Target ball adaptive tracking system and method suitable for complex undulating environment

By introducing laser tracking, image acquisition and controller into the target ball testing system, combining time loop neural network and chaos mapping model, adaptive target ball tracking in complex ups and downs is achieved, and the measurement accuracy reduction and frequent manual correction are solved. The problem of improving measurement accuracy and prediction capabilities.

CN118913106BActive Publication Date: 2024-12-06SICHUAN LAMBDA TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411400041.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-10-09
Publication Date
2024-12-06
Estimated Expiration
2044-10-09

AI Technical Summary

Technical Problem

In complex ups and downs, the self-positioning of the target ball testing system is affected, resulting in a reduction in measurement accuracy. Frequent manual light drawing is required to correct, resulting in large errors in tracking data.

Method used

A target ball adaptive tracking system including a laser tracking mechanism, an image acquisition mechanism and a controller is designed. The target position information is collected in real time through the image acquisition mechanism, and the controller compares real-time and theoretical position information, and uses the time loop neural network model and chaos. The mapping model is used to predict and the target ball mechanism is remotely controlled to achieve adaptive light response.

Benefits of technology

It improves measurement accuracy, reduces the working intensity of artificial light-induced light, can adjust the orientation of the target ball from three dimensions, and improves the global retrieval ability and prediction accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118913106B_ABST
    Figure CN118913106B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of testing technology, and in particular to a target ball adaptive tracking system and method suitable for complex undulating environments, comprising a laser tracking mechanism, an image acquisition mechanism, a controller and a target ball mechanism. The target ball adaptive tracking system suitable for complex undulating environments is provided by setting the laser tracking mechanism, the image acquisition mechanism, the controller and the target ball mechanism, wherein the controller can adjust the steering of the target ball mechanism based on the real-time position of the target to be tracked acquired by the image acquisition mechanism, so that the target ball mechanism can be adaptively aligned with the laser tracking mechanism. On the one hand, the overall measurement accuracy can be improved, and on the other hand, the adaptive alignment method can greatly reduce the workload of artificial light guidance.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of testing technology, and in particular to a target ball adaptive tracking system and method suitable for complex undulating environments. Background Art

[0002] The target ball test system mainly includes a target ball and a laser tester. The target ball is used to find the center point of the target to be tracked. During measurement, the laser emitted by the tracking head of the laser tester hits the reflective part of the target ball. The target ball acts as a reflector to reflect the laser back to the tracking head of the laser tester. When the target to be tracked moves, manual light guidance operation is required.

[0003] Conventional target ball test systems are mainly used in closed indoor environments to test or measure some industrial robots. In larger application scenarios, they can be used in position tracking or position tracking tests of equipment such as drones and unmanned boats.

[0004] Taking the unmanned boat tracking test as an example, when the test is carried out outdoors, the data acquisition end is usually set on a fixed part to collect data, forming a fixed and mobile bidirectional model. In addition to the pre-set designated navigation track, it will also be affected by water currents and waves. The water currents and waves continue to act on the unmanned boat, making the unmanned boat in a complex undulating environment. This complex undulating environment will affect the self-positioning of the target ball test system, and frequently trigger manual light guidance, otherwise it will cause large errors in other tracking data. This complex undulating environment has short-term regularity and will not change frequently. Summary of the invention

[0005] In view of the above problems, the present invention provides a target ball adaptive tracking system and method suitable for complex undulating environments.

[0006] The technical solution adopted is: a target ball adaptive tracking system suitable for complex undulating environments, including a laser tracking mechanism, an image acquisition mechanism, a controller and a target ball mechanism;

[0007] The laser tracking mechanism can track the position of the target ball mechanism, and the laser tracking mechanism is installed on the fixed rod;

[0008] The target ball mechanism is installed on the target to be tracked, and the target to be tracked is in a complex undulating environment;

[0009] The image acquisition mechanism can acquire the real-time position information of the target to be tracked during this acquisition cycle, and can transmit the real-time position information to the controller, and the image acquisition mechanism is installed on the fixed rod;

[0010] The controller compares the real-time position information and theoretical position information of the target to be tracked in a complex undulating environment and obtains the estimated position information of the target to be tracked in the next acquisition cycle through a target ball adaptive tracking method, and remotely controls the rotation of the target ball mechanism.

[0011] Optionally, the target ball mechanism includes a target ball, a containing sleeve, a first direction adjustment component, a second direction adjustment component, a third direction adjustment component and a fixing seat;

[0012] The target ball is arranged in the containing sleeve;

[0013] The first direction adjustment component is arranged on the accommodation sleeve, and the first direction adjustment component can adjust the position of the target ball in the X-axis direction;

[0014] The second direction adjustment component is used to connect the accommodating sleeve and the third direction adjustment component, and the second direction adjustment component can adjust the position of the target ball in the Y-axis direction;

[0015] The third direction adjustment component is used to connect the second direction adjustment component and the fixing seat, and the third direction adjustment component can adjust the position of the target ball in the Z-axis direction;

[0016] The fixing seat is used to be connected with the target to be tracked.

[0017] Optionally, a receiving cavity is provided in the receiving sleeve and an opening is provided at one end of the receiving sleeve;

[0018] A bottom plate is provided at the bottom of the accommodating cavity;

[0019] The target ball can be placed in the accommodating cavity.

[0020] Optionally, the first direction adjustment component includes a first motor group, a first rotating shaft, a first gear and an adjustment part in the accommodating cavity;

[0021] The first motor group is arranged on the top of the housing and is connected to the controller signal wirelessly;

[0022] One end of the first rotating shaft is connected to the power output end of the first motor group;

[0023] The first gear is sleeved on the first rotating shaft and passes through the top of the accommodating sleeve to mesh with the adjusting portion in the accommodating cavity.

[0024] Optionally, the adjusting part in the accommodating cavity includes a first screw rod, a second screw rod, a screw sleeve, a connecting seat and a limiting rod;

[0025] The target ball is installed on the connecting seat, and the target ball faces away from the accommodating cavity;

[0026] One end of the first screw is connected to the connecting seat, and the other end of the first screw is connected to one end of the screw sleeve;

[0027] One end of the second screw is fixed to the bottom plate, the other end of the second screw is connected to the other end of the screw sleeve, and the outer peripheral threads of the first screw and the second screw rotate in opposite directions;

[0028] The screw sleeve is a hollow tubular structure, and a convex tooth is arranged on the outer periphery of the screw sleeve, and the convex tooth can mesh with the first gear, and there is a gap between the first screw rod located in the screw sleeve and the second screw rod located in the screw sleeve;

[0029] One end of the limiting rod is connected to the connecting seat, and the other end of the limiting rod passes through the bottom plate.

[0030] Optionally, the second direction adjustment component includes an arc seat, a connecting sleeve and a connecting base;

[0031] The arc seat is installed at the bottom of the accommodating sleeve and is movably connected with the connecting sleeve. An arc tooth plate is arranged in the arc seat.

[0032] One end of the connecting sleeve is connected to the arc seat, and the other end of the connecting sleeve is connected to the connecting base, and a toothed truncated cone and a second rotating shaft are arranged in the connecting sleeve, and the toothed truncated cone can mesh with the arc-shaped tooth plate, and one end of the second rotating shaft is connected to the bottom of the toothed truncated cone;

[0033] A second motor group is arranged on the connection base, and a power output end of the second motor group is connected to the second rotating shaft, and the second motor group is connected to the controller signal in a wireless manner.

[0034] Optionally, the third direction adjustment component includes a third motor group and a support platform, the second direction adjustment component is arranged on the support platform, the third motor group is arranged on a fixed seat, and the power output end of the third motor group is connected to the support platform, and the third motor group is connected to the controller signal via a wireless manner.

[0035] The present application also provides a target ball adaptive tracking method applicable to complex undulating environments, which is implemented based on a target ball adaptive tracking system applicable to complex undulating environments, and includes the following steps:

[0036] S1. Initialize data and establish the initial coordinate values ​​of the target to be tracked on the XYZ axis;

[0037] S2. The laser tracking mechanism is aligned with the target ball mechanism on the target to be tracked, the controller controls the target to be tracked to move according to the specified coordinate data set, and the image acquisition mechanism acquires the actual position of the target to be tracked according to the specified period to obtain the actual coordinate data set;

[0038] S3. Compare the actual coordinate data set and the specified coordinate data set to obtain a difference coordinate data set, and divide the difference coordinate data set into a training data set and a test data set;

[0039] S4. Establish a time recurrent neural network model based on the training data set, and optimize the hyperparameters of the time recurrent neural network model through a chaotic mapping model;

[0040] S5. Based on the inter-cyclic neural network model with optimized hyperparameters, the feature data of the test data set is substituted to obtain the predicted coordinate data set of the next cycle;

[0041] S6. The controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set so that the target ball turns and aligns with the laser tracking mechanism.

[0042] The present application also provides a target ball adaptive tracking method applicable to complex undulating environments, which is implemented based on a target ball adaptive tracking system applicable to complex undulating environments, and includes the following steps:

[0043] S1. Initialize data and establish the initial coordinate values ​​of the target to be tracked on the XYZ axis;

[0044] S2. The laser tracking mechanism is aligned with the target ball mechanism on the target to be tracked, and the image acquisition mechanism collects the actual position of the target to be tracked according to the specified period to obtain the actual coordinate data set;

[0045] S3. Compare the actual coordinate data set with the initial coordinate value to obtain a difference coordinate data set, and divide the difference coordinate data set into a training data set and a test data set;

[0046] S4. Establish a time recurrent neural network model based on the training data set, and optimize the hyperparameters of the time recurrent neural network model through a chaotic mapping model;

[0047] S5. Based on the inter-cyclic neural network model with optimized hyperparameters, the feature data of the test data set is substituted to obtain the predicted coordinate data set of the next cycle;

[0048] S6. The controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set so that the target ball turns and aligns with the laser tracking mechanism.

[0049] The beneficial effects of the present invention include at least one of the following:

[0050] 1. An adaptive target ball tracking system suitable for complex undulating environments is set up, which includes a laser tracking mechanism, an image acquisition mechanism, a controller and a target ball mechanism. The controller can adjust the steering of the target ball mechanism based on the real-time position of the target to be tracked acquired by the image acquisition mechanism, so that it can adaptively align with the laser tracking mechanism. On the one hand, the overall measurement accuracy can be improved. On the other hand, the use of adaptive alignment can greatly reduce the workload of artificial light guidance.

[0051] 2. A first direction adjustment component, a second direction adjustment component and a third direction adjustment component are provided on the target ball mechanism, so that the direction of the target ball can be adjusted from three dimensions.

[0052] 3. A corresponding target ball adaptive tracking method is provided, and the chaotic mapping model is used to optimize the time recurrent neural network model to improve the global retrieval capability and prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of a target ball adaptive tracking system suitable for complex undulating environments;

[0054] Figure 2 It is another schematic diagram of the target ball adaptive tracking system suitable for complex undulating environments;

[0055] Figure 3 It is a schematic diagram of the target ball mechanism structure;

[0056] Figure 4 is a schematic diagram of the internal structure of the housing;

[0057] Figure 5 is a schematic diagram of the structure of the target ball;

[0058] Figure 6 It is a schematic diagram of the structures of the first direction adjustment component, the second direction adjustment component and the third direction adjustment component;

[0059] Figure 7 is a schematic diagram of the structure of the second direction adjustment component;

[0060] Figure 8 Schematic diagram of the target ball in the XYZ axis coordinates.

[0061] The reference numerals are:

[0062] 1 is a containing sleeve, 2 is a containing sleeve port, 3 is a target ball, 5 is an arc seat, 6 is a connecting sleeve, 7 is a connecting base, 8 is a supporting platform, 9 is a third motor group, 10 is a fixing seat, 11 is a first motor group, 12 is a limiting rod, 13 is a first gear, 14 is a first rotating shaft, 15 is a second motor group, 16 is a supporting leg, 17 is a second rotating shaft, 18 is a toothed frustum, 19 is an arc-shaped toothed plate, 20 is a connecting seat, 21 is a containing cavity, 22 is a bottom plate, 23 is a screw sleeve, 24 is a convex tooth, 25 is a first screw rod, 26 is a second screw rod, 27 is a fixing rod, 28 is an image acquisition mechanism, and 29 is a laser tracking mechanism. DETAILED DESCRIPTION

[0063] The following describes the implementation of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific implementations, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0064] It should be noted that the illustrations provided in the following embodiments are only used to schematically illustrate the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0065] like Figure 1 and Figure 2 As shown, a target ball adaptive tracking system suitable for complex undulating environments includes a laser tracking mechanism 29, an image acquisition mechanism 28, a controller and a target ball mechanism;

[0066] The laser tracking mechanism 29 can track the position of the target ball mechanism, and the laser tracking mechanism 29 is installed on the fixed rod 27;

[0067] The target ball mechanism is installed on the target to be tracked, and the target to be tracked is in a complex undulating environment;

[0068] The image acquisition mechanism 28 can acquire the real-time position information of the target to be tracked during this acquisition cycle, and can transmit the real-time position information to the controller. The image acquisition mechanism 28 is installed on the fixed rod 27;

[0069] The controller compares the real-time position information and theoretical position information of the target to be tracked in a complex undulating environment and obtains the estimated position information of the target to be tracked in the next acquisition cycle through a target ball adaptive tracking method, and remotely controls the rotation of the target ball mechanism.

[0070] The purpose of such a design is to provide a target ball adaptive tracking system suitable for complex and undulating environments, which includes a laser tracking mechanism, an image acquisition mechanism, a controller and a target ball mechanism. The controller can adjust the steering of the target ball mechanism based on the real-time position of the target to be tracked acquired by the image acquisition mechanism, so that it can adaptively align with the laser tracking mechanism. On the one hand, it can improve the overall measurement accuracy. On the other hand, the use of adaptive alignment can greatly reduce the workload of artificial light guidance.

[0071] It should be pointed out that Figure 1In the figure, a target ball adaptive tracking system suitable for complex undulating environments is applied in an unmanned boat scene, wherein the target ball mechanism is fixed on the unmanned boat through supporting legs 16, and an image acquisition mechanism 28 and a laser tracking mechanism 29 respectively track and acquire images of the unmanned boat and the target ball mechanism in real time, wherein the complex undulating environment is mainly affected by waves and water currents.

[0072] Figure 2 The target ball adaptive tracking system suitable for complex undulating environments is applied in UAV scenarios, wherein the target ball mechanism is directly fixed on the lower part of the UAV, and the image acquisition mechanism 28 and the laser tracking mechanism 29 respectively track and acquire images of the UAV and the target ball mechanism in real time, wherein the complex undulating environment is mainly affected by airflow.

[0073] In this embodiment, Figure 3 As shown, the target ball mechanism includes a target ball 3, a receiving sleeve 1, a first direction adjustment component, a second direction adjustment component, a third direction adjustment component and a fixing seat 10;

[0074] The target ball 3 is arranged in the containing sleeve 1;

[0075] The first direction adjustment component is disposed on the accommodating sleeve 1, and the first direction adjustment component can adjust the position of the target ball 3 in the X-axis direction;

[0076] The second direction adjustment component is used to connect the accommodating sleeve 1 and the third direction adjustment component, and the second direction adjustment component can adjust the position of the target ball 3 in the Y-axis direction;

[0077] The third direction adjustment component is used to connect the second direction adjustment component and the fixing seat 10, and the third direction adjustment component can adjust the position of the target ball 3 in the Z-axis direction;

[0078] The fixing base 10 is used to be connected to the target to be tracked.

[0079] The purpose of such a design is to provide the target ball mechanism with a first direction adjustment component, a second direction adjustment component and a third direction adjustment component, so that the direction of the target ball can be adjusted from three dimensions.

[0080] In this embodiment, Figure 4 and 5 As shown, a receiving cavity 21 is provided in the receiving sleeve 1 and an opening is provided at one end of the receiving sleeve 1;

[0081] A bottom plate 22 is provided at the bottom of the accommodating cavity 21;

[0082] The target ball 3 can be placed in the accommodating cavity 21 .

[0083] Meanwhile, the first direction adjustment component includes a first motor group 11, a first rotating shaft 14, a first gear 13 and an adjustment part in the accommodating cavity;

[0084] The first motor group 11 is disposed on the top of the housing 1 and is connected to the controller signal wirelessly;

[0085] One end of the first rotating shaft 14 is connected to the power output end of the first motor group 11;

[0086] The first gear 13 is sleeved on the first rotating shaft 14 and passes through the top of the accommodating sleeve 1 to engage with the adjusting part in the accommodating cavity. The adjusting part in the accommodating cavity includes a first screw rod 25, a second screw rod 26, a screw sleeve 23, a connecting seat 20 and a limiting rod 12;

[0087] The target ball 3 is mounted on the connecting seat 20, and the target ball 3 faces away from the accommodating cavity 21;

[0088] One end of the first screw rod 25 is connected to the connecting seat 20, and the other end of the first screw rod 25 is connected to one end of the screw sleeve 23;

[0089] One end of the second screw 26 is fixed on the bottom plate 22, and the other end of the second screw 26 is connected to the other end of the screw sleeve 23, and the outer peripheral threads of the first screw 25 and the second screw 26 rotate in opposite directions;

[0090] The screw sleeve 23 is a hollow tubular structure, and a convex tooth 24 is arranged on the outer periphery of the screw sleeve 23, and the convex tooth 24 can mesh with the first gear 13, and there is a gap between the first screw rod 25 located in the screw sleeve 23 and the second screw rod 26 located in the screw sleeve 23;

[0091] One end of the limiting rod 12 is connected to the connecting seat 20 , and the other end of the limiting rod 12 passes through the bottom plate 22 .

[0092] In specific use, when it is necessary to adjust the position of the target ball in the X-axis direction, the controller remotely controls the first motor group to drive the first rotating shaft to rotate, thereby driving the meshing convex tooth to rotate through the first gear. Since the first screw, the second screw and the screw sleeve are connected by threads, and the first screw and the second screw use opposite threads, and two matching threads are also used at the connection between the screw sleeve and the first screw and the second screw, it can be ensured that when the convex tooth rotates, it drives the connecting seat toward or away from the port 2 of the accommodating sleeve.

[0093] At the same time, a limiting rod is provided, and the limiting rod passes through the bottom plate, so that under the restriction of the limiting rod and the bottom plate, the connecting seat will not rotate together.

[0094] In this embodiment, Figure 6 and Figure 7 As shown, the second direction adjustment component includes an arc seat 5, a connecting sleeve 6 and a connecting base 7;

[0095] The arc seat 5 is installed at the bottom of the accommodating sleeve 1 and is movably connected to the connecting sleeve 6. An arc tooth plate 19 is arranged in the arc seat 5.

[0096] One end of the connecting sleeve 6 is connected to the arc seat 5, and the other end of the connecting sleeve 6 is connected to the connecting base 7. A toothed truncated cone 18 and a second rotating shaft 17 are arranged in the connecting sleeve 6. The toothed truncated cone 18 can mesh with the arc tooth plate 19, and one end of the second rotating shaft 17 is connected to the bottom of the toothed truncated cone 18.

[0097] The connecting base 7 is provided with a second motor group 15, and a power output end of the second motor group 15 is connected to a second rotating shaft 17, and the second motor group 15 is connected to a controller signal in a wireless manner.

[0098] The purpose of this design is that when the position of the target ball in the Y-axis direction needs to be adjusted, the controller remotely controls the second motor group so that the second motor group drives the second rotating shaft to rotate, and at the same time the toothed cone at the end of the second rotating shaft rotates together, so that the arc-shaped toothed plate meshing with it rotates at a certain angle.

[0099] It should be pointed out that the connecting sleeve and the arc seat are movably connected, and the arc seat can rotate on the connecting sleeve at an angle of 0°-30°.

[0100] In this embodiment, Figure 6 and Figure 8 As shown, the third direction adjustment component includes a third motor group 9 and a support platform 8, the second direction adjustment component is arranged on the support platform 8, the third motor group 9 is arranged on a fixed seat 10, and the power output end of the third motor group 9 is connected to the support platform 8, and the third motor group 9 is connected to the controller signal by wireless means.

[0101] The purpose of this design is that when the position of the target ball in the Z-axis direction needs to be adjusted, the controller remotely controls the third motor group to drive the support platform to rotate, thereby causing the entire support platform to be displaced in the Z-axis.

[0102] It should be pointed out that in order to facilitate more precise adjustment of the position of the XYZ axes, the first motor group, the second motor group and the third motor group use stepper motors, and the specific rotation angle and size are controlled by gears or tooth grooves. In addition to the motors, the three motor groups also contain necessary circuits, power supply components and wireless transmission modules.

[0103] In this embodiment, a target ball adaptive tracking method suitable for complex undulating environments is also provided, comprising the following steps:

[0104] S1. Initialize data and establish the initial coordinate values ​​of the target to be tracked on the XYZ axis;

[0105] S2. The laser tracking mechanism 29 is aligned with the target ball mechanism on the target to be tracked, the controller controls the target to be tracked to move according to the specified coordinate data set, and the image acquisition mechanism 28 acquires the actual position of the target to be tracked according to the specified period to obtain the actual coordinate data set;

[0106] S3. Compare the actual coordinate data set and the specified coordinate data set to obtain a difference coordinate data set, and divide the difference coordinate data set into a training data set and a test data set;

[0107] S4. Establish a time recurrent neural network model based on the training data set, and optimize the hyperparameters of the time recurrent neural network model through a chaotic mapping model;

[0108] S5. Based on the inter-cyclic neural network model with optimized hyperparameters, the feature data of the test data set is substituted to obtain the predicted coordinate data set of the next cycle;

[0109] S6. The controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set so that the target ball turns and aligns with the laser tracking mechanism 29.

[0110] It should be pointed out that this method is mainly applicable to the directional cruising scenario of the target to be tracked, that is, in addition to being affected by the external environment, the target to be tracked itself will also move.

[0111] In a specific implementation, this embodiment provides a specific achievable method. In S2, the abnormal data that may exist in the actual coordinate data set is first processed to remove unreasonable data or outlier noise points, and then the coordinate data in the actual coordinate data set is normalized. The calculation formula can be:

[0112] ;

[0113] in, is the value after data normalization, is the data value before data normalization, is the minimum value in the sample data, is the maximum value in the sample data.

[0114] In S3, the training data set and the test data set include the training data set trainX, trainY, trainZ and the test data set testX, testY, testZ respectively.

[0115] In S4, a time recurrent neural network model is established, which includes an LSTM layer, a dropout layer, a fully connected layer, and an output layer. At the same time, the model contains 6 hyperparameters. The chaos mapping model is used to optimize the hyperparameters of the time recurrent neural network model, and the upper bound of the model hyperparameter vector structure search space is set. and the lower bound , number of samples , variable dimension .

[0116] in ;

[0117] Indicates the maximum value of the first hyperparameter sliding window size look_back, Respectively represent the maximum number of hidden units of the first to third layers of LSTM, Indicates the maximum value of the discard rate, Indicates the maximum value of the training batch size batch_size; Indicates the minimum value of the first hyperparameter sliding window size look_back, Respectively represent the minimum number of hidden units of the first to third layers of LSTM, Indicates the minimum value of the discard rate, Indicates the minimum value of the training batch size batch_size.

[0118] Randomly generate a Dimensional vector , each component of the vector is between 0 and 1;

[0119] Will Each component of is mapped to the value interval of the variable, the formula is as follows:

[0120] ;

[0121] In S5, set the maximum number of iterations Maxlter, and then take the model hyperparameters obtained in S4. is the sliding window size, is the number of hidden units of the 3-layer LSTM, is the dropout rate of the Dropout layer, is the training batch size batch_size.

[0122] Based on the above results, trainX, trainY, and trainZ are trained, and the trained model is used to predict testX, testY, and testZ to generate pred;

[0123] According to the prediction generated pred and The mean square error is the fitness of the initialized sample. The mean square error calculation formula is as follows:

[0124] ;

[0125] The minimum fitness value is recorded as GbestScore, the index of the minimum fitness value is GbestPositon, and the maximum number of iterations MaxIter and iteration calculator t are set. When the iteration calculator t is greater than the maximum number of iterations MaxIter, S5 obtains the optimal hyperparameters, rebuilds the optimal LSTM model, and , and Carry out model training, Make a prediction to generate testPredict, convert the prediction result of the test set testPredict into the range of the original data, and calculate The mean square error with testPredict, The mean square error of testPredict is calculated to obtain the influence error of airflow or water flow in the XYZ axis direction. At the same time, based on the obtained mean square error, the influence of airflow or water flow in the XYZ axis direction in the next acquisition cycle can be predicted.

[0126] In S6, the impact value of the airflow or water flow in the XYZ axis direction in the next acquisition cycle is predicted, so that the controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set to make the target ball turn and align with the laser tracking mechanism 29.

[0127] This provides a suitable target ball adaptive tracking method, uses a chaotic mapping model to optimize the time recurrent neural network model, and improves the global retrieval capability and prediction accuracy.

[0128] In this embodiment, another target ball adaptive tracking method applicable to complex undulating environments is provided, comprising the following steps:

[0129] S1. Initialize data and establish the initial coordinate values ​​of the target to be tracked on the XYZ axis;

[0130] S2. The laser tracking mechanism 29 is aligned with the target ball mechanism on the target to be tracked, and the image acquisition mechanism 28 acquires the actual position of the target to be tracked according to the specified period to obtain the actual coordinate data set;

[0131] S3. Compare the actual coordinate data set with the initial coordinate value to obtain a difference coordinate data set, and divide the difference coordinate data set into a training data set and a test data set;

[0132] S4. Establish a time recurrent neural network model based on the training data set, and optimize the hyperparameters of the time recurrent neural network model through a chaotic mapping model;

[0133] S5. Based on the inter-cyclic neural network model with optimized hyperparameters, the feature data of the test data set is substituted to obtain the predicted coordinate data set of the next cycle;

[0134] S6. The controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set so that the target ball turns and aligns with the laser tracking mechanism 29.

[0135] It should be pointed out that this method is mainly applicable to the test of the target to be tracked hovering or stagnating, that is, no additional location is set except for the influence of the partial environment.

[0136] In a specific implementation, this embodiment provides a specific achievable method. In S2, the abnormal data that may exist in the actual coordinate data set is first processed to remove unreasonable data or outlier noise points, and then the coordinate data in the actual coordinate data set is normalized. The calculation formula can be:

[0137] ;

[0138] in, is the value after data normalization, is the data value before data normalization, is the minimum value in the sample data, is the maximum value in the sample data.

[0139] In S3, the training data set and the test data set include the training data set trainX, trainY, trainZ and the test data set testX, testY, testZ respectively.

[0140] In S4, a time recurrent neural network model is established, which includes an LSTM layer, a dropout layer, a fully connected layer, and an output layer. At the same time, the model contains 6 hyperparameters. The chaos mapping model is used to optimize the hyperparameters of the time recurrent neural network model, and the upper bound of the model hyperparameter vector structure search space is set. and the lower bound , number of samples , variable dimension .

[0141] in ;

[0142] Indicates the maximum value of the first hyperparameter sliding window size look_back, Respectively represent the maximum number of hidden units of the first to third layers of LSTM, Indicates the maximum value of the discard rate, Indicates the maximum value of the training batch size batch_size; Indicates the minimum value of the first hyperparameter sliding window size look_back, Respectively represent the minimum number of hidden units of the first to third layers of LSTM, Indicates the minimum value of the discard rate, Indicates the minimum value of the training batch size batch_size.

[0143] Randomly generate a Dimensional vector , each component of the vector is between 0 and 1;

[0144] Will Each component of is mapped to the value interval of the variable, the formula is as follows:

[0145] ;

[0146] In S5, set the maximum number of iterations Maxlter, and then take the model hyperparameters obtained in S4. is the sliding window size, is the number of hidden units of the 3-layer LSTM, is the dropout rate of the Dropout layer, is the training batch size batch_size.

[0147] Based on the above results, trainX, trainY, and trainZ are trained, and the trained model is used to predict testX, testY, and testZ to generate pred;

[0148] According to the prediction generated pred and The mean square error is the fitness of the initialized sample. The mean square error calculation formula is as follows:

[0149] ;

[0150] The minimum fitness value is recorded as GbestScore, the index of the minimum fitness value is GbestPositon, and the maximum number of iterations MaxIter and iteration calculator t are set. When the iteration calculator t is greater than the maximum number of iterations MaxIter, S5 obtains the optimal hyperparameters, rebuilds the optimal LSTM model, and , and Carry out model training, Make a prediction to generate testPredict, convert the prediction result of the test set testPredict into the range of the original data, and calculate The mean square error with testPredict, The mean square error of testPredict is calculated to obtain the influence error of airflow or water flow in the XYZ axis direction. At the same time, based on the obtained mean square error, the influence of airflow or water flow in the XYZ axis direction in the next acquisition cycle can be predicted.

[0151] In S6, the impact value of the airflow or water flow in the XYZ axis direction in the next acquisition cycle is predicted, so that the controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set to make the target ball turn and align with the laser tracking mechanism 29.

[0152] This provides a suitable target ball adaptive tracking method, uses a chaotic mapping model to optimize the time recurrent neural network model, and improves the global retrieval capability and prediction accuracy.

[0153] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The target ball adaptive tracking system is suitable for complex undulating environments, characterized by: It includes a laser tracking mechanism (29), an image acquisition mechanism (28), a controller and a target ball mechanism; The laser tracking mechanism (29) can track the position of the target ball mechanism, and the laser tracking mechanism (29) is installed on the fixing rod (27); The target ball mechanism is installed on the target to be tracked, and the target to be tracked is in a complex undulating environment; The image acquisition mechanism (28) can acquire real-time position information of the target to be tracked during this acquisition cycle, and can transmit the real-time position information to the controller, and the image acquisition mechanism (28) is mounted on the fixed rod (27); The controller compares the real-time position information and theoretical position information of the target to be tracked in a complex undulating environment and obtains the estimated position information of the target to be tracked in the next acquisition cycle through a target ball adaptive tracking method, and remotely controls the rotation of the target ball mechanism; The target ball mechanism comprises a target ball (3), a containing sleeve (1), a first direction adjustment component, a second direction adjustment component, a third direction adjustment component and a fixing seat (10); The target ball (3) is arranged in the containing sleeve (1); The first direction adjustment component is arranged on the containing sleeve (1), and the first direction adjustment component can adjust the position of the target ball (3) in the X-axis direction; The second direction adjustment component is used to connect the containing sleeve (1) and the third direction adjustment component, and the second direction adjustment component can adjust the position of the target ball (3) in the Y-axis direction; The third direction adjustment component is used to connect the second direction adjustment component and the fixing seat (10), and the third direction adjustment component can adjust the position of the target ball (3) in the Z-axis direction; The fixing seat (10) is used to connect with the target to be tracked; The accommodating sleeve (1) is provided with an accommodating cavity (21) and one end of the accommodating sleeve (1) is provided with an opening; A bottom plate (22) is provided at the bottom of the accommodating cavity (21); The target ball (3) can be placed in the accommodating cavity (21); The first direction adjustment component comprises a first motor group (11), a first rotating shaft (14), a first gear (13) and an adjustment portion in the accommodating cavity; The first motor group (11) is arranged on the top of the housing (1) and is connected to the controller signal by wireless means; One end of the first rotating shaft (14) is connected to the power output end of the first motor group (11); The first gear (13) is sleeved on the first rotating shaft (14) and passes through the top of the accommodating sleeve (1) to mesh with the adjusting portion in the accommodating cavity; The adjusting portion in the accommodating cavity comprises a first screw rod (25), a second screw rod (26), a screw sleeve (23), a connecting seat (20) and a limiting rod (12); The target ball (3) is mounted on the connecting seat (20), with the target ball (3) facing away from the accommodating cavity (21); One end of the first screw rod (25) is connected to the connecting seat (20), and the other end of the first screw rod (25) is connected to one end of the screw sleeve (23); One end of the second screw rod (26) is fixed on the bottom plate (22), the other end of the second screw rod (26) is connected to the other end of the screw sleeve (23), and the outer peripheral threads of the first screw rod (25) and the second screw rod (26) rotate in opposite directions; The screw sleeve (23) is a hollow tubular structure, and a convex tooth (24) is arranged on the outer periphery of the screw sleeve (23), and the convex tooth (24) can mesh with the first gear (13), and there is a gap between the first screw rod (25) located in the screw sleeve (23) and the second screw rod (26) located in the screw sleeve (23); One end of the limiting rod (12) is connected to the connecting seat (20), and the other end of the limiting rod (12) passes through the bottom plate (22); The second direction adjustment component comprises an arc-shaped seat (5), a connecting sleeve (6) and a connecting base (7); The arc seat (5) is installed at the bottom of the accommodating sleeve (1) and is movably connected to the connecting sleeve (6). An arc tooth plate (19) is arranged inside the arc seat (5); One end of the connecting sleeve (6) is connected to the arc seat (5), and the other end of the connecting sleeve (6) is connected to the connecting base (7). A toothed truncated cone (18) and a second rotating shaft (17) are arranged inside the connecting sleeve (6). The toothed truncated cone (18) can mesh with the arc tooth plate (19), and one end of the second rotating shaft (17) is connected to the bottom of the toothed truncated cone (18); A second motor group (15) is disposed on the connection base (7), and a power output end of the second motor group (15) is connected to the second rotating shaft (17), and the second motor group (15) is connected to the controller signal in a wireless manner.

2. The target ball adaptive tracking system suitable for complex undulating environments according to claim 1, characterized in that: The third direction adjustment component comprises a third motor group (9) and a support platform (8), the second direction adjustment component is arranged on the support platform (8), the third motor group (9) is arranged on a fixing seat (10), and a power output end of the third motor group (9) is connected to the support platform (8), and the third motor group (9) is connected to a controller signal in a wireless manner.

3. A target ball adaptive tracking method suitable for complex undulating environments, which is implemented based on the target ball adaptive tracking system suitable for complex undulating environments according to claim 1, characterized in that: The following steps are involved: S1. Initialize data and establish the initial coordinate values ​​of the target to be tracked on the XYZ axis; S2. The laser tracking mechanism (29) is aligned with the target ball mechanism on the target to be tracked, the controller controls the target to be tracked to move according to the specified coordinate data set, and the image acquisition mechanism (28) acquires the actual position of the target to be tracked according to the specified period to obtain the actual coordinate data set; S3. Compare the actual coordinate data set and the specified coordinate data set to obtain a difference coordinate data set, and divide the difference coordinate data set into a training data set and a test data set; S4. Establish a time recurrent neural network model based on the training data set, and optimize the hyperparameters of the time recurrent neural network model through a chaotic mapping model; S5. Based on the inter-cyclic neural network model with optimized hyperparameters, the feature data of the test data set is substituted to obtain the predicted coordinate data set of the next cycle; S6. The controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set so that the target ball is turned and aligned with the laser tracking mechanism (29).

4. A target ball adaptive tracking method suitable for complex undulating environments, characterized in that: The target ball adaptive tracking system suitable for complex undulating environments according to claim 1 is implemented, comprising the following steps: S1. Initialize data and establish the initial coordinate values ​​of the target to be tracked on the XYZ axis; S2. The laser tracking mechanism (29) is aligned with the target ball mechanism on the target to be tracked, and the image acquisition mechanism (28) acquires the actual position of the target to be tracked at a specified period to obtain an actual coordinate data set; S3. Compare the actual coordinate data set with the initial coordinate value to obtain a difference coordinate data set, and divide the difference coordinate data set into a training data set and a test data set; S4. Establish a time recurrent neural network model based on the training data set, and optimize the hyperparameters of the time recurrent neural network model through a chaotic mapping model; S5. Based on the inter-cyclic neural network model with optimized hyperparameters, the feature data of the test data set is substituted to obtain the predicted coordinate data set of the next cycle; S6. The controller wirelessly controls the first direction adjustment component, the second direction adjustment component and the third direction adjustment component on the target ball mechanism based on the predicted coordinate data set so that the target ball is turned and aligned with the laser tracking mechanism (29).