Intelligent sensing and testing system for navigation situation of inland ship and evaluation method of intelligent sensing and testing system

Abstract

The invention relates to an inland ship navigation situation intelligent sensing test system and an evaluation method thereof. The inland ship navigation situation intelligent sensing test system comprises a sensing acquisition subsystem, a data augmentation subsystem and an inland ship navigation situation intelligent sensing test and evaluation subsystem. The sensing acquisition subsystem comprises sensing equipment such as a radar, a camera, an AIS (Automatic Identification System), a GPS (Global Positioning System) and a compass, and is used for acquiring ship navigation data and generating a real data test set; the data augmentation subsystem comprises ship manipulation simulation, radar image simulation, AIS (Automatic Identification System) data simulation, video image simulation and the like; and the test and evaluation subsystem is used for updating the real data test set generated by the intelligent ship navigation sensing equipment acquisition subsystem and the virtual data test set generated by the virtual simulation test subsystem into the test and evaluation subsystem. According to the invention, the inland ship navigation situation intelligent sensing method has the advantages of high test and evaluation precision and good stability, and the experimental test cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of navigation technology, and in particular to an intelligent perception test system for inland river vessel navigation status and an evaluation method thereof. Background Art

[0002] Navigation situation awareness is an important capability of ship intelligence. Intelligent perception of inland river ship navigation situation integrates advanced perception, big data, artificial intelligence and automatic control technologies, and has significant characteristics such as safety, intelligence and efficiency. Driven by advanced technologies such as artificial intelligence, the level of intelligent perception of inland river ship navigation situation continues to improve, and the complexity of the system is further increased. In addition, the curvature of inland waterways, the changeable hydrology and the complex encounter situations have all posed challenges to the testing and evaluation of intelligent perception methods for inland river ship navigation situation. However, traditional testing methods are prone to problems such as inaccurate simulation data and unstable systems. Summary of the invention

[0003] The purpose of the present invention is to provide an inland river vessel navigation situation intelligent perception test system and an evaluation method thereof, which can make the inland river vessel navigation situation intelligent perception method test and evaluation have the advantages of high precision and good stability, and reduce the experimental test cost.

[0004] To achieve the above-mentioned object, the present invention provides the following technical solutions: an inland river vessel navigation situation intelligent perception test system, the inland river vessel navigation situation intelligent perception test system comprising a perception collection subsystem, a data augmentation subsystem and an inland river vessel navigation situation intelligent perception test and evaluation subsystem;

[0005] The sensing and acquisition subsystem includes sensing devices such as radar, camera, AIS, GPS and compass, which are used to collect ship navigation data and generate a real data test set;

[0006] The data augmentation subsystem, including ship maneuvering simulation, radar image simulation, AIS data simulation and video image simulation, is responsible for simulating the ship navigation environment data in the virtual world and generating a virtual data test set.

[0007] The inland river vessel navigation situation intelligent perception test and evaluation subsystem is used to update the real data test set generated by the intelligent ship navigation perception equipment acquisition subsystem and the virtual data test set generated by the virtual simulation test subsystem into the test and evaluation subsystem.

[0008] Furthermore, the ship maneuvering simulation adopts a steering response model to establish a model, and uses methods such as the Euler method or the Runge-Kutta method to perform integral solution operations on the state to achieve the simulation of the ship's motion state.

[0009] Furthermore, the radar image simulation adopts a scan line intersection algorithm, a ray tracing algorithm or an echo intensity algorithm based on the radar equation.

[0010] Furthermore, the AIS data simulation method comprises the following steps:

[0011] A1: Based on the collected ship navigation data, the longitude and latitude, speed and heading are obtained, and then classified according to the ship dynamic information and ship static and navigation related information of AIS;

[0012] A2: Encode AIS data according to ITU-R M.1371 protocol to generate AIS simulation message;

[0013] A3: Determine the sending interval according to the AIS ship's sailing speed and send AIS simulation messages.

[0014] Furthermore, the radar image of the data augmentation subsystem adopts the K distribution model to randomly generate clutter images, which are superimposed on the radar echo image, and the Rayleigh distribution is used to describe the amplitude distribution of rain and snow clutter to achieve radar image data augmentation. The video image adopts the geometric augmentation method to change the position of the target in the image through operations such as image translation and rotation to achieve video image data augmentation.

[0015] An evaluation method for an inland river vessel navigation situation intelligent perception test system, the evaluation method comprising the following steps:

[0016] Step S1: Construct the evaluation model of the target layer, decision layer and scheme layer according to the hierarchical analysis method;

[0017] Step S2: By comparing the indicators at each level with each other, the relative importance matrix c of the fuzzy hierarchical analysis method is obtained. ij ,

[0018] c ij is an n*n matrix, c ij Indicates the importance of indicator i corresponding to j and c ii =1, c ji =1 / c ij ;

[0019] The relative importance matrix c in the fuzzy analytic hierarchy process ij Each item uses three numbers to fuzzy the exact number of the general pairwise discriminant matrix to determine the score D of each indicator of the solution i :

[0020]

[0021] In the formula, Indicates pessimistic score, Indicates optimism score, represents the maximum possible score.

[0022] Step S3: Relative importance matrix c ij Processing is performed to calculate the geometric mean of each judgment matrix row element and the fuzzy weight w of each indicator. i :

[0023]

[0024] Where n is the number of rows and columns of the matrix;

[0025] Then use the following formula to defuzzify:

[0026]

[0027] The standardized weight formula is as follows:

[0028]

[0029] Step S4: The evaluation of target recognition rate and target recognition accuracy is obtained through the data collected by the sensor; the evaluation of parameters such as target recognition missed detection rate, target recognition accuracy, target tracking accuracy, target type accuracy, target position accuracy, target speed accuracy, target heading accuracy, target data update rate, etc. is evaluated by the expert group on the simulation results. The evaluation matrix R is obtained through standardization processing, and the result Q is finally obtained. i .

[0030] Furthermore, in step S4:

[0031] The membership function of the missed detection rate of the target recognition is:

[0032] f(x1)=1-x1,0≤x1≤1

[0033] Where x1 is the missed detection rate of the target.

[0034] The membership function of the target recognition accuracy is:

[0035] f(x2)=x2,0≤x2≤1

[0036] Where x2 is the target recognition accuracy.

[0037] The membership function of the target tracking accuracy is:

[0038] f(x3)=x3,0≤x3≤1

[0039] Where x3 is the target tracking accuracy.

[0040] The membership function of the target type accuracy is:

[0041] f(x4)=x4,0≤x4≤1

[0042] Where x4 is the target type accuracy.

[0043] The membership function of the target position accuracy is:

[0044]

[0045] Where x5 is the target position accuracy.

[0046] The membership function of the target speed accuracy is:

[0047]

[0048] Where x6 is the target speed accuracy.

[0049] The membership function of the target heading accuracy is:

[0050]

[0051] Where x7 is the target heading accuracy.

[0052] The membership function of the target data update rate is:

[0053]

[0054] Where x8 is the target data update rate.

[0055] Furthermore, the result Q in step S4 i for:

[0056] Q i =FW i ×R.

[0057] Beneficial effects of the present invention: The present invention establishes an inland river ship navigation situation intelligent perception test system. The intelligent ship navigation perception equipment acquisition subsystem is responsible for collecting ship navigation data in the real world and generating a real data test set. The data augmentation subsystem is responsible for simulating the ship navigation environment data in the virtual world. The virtual data test set is generated to update the real data set generated by the intelligent ship navigation perception equipment acquisition subsystem and the virtual test set generated by the virtual simulation test subsystem to the test and evaluation subsystem. The inland river ship navigation situation intelligent perception system under test conducts integrated testing by interacting with the test and evaluation subsystem to effectively evaluate the intelligent level of inland river ship navigation situation perception, environmental cognition ability and adaptability.

[0058] In the evaluation of the perception test system, the present invention introduces multiple parameters such as target recognition rate, target recognition accuracy, target recognition missed detection rate, target tracking accuracy, target type accuracy, target position accuracy, target speed accuracy, target heading accuracy, and target data update rate. The factors considered in the evaluation process are more comprehensive, which is more conducive to obtaining accurate evaluation results. For the evaluation of parameter importance, a combination of qualitative and quantitative methods is adopted according to the source of parameter data to solve the problem that some parameters have uncertainty and ambiguity. In the evaluation process, the fuzzy hierarchical analysis method is used to first calculate the fuzzy weights of each parameter, and then the weight calculation formula is defuzzified and standardized to achieve accurate calculation of the evaluation results.

[0059] Compared with the traditional full simulation system, the system proposed by the present invention is closer to the real traffic scene, making up for the current inadequacy of the intelligent navigation test capability of inland ships. In addition, the present invention also proposes ship motion simulation, radar image simulation, AIS data simulation technology, radar image augmentation, video image augmentation, AIS data augmentation method, which provides offline debugging and testing conditions for the development of the intelligent perception system of inland ship navigation situation, simulates the real inland ship navigation scene, thereby improving the system development efficiency and reducing the experimental test cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of an intelligent perception test system for inland river vessel navigation situation of the present invention;

[0061] Figure 2 A schematic diagram of an intelligent perception test system for inland river vessel navigation situation of the present invention;

[0062] Figure 3 The radar test data of the test data set in the specific embodiment of the present invention;

[0063] Figure 4 It is the video test data of the test data set in the specific embodiment of the present invention;

[0064] Figure 5 AIS test data of the test data set in a specific embodiment of the present invention;

[0065] Figure 6 is the compass test data of the test data set in the specific embodiment of the present invention;

[0066] Figure 7 It is a simulation test interface in a specific embodiment of the present invention;

[0067] Figure 8 Evaluation of the evaluation interface in the specific embodiment of the present invention;

[0068] Fig. 9is a parameter curve in a specific embodiment of the present invention;

[0069] Fig.10 This is a simulation test of the intelligent perception system for inland river vessel navigation status in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0070] The present invention will be further described below in conjunction with the accompanying drawings.

[0071] See also Figure 1 , the present invention provides an embodiment: an inland river ship navigation situation intelligent perception test system, the inland river ship navigation situation intelligent perception test system includes a perception acquisition subsystem, a data augmentation subsystem and an inland river ship navigation situation intelligent perception test and evaluation subsystem;

[0072] The perception acquisition subsystem includes perception devices such as radar, camera, AIS, GPS and compass, which are used to collect ship navigation data and generate real data test sets; the intelligent ship navigation perception device acquisition subsystem is responsible for collecting ship navigation data in the real world and generating real data test sets. On the one hand, the obtained real data test set is updated to the test and evaluation subsystem for testing; on the other hand, the real data is also synchronously updated to the data augmentation subsystem for simulation.

[0073] The data augmentation subsystem, including ship maneuvering simulation, radar image simulation, AIS data simulation and video image simulation, is responsible for simulating the ship navigation environment data in the virtual world and generating a virtual data test set.

[0074] The inland river ship navigation situation intelligent perception test and evaluation subsystem is used to update the real data test set generated by the intelligent ship navigation perception equipment acquisition subsystem and the virtual data test set generated by the virtual simulation test subsystem into the test and evaluation subsystem. The inland river ship navigation situation intelligent perception system under test interacts with the test and evaluation subsystem to carry out integrated testing, effectively evaluating the inland river ship navigation situation perception intelligence level, environmental cognition ability and adaptability, etc.

[0075] Please continue reading Figure 1 As shown, in one embodiment of the present invention, the ship maneuvering simulation adopts a steering response model to establish a model, and uses the Euler method or Runge-Kutta method to perform integral solution calculations on the state to achieve the simulation of the ship's motion state.

[0076] Please continue reading Figure 1As shown, in one embodiment of the present invention, the radar image simulation adopts a scanning line intersection algorithm, a ray tracing algorithm or an echo intensity algorithm based on the radar equation. The three main requirements of radar image simulation are: first, there must be a clear occlusion judgment algorithm, second, the calculation of the echo angle must be reasonable, and third, real-time performance must be guaranteed. Common radar image simulation methods include the following:

[0077] 1) Scan line intersection algorithm. The scan line intersection algorithm is based on target data, which consists of several line segments. The scan line intersection algorithm first finds the intersection point between the scan line and the target, and then uses the Bresenham line drawing algorithm to draw a line segment of any length from this point to the edge of the target, producing an effect similar to using a pencil to smear along the coastline on the nautical chart toward the land side. Although the real-time performance of the simulated radar image generated by the scan line intersection algorithm is met, the authenticity is poor due to the lack of a suitable occlusion judgment algorithm and echo intensity algorithm.

[0078] 2) Ray tracing algorithm. The ray tracing algorithm simulates the diffraction of electromagnetic waves on the target through rays, calculates the projection coordinates of each ray tube and the output field strength of the ray tube, then accumulates the contributions of all ray tubes in the image domain, and finally obtains the radar image through windowing. The ray tracing algorithm has a high sense of reality, but to ensure real-time performance, not too many rays can be used.

[0079] 3) Echo intensity algorithm based on radar equation. Highly realistic simulated radar images cannot be separated from a reasonable echo intensity algorithm, which can determine whether an unobstructed target should be displayed and better simulate gain and tuning functions. The echo intensity algorithm based on radar equation adopts the propagation attenuation algorithm of radar waves and the radar antenna azimuth diagram function to achieve the lateral expansion and radial expansion effects of the echo and the sidelobe false echo, which further improves the realism of the simulated radar image.

[0080] Please continue reading Figure 1 As shown, in one embodiment of the present invention, the method for simulating AIS data includes:

[0081] Next steps:

[0082] A1: Based on the collected ship navigation data, the longitude and latitude, speed and heading are obtained, and then classified according to the ship dynamic information and ship static and navigation related information of AIS;

[0083] A2: Encode AIS data according to ITU-R M.1371 protocol to generate AIS simulation message;

[0084] A3: Determine the sending interval according to the AIS ship's sailing speed and send AIS simulation messages.

[0085] Please continue reading Figure 1 As shown, in one embodiment of the present invention, the radar image of the data augmentation subsystem adopts the K distribution model to randomly generate a clutter image, which is superimposed on the radar echo image, and the Rayleigh distribution is used to describe the amplitude distribution of rain and snow clutter to achieve radar image data augmentation. The video image adopts a geometric augmentation method to change the position of the target in the image through operations such as image translation and rotation to achieve video image data augmentation. The K distribution model can describe the amplitude distribution law of various wind and wave clutter. Wind and wave clutter refers to the scattered echo on the surface of the incoming water. When generating a radar echo image, a clutter image can be randomly generated using a typical wind and wave clutter amplitude distribution and superimposed on the radar echo image to achieve the purpose of data augmentation; rain and snow in the air can be regarded as a large number of uniformly distributed scattering particles within a certain range. The arbitrary change of each scattering particle echo causes the total echo amplitude to change randomly. When generating a radar echo image, a clutter image can also be randomly generated using the rain and snow clutter amplitude distribution and superimposed on the radar echo image to achieve the purpose of data augmentation. Compared with the amplitude distribution of sea clutter, the amplitude distribution of rain and snow clutter can be described by Rayleigh distribution regardless of the radar resolution. Specific embodiment 1:

[0087] Aiming at the testing requirements of the intelligent perception method for inland river ship navigation situation, the present invention constructs a variety of different types of test data sets such as following, meeting, crossing, overtaking, bridge area, curve, anchorage, rain and snow, wind and waves, etc. Each test data set includes radar data, video data, AIS data and compass data, etc. Figures 3 to 6 As shown, these data are all marked with timestamps, and the test system synchronizes the time between different data according to the timestamps to perform simulation tests.

[0088] The entire experimental process is mainly divided into three steps: scenario selection, simulation testing, and auxiliary evaluation.

[0089] First, test scene selection. Test scenes are selected according to different test items. This paper constructs a variety of different types of test data sets such as following, meeting, crossing, overtaking, bridge area, curve, anchorage, rain and snow, wind and waves, etc.

[0090] Secondly, simulation test. In the simulation test mode, the system will display the target ship's track and distance from the ship and other parameter information, such as Figure 7 When the test is completed, the system will automatically save the test data to assist in evaluation.

[0091] Finally, assist evaluation. Enter the assist evaluation mode, such as Figure 8 , 9 As shown, load the test result data saved locally, click Play, and select the playback speed during playback. After playback is completed, click the Output icon to generate an auxiliary evaluation document.

[0092] More than 200 simulation tests were carried out using the inland river vessel navigation situation intelligent perception test system, covering various common inland river traffic scenarios such as following, meeting, crossing, overtaking, bridge areas, ferry crossings, daytime, nighttime, rain and snow, wind and waves. The experiments were sufficient, covered a wide range, and achieved good results.

[0093] The experiment shows that the inland river ship navigation situation intelligent perception test system provides offline debugging and testing conditions for the research of inland river ship navigation situation intelligent perception methods, such as Fig.10 As shown, it is possible to simulate the actual inland ship navigation scene without having to use the actual ship navigation environment, thereby improving the verification efficiency of the intelligent perception method for inland ship navigation status, shortening the cycle, reducing the test risk, and reducing the experimental test cost.

[0094] See also Figure 2 The present invention provides an embodiment: an evaluation method for an inland river vessel navigation situation intelligent perception test system, the evaluation method comprising the following steps:

[0095] Step S1: Construct the evaluation model of the target layer, decision layer and scheme layer according to the hierarchical analysis method;

[0096] Step S2: By comparing the indicators at each level with each other, the relative importance matrix c of the fuzzy hierarchical analysis method is obtained. ij ,

[0097] c ij is an n*n matrix, c ij Indicates the importance of indicator i corresponding to j and c ii =1, c ji =1 / c ij ;

[0098] For example, in the traditional comparison matrix, indicator 1 is compared with indicator 2, as shown in Table 1:

[0099]

[0100] Table 1 Importance scale table

[0101] The relative importance matrix c in the fuzzy analytic hierarchy process ij Each item uses three numbers to fuzzy the exact numbers of the general pairwise discriminant matrix. The triangular fuzzy number is formed by collecting three points, which can well solve the problem that the indicators cannot be accurately measured in the pairwise comparison process and can only be fuzzy evaluated in natural language. Determine the score D of each indicator of the scheme i :

[0102]

[0103] In the formula, Indicates pessimistic score, Indicates optimism score, represents the maximum possible score.

[0104] Step S3: Relative importance matrix c ij Processing is performed to calculate the geometric mean of each judgment matrix row element and the fuzzy weight w of each indicator. i ,Right now:

[0105]

[0106] Where n is the number of rows and columns of the matrix;

[0107] Then use the following formula to defuzzify:

[0108]

[0109] The standardized weight formula is as follows:

[0110]

[0111] Step S4: The evaluation of target recognition rate and target recognition accuracy is obtained through the data collected by the sensor; the evaluation of parameters such as target recognition missed detection rate, target recognition accuracy, target tracking accuracy, target type accuracy, target position accuracy, target speed accuracy, target heading accuracy, target data update rate, etc. is evaluated by the expert group on the simulation results. The evaluation matrix R is obtained through standardization processing, and the result Q is finally obtained. i . Where Q i It is the evaluation result of the perception test system obtained by the i-th indicator; the present invention takes the intelligent perception system of inland river ship navigation situation as the research object, combines the actual situation of inland river ship intelligent navigation, uses fuzzy hierarchical analysis method to establish an evaluation model, and scientifically establishes the test evaluation model of the intelligent perception system of inland river ship navigation situation. The present invention evaluates the intelligent perception system of inland river ship navigation situation, and mainly considers the missed detection rate, target recognition accuracy, target tracking accuracy, target type accuracy, target position accuracy, target speed accuracy, target heading accuracy, target data update rate and other indicators of the navigation situation perception system in combination with the actual situation. The intelligent perception system of inland river ship navigation situation is different from the general perception system. The missed detection rate, accuracy rate and tracking accuracy rate of target recognition are the basis of the reliability of the intelligent perception system of inland river ship navigation situation, and the recognition accuracy of each characteristic attribute of the target is an important condition. Generally, the higher the target recognition rate, the higher the accuracy rate, the higher the accuracy of the target position, speed, heading, etc., the stronger the perception ability; otherwise, the lower.

[0112] Please continue reading Figure 2 As shown, in one embodiment of the present invention, in step S4:

[0113] The membership function of the missed detection rate of the target recognition is:

[0114] f(x1)=1-x1,0≤x1≤1

[0115] Where x1 is the missed detection rate of the target.

[0116] The membership function of the target recognition accuracy is:

[0117] f(x2)=x2,0≤x2≤1

[0118] Where x2 is the target recognition accuracy.

[0119] The membership function of the target tracking accuracy is:

[0120] f(x3)=x3,0≤x3≤1

[0121] Where x3 is the target tracking accuracy.

[0122] The membership function of the target type accuracy is:

[0123] f(x4)=x4,0≤x4≤1

[0124] Where x4 is the target type accuracy.

[0125] The membership function of the target position accuracy is:

[0126]

[0127] Where x5 is the target position accuracy.

[0128] The membership function of the target speed accuracy is:

[0129]

[0130] Where x6 is the target speed accuracy.

[0131] The membership function of the target heading accuracy is:

[0132]

[0133] Where x7 is the target heading accuracy.

[0134] The membership function of the target data update rate is:

[0135]

[0136] Where x8 is the target data update rate.

[0137] Please continue reading Figure 2 As shown, in one embodiment of the present invention, the result Q in step S4 is i for:

[0138] Q i =FW i ×R.

[0139] In summary: the present invention establishes an inland river ship navigation situation intelligent perception test system, the intelligent ship navigation perception device acquisition subsystem is responsible for collecting ship navigation data in the real world and generating a real data test set, the data augmentation subsystem is responsible for simulating the ship navigation environment data in the virtual world, and generating a virtual data test set to update the real data set generated by the intelligent ship navigation perception device acquisition subsystem and the virtual test set generated by the virtual simulation test subsystem to the test and evaluation subsystem. The inland river ship navigation situation intelligent perception system under test interacts with the test and evaluation subsystem to carry out integrated testing, effectively evaluating the inland river ship navigation situation perception intelligence level, environmental cognition ability and adaptability, etc. The present invention also proposes ship motion simulation, radar image simulation, AIS data simulation technology, radar image augmentation, video image augmentation, AIS data augmentation method, which provides offline debugging and testing conditions for the development of the inland river ship navigation situation intelligent perception system, simulates the real inland river ship navigation scene, thereby improving the system development efficiency and reducing the experimental test cost.

[0140] The above description is only a preferred embodiment of the present invention and should not be construed as limiting the present application. All equivalent changes and modifications made within the scope of the patent application of the present invention should fall within the scope of the present invention.

Claims

1. An intelligent perception test system for inland river vessel navigation situation, characterized in that: The inland river vessel navigation situation intelligent perception test system includes a perception acquisition subsystem, a data augmentation subsystem and an inland river vessel navigation situation intelligent perception test and evaluation subsystem; The sensing and acquisition subsystem includes sensing devices such as radar, camera, AIS, GPS and compass, which are used to collect ship navigation data and generate a real data test set; The data augmentation subsystem includes ship maneuvering simulation, radar image simulation, AIS data simulation and video image simulation, etc., and is responsible for simulating the ship navigation environment data in the virtual world and generating a virtual data test set; The inland river vessel navigation situation intelligent perception test and evaluation subsystem is used to update the real data test set generated by the intelligent ship navigation perception equipment acquisition subsystem and the virtual data test set generated by the virtual simulation test subsystem into the test and evaluation subsystem.

2. According to claim 1, the intelligent perception test system for inland river vessel navigation situation is characterized in that: The ship maneuvering simulation adopts a steering response model to establish a model, and uses methods such as the Euler method or the Runge-Kutta method to perform integral solution operations on the state to achieve the simulation of the ship's motion state.

3. The intelligent perception test system for inland river vessel navigation situation according to claim 1 is characterized by: The radar image simulation adopts a scan line intersection algorithm, a ray tracing algorithm or an echo intensity algorithm based on the radar equation.

4. The intelligent perception test system for inland river vessel navigation situation according to claim 1 is characterized by: The method for AIS data simulation comprises the following steps: A1: Based on the collected ship navigation data, the longitude and latitude, speed and heading are obtained, and then classified according to the ship dynamic information and ship static and navigation related information of AIS; A2: Encode AIS data according to ITU-R M.1371 protocol to generate AIS simulation message; A3: Determine the sending interval according to the AIS ship's sailing speed and send AIS simulation messages.

5. The inland river vessel navigation situation intelligent perception test system according to claim 1 is characterized by: The radar image of the data augmentation subsystem adopts the K distribution model to randomly generate clutter images, which are superimposed on the radar echo image, and the Rayleigh distribution is used to describe the amplitude distribution of rain and snow clutter to achieve radar image data augmentation. The video image adopts the geometric augmentation method to change the position of the target in the image through operations such as image translation and rotation to achieve video image data augmentation.

6. An evaluation method applicable to the inland river vessel navigation situation intelligent perception test system according to claim 1, characterized in that: The evaluation method comprises the following steps: Step S1: Construct the evaluation model of the target layer, decision layer and scheme layer according to the hierarchical analysis method; Step S2: By comparing the indicators at each level with each other, the relative importance matrix c of the fuzzy hierarchical analysis method is obtained. ij , c ij is an n*n matrix, c ij Indicates the importance of indicator i corresponding to j and c ii =1, c ji =1 / c ij ; The relative importance matrix c in the fuzzy analytic hierarchy process ij Each item uses three numbers to fuzzy the exact number of the general pairwise discriminant matrix to determine the score D of each indicator of the solution i : In the formula, Indicates pessimistic score, Indicates optimism score, represents the maximum possible score. Step S3: Relative importance matrix c ij Processing is performed to calculate the geometric mean of each judgment matrix row element and the fuzzy weight w of each indicator. i : Where n is the number of rows and columns of the matrix; Then use the following formula to defuzzify: The standardized weight formula is as follows: Step S4: evaluation of target recognition rate and target recognition accuracy, obtained through data collected by sensors; The expert group evaluates the simulation results for the missed detection rate, target recognition accuracy, target tracking accuracy, target type accuracy, target position accuracy, target speed accuracy, target heading accuracy, target data update rate and other parameters of target recognition. The evaluation matrix R is obtained through standardization, and the result Q is finally obtained. i .

7. The evaluation method of an inland river vessel navigation situation intelligent perception test system according to claim 6 is characterized by: In the step S4, The membership function of the missed detection rate of the target recognition is: f(x)=1-x1,0≤x1≤1 Where x1 is the missed detection rate of the target. The membership function of the target recognition accuracy is: f(x2)=x2,0≤x2≤1 Where x2 is the target recognition accuracy. The membership function of the target tracking accuracy is: f(x3)=x3,0≤x3≤1 Where x3 is the target tracking accuracy. The membership function of the target type accuracy is: f(x4)=x4,0≤x4≤1 Where x4 is the target type accuracy. The membership function of the target position accuracy is: Where x5 is the target position accuracy. The membership function of the target speed accuracy is: Where x6 is the target speed accuracy. The membership function of the target heading accuracy is: Where x7 is the target heading accuracy. The membership function of the target data update rate is: Where x8 is the target data update rate.

8. The evaluation method of an inland river vessel navigation situation intelligent perception test system according to claim 6 is characterized by: The result Q in step S4 i for: Q i =FW i ×R。

Images