to the virtual space after being mapped by S4.

[0057] S6. Use C# language to build the data processing logic of the virtual space to drive the twin model, and combine it with time rollback processing to achieve real-time simulation of the physical workshop.

[0058] The time rollback processing expression is:

[0059] T response =T image_process +T transport +T data_process

[0060] T v =T n -T response

[0061] Where T response It indicates the response time from when the camera reads the physical image information to when the virtual system reproduces the image information; T image_process T represents the processing time of the target detection and recognition system for the information read by the camera; transport Indicates the transit time of data transmitted through the port; Tdata_process Indicates the logical processing time of data; T n Indicates the current time; T v Represents the running time of the workshop production line in the virtual space.

[0062] The specific implementation modes of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above implementation modes, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A digital twin simulation method combined with target recognition technology, characterized in that: The following steps are involved: S1. Build a twin model of the physical workshop; S2. Build a target detection and recognition system to extract non-steady-state information from the physical workshop; S3. Build a noise reduction system to eliminate false detection data generated by the target detection and recognition system; S4, establishing a mapping relationship between physical space and virtual space; S5, through the port communication mechanism, transmits the valid data processed by S3 to the virtual space after being mapped by S4; S6. Build data processing logic in the virtual space to drive the twin model, and combine it with time rollback processing to achieve real-time simulation of the physical workshop.

2. According to claim 1, a digital twin simulation method combined with target recognition technology is characterized in that: The S1 includes: The static data of the workshop production line is measured through physical measuring instruments. Based on the measured data, the physical workshop is three-dimensionally modeled, and then a twin model of the physical workshop is constructed.

3. The digital twin simulation method combined with target recognition technology according to claim 1 is characterized in that: The S2 includes: S2.1, design target detection algorithm; S2.

2. The image information collected from the workshop is trained through an algorithm to obtain a weight file. Combined with the weight file, the algorithm is used to track the workshop operation image in real time. While tracking the workshop image, the position parameters and quantity information target data are output.

4. The digital twin simulation method combined with target recognition technology according to claim 1, characterized in that: The S3 includes: S3.

1. Use the interquartile range method to obtain the fluctuation range of the normal value of the data, and the expression is: IQR=Q3-Q1 Where IQR represents the interquartile range; Q1 and Q3 represent the 25% and 75% quantiles in the data, respectively; B L and B U are the lower and upper limits of normal values, respectively; S3.

2. By using the fluctuation range of normal values, the data obtained by the target detection and recognition system that is beyond the normal range is eliminated.

5. According to the digital twin simulation method combined with target recognition technology according to claim 1, the interval mapping method is adopted in S4, and the expression is: V virtual =V vL +N detect ·(V vU -V vL ) Where V dL and V dU Respectively represent the upper and lower bounds of the coordinate values ​​of the target object in the detection image; V d Indicates the real-time coordinate value of the target object in the detection image; N detect Indicates V d Normalized value; V vL and V vU Respectively represent the upper and lower bounds of the coordinate values ​​of the target object in the virtual space, V virtual Indicates the real-time coordinate value of the target object in the virtual space.

6. According to the digital twin simulation method combined with target recognition technology in claim 1, the time backoff processing expression in S6 is: T response =T image_process +T transport +T data_process T v =T n -T response Where, T response It indicates the response time from when the camera reads the physical image information to when the virtual system reproduces the image information; T image_process T represents the processing time of the target detection and recognition system for the information read by the camera; transport Indicates the transit time of data transmitted through the port; T data_process Indicates the logical processing time of data; T n Indicates the current time; T v Represents the running time of the workshop production line in the virtual space.

Images