Identification method and system based on multi-sensor information fusion

Abstract

The invention discloses an identification method and system based on multi-sensor information fusion. According to the method, firstly, a one-dimensional recognition framework and a two-dimensional recognition framework are constructed, and sensor detection results and reliability thereof are described respectively; grouping is carried out according to sensor feature information, calculation is simplified, and the information fusion difficulty is reduced. And then the reliability of the sensor is evaluated by adopting an AHP-TOPSIS method, and evidence is corrected. And finally, integrating each group of evidences by applying a D-S evidence theory fusion rule to generate a final fusion detection result. The method effectively fuses multi-sensor data, improves decision-making efficiency, and is suitable for intelligent identification tasks in a complex environment.

Description

Technical Field

[0001] The present invention relates to the technical field of environment perception of unmanned vehicles, and in particular to a recognition method and system based on multi-sensor information fusion. Background Art

[0002] Unmanned vehicles need to have the ability of intelligent environmental perception, that is, to be able to perceive the surrounding environment without human operation. In order to achieve this capability, unmanned vehicles need to integrate the data information of various sensors to obtain real-time information about the surrounding environment and its own status. Analyzing and processing the sensor perception data through the decision-making planning layer and fully and accurately perceiving the vehicle's surrounding environment are the key to ensuring driving safety and intelligence.

[0003] Detection methods for unmanned vehicles include: monocular vision, binocular vision, LiDAR, millimeter-wave radar, forward-looking infrared radar, ultrasonic sensors, etc. The redundant perception system based on sensor fusion technology uses multiple sensors with complementary characteristics to enhance its perception ability. It has broad development prospects in the field of autonomous driving and has become one of the development trends of intelligent environmental perception technology. Using sensor fusion detection to detect targets is one of the basic tasks to realize the recognition of pedestrians, vehicles and other targets in road scenes in intelligent environments and provide positioning information, position posture, physical size and other parameters to autonomous driving vehicles. The detection effect directly affects the processing results of the subsequent decision-making and planning system.

[0004] Dempster-Shafer theory (DS evidence theory) is widely used in target detection, pattern recognition, risk assessment, fault diagnosis, image processing, and multi-attribute decision-making due to its unique advantages in multi-sensor information fusion. However, the traditional DS evidence theory synthesis rules assume that the evidence is independent, non-conflicting, and equally important. This relatively strict assumption not only limits its application and development, but also is far from the actual situation. Summary of the invention

[0005] Based on this, the embodiments of the present application provide a recognition method and system based on multi-sensor information fusion, which can quickly detect various objects in a complex traffic environment, combine the advantages of multiple sensors, and make fast, accurate and highly robust judgments on the detection information.

[0006] In a first aspect, a recognition method based on multi-sensor information fusion is provided, the method comprising:

[0007] Constructing a recognition framework; wherein the recognition framework includes a one-dimensional recognition framework and a two-dimensional recognition framework, wherein the one-dimensional recognition framework is used to describe the detection result given by the sensor, and the two-dimensional recognition framework is used to describe the characteristic information of the sensor;

[0008] Grouping sensors according to their characteristic information;

[0009] The analytic hierarchy process and ideal point approximation method are used to obtain the parameters of the sensor, and the detection results given by the sensor are corrected according to the parameters;

[0010] The DS evidence theory fusion rules are used to fuse the evidence between different sensor groups to obtain the target recognition result.

[0011] Optionally, the sensors are grouped according to characteristic information of the sensors, specifically including:

[0012] The K-means algorithm is used to group sensors. Given a sample set D = {x1, x2, ..., x m}, the corresponding cluster set C = {c1, c, …, c k}, the square error of the K-means clustering algorithm minimization formula is as follows

[0013]

[0014] in, is cluster C i The smaller the value of E, the higher the similarity of samples in the cluster.

[0015] Optionally, grouping the sensors according to characteristic information of the sensors further includes:

[0016] The greedy strategy is used to minimize the equation and the approximate solution is approached through iterative optimization. The mean vector is first initialized, and then the current clustering division and the mean vector are iteratively updated; if the clustering result remains unchanged after the iterative update, the current clustering division result is returned.

[0017] Optionally, a hierarchical analysis method and an ideal point approximation method are used to obtain parameters of the sensor, and the detection results given by the sensor are corrected according to the parameters, including:

[0018] Normalize the initial decision matrix; Since the dimensions of each indicator are different, in order to eliminate the influence between different dimensions, the initial decision matrix is ​​linearly normalized to obtain the normalized decision matrix V = {v ij} m×n , specifically:

[0019]

[0020] The weights of sensor characteristic information indicators are determined by using the hierarchical analysis method, and a judgment matrix is ​​constructed based on the relative importance scale. First, a hierarchical structure is established, with the top layer being the target layer, the bottom object layer representing all sensors in the example, and the middle criterion layer representing the specific characteristic information indicator definition of the sensor.

[0021] Determine the ideal solution v + and negative ideal solution v - ;

[0022]

[0023]

[0024] Among them, J + , J - Represent profitability index and cost index respectively, i and j represent different factors respectively;

[0025] The distances between the evaluation object and the ideal solution and the negative ideal solution are calculated respectively, and the closeness of each evaluation object to the ideal solution is calculated.

[0026] Optionally, the distances between the evaluation object and the ideal solution and the negative ideal solution are calculated respectively, and the closeness of each evaluation object to the ideal solution is calculated, specifically including:

[0027] Determine the weighted Euclidean distances between the evaluation object and the ideal solution and the negative ideal solution respectively. and

[0028]

[0029]

[0030] Calculate the closeness T between each evaluation object and the ideal solution i As correction factor:

[0031]

[0032] Where 0≤T i ≤1.

[0033] Optionally, evidences from different sensor groups are fused using DS evidence theory fusion rules to obtain target recognition results, including:

[0034] The weighted average method is used to fuse the evidence within the group. The fusion of group m includes:

[0035]

[0036] Among them, R m is the number of evidences for group m, ω m，n is the weight of evidence n for group m, ω m，n It is calculated by the Pignistic probability distance between the evidence within the group;

[0037]

[0038] in, represents the sum of the Pignistic similarities between evidence n in group m and all evidence in group m, represents the sum of the Pignistic similarities of all evidences in group m; where E a With E b The formula for calculating the sum of the evidence similarity is:

[0039]

[0040] The calculation method is as follows:

[0041]

[0042] After the above fusion process, each group of evidence is transformed into an evidence body, and j new evidence bodies E′={E′1, E′2, …, E′ j}; The final result of the fusion of each set of evidence is a piece of evidence containing the original evidence source information;

[0043] After integrating the evidence within the group, Dempster's rule is used to integrate the evidence between groups:

[0044]

[0045] And the target recognition result is determined from all fusion results according to the preset judgment threshold.

[0046] In a second aspect, a recognition system based on multi-sensor information fusion is provided, the system comprising:

[0047] A construction module, used to construct a recognition framework; wherein the recognition framework includes a one-dimensional recognition framework and a two-dimensional recognition framework, wherein the one-dimensional recognition framework is used to describe the detection result given by the sensor, and the two-dimensional recognition framework is used to describe the characteristic information of the sensor;

[0048] A grouping module, used for grouping sensors according to characteristic information of the sensors;

[0049] A correction module is used to obtain the parameters of the sensor by using the hierarchical analysis method and the ideal point approximation method, and to correct the detection results given by the sensor according to the parameters;

[0050] The fusion module is used to use the DS evidence theory fusion rules to fuse the evidence between different sensor groups to obtain the target recognition result.

[0051] In a third aspect, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the identification method described in any one of the first aspects is implemented.

[0052] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the identification method described in any one of the first aspects is implemented.

[0053] In a fifth aspect, a computer program product is provided, comprising a computer program / instruction, which, when executed by a processor, implements any identification method described in the first aspect.

[0054] The beneficial effects brought by the technical solution provided by the embodiment of the present application include at least:

[0055] (1) This application adopts an unmanned driving environment perception system based on multiple sensors. By reasonably arranging the multi-sensor fusion strategy on the vehicle body, it can make up for the shortcomings of false detection and missed detection of a single sensor and meet the perception needs of unmanned vehicles in tunnel environments.

[0056] (2) At the decision-making level, the decision-making information fusion model based on the two-dimensional evidence theory can control the entire process from data collection to decision-making, redistribute the basic probability weights of the evidence obtained at the target level, prevent error interference, and significantly improve the accuracy of real-time decision-making and reduce the false detection rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0058] Figure 1 The overall architecture diagram of fusion recognition provided in the embodiment of the present application;

[0059] Figure 2 A flowchart of the steps of the identification method based on multi-sensor information fusion provided in the embodiment of the present application;

[0060] Figure 3 A block diagram of an identification system based on multi-sensor information fusion provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] In the description of the present invention, the terms "comprises", "has" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may also include other steps or units that are not explicitly listed but are inherent to these processes, methods, products or apparatuses, or steps or units added based on further optimization schemes conceived by the present invention.

[0063] The purpose of the present invention is to provide a multi-sensor information fusion method for unmanned vehicles to address the corresponding deficiencies of the prior art, quickly detect various objects in a complex traffic environment, combine the advantages of multiple sensors, and make fast, accurate and highly robust judgments on the detection information.

[0064] To achieve the above-mentioned purpose, the present invention provides a technical solution of a fusion detection method for unmanned vehicles based on evidence theory:

[0065] The sensor detection system includes n sensors of different types, where n is an integer greater than or equal to 2. The method includes the following steps:

[0066] First, a two-dimensional recognition framework is constructed. The one-dimensional recognition framework describes the detection results given by the sensor, and the two-dimensional recognition framework describes the characteristic information of the sensor to represent their reliability. Then, the detection results and the characteristic information of the sensor are converted into evidence, which can be represented by basic probability distribution.

[0067] Secondly, sensors are grouped according to feature information. After establishing a two-dimensional recognition framework, sensors can be further grouped according to feature information due to strong conflicts between evidence, a large number of sensors, and complex calculations. Feature information can reflect the reliability of sensor detection information. In the present invention, sensors are grouped according to data such as type and working environment. This can simplify the calculation complexity, reduce the difficulty of information fusion, and improve decision-making efficiency.

[0068] Then, the evidence correction factor is generated. The characteristic information of the sensor will affect the reliability of its recommendation. We use the AHP (Analytic Hierarchy Process)-TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) method to obtain the reliability of the sensor and use the obtained parameters to correct one dimension of the evidence.

[0069] Finally, the sensor fusion detection results are generated. This model is based on the information provided by multiple sensors on the same object on the same day. After sensor grouping and evidence correction, this application uses the DS evidence theory fusion rule to fuse the evidence between different groups to obtain the final detection results. Figure 1 The overall architecture diagram of the fusion recognition of the proposed method is shown.

[0070] Please refer to Figure 2 , which shows a flow chart of an identification method based on multi-sensor information fusion in one embodiment, the method may include the following steps:

[0071] S1, build recognition framework.

[0072] The recognition framework includes a one-dimensional recognition framework and a two-dimensional recognition framework. The one-dimensional recognition framework is used to describe the detection results given by the sensor, and the two-dimensional recognition framework is used to describe the characteristic information of the sensor.

[0073] S2, grouping sensors according to their characteristic information.

[0074] S3, uses the hierarchical analysis method and the ideal point approximation method to obtain the parameters of the sensor, and corrects the detection results given by the sensor according to the parameters.

[0075] S4 uses the DS evidence theory fusion rules to fuse the evidence between different sensor groups to obtain the target recognition result.

[0076] The specific embodiments of the above method are given below:

[0077] 1. Construction of two-dimensional recognition framework

[0078] Step 1: (Recognition framework) The set of all possible recognitions of the unmanned driving system is Θ = {θ1, θ2, ..., θ n The elements in the set are mutually exclusive and complete, that is, for any i, j∈{1, 2, …, n} and in, represents the empty set. Θ is called the recognition frame. The power set of Θ is denoted by 2 Θ , which can be expressed as:

[0079]

[0080] Step 2: (Basic Probability Allocation BPA) is a function m that satisfies the following condition: 2 θ →[0, 1]:

[0081]

[0082] Step 3: (Two-dimensional recognition framework) For any decision problem, in addition to the traditional information and recognition framework Θ, a one-dimensional evidence information and recognition framework Ψ is added to formally represent the feature information generated in the process of evidence acquisition or the quality information of different evidence sources. We can express the two-dimensional recognition framework as follows:

[0083] γ={Ψ1,Ψ2,…,Ψ n}→{Θ1, Θ2, …, Θ m} (3)

[0084] Step 4: (Correction Factor) Evidence E i The correction factor α i is a value between 0 and 1. The higher the reliability of the source of evidence, the higher the correction factor α. i The closer the value of is to 1. It can be determined according to the characteristics of the source of evidence, and can be constructed from the characteristics of the source of evidence {C|c1, c2, ..., c I} to the evidence correction factor α = f({C|c1, c2, ..., c I}). The specific correction formula is as follows:

[0085]

[0086] 2. Group sensors based on feature information

[0087] Due to the large number of sensors, complex calculations, and conflicts between evidence, it is difficult to directly correct and fuse the evidence. Therefore, it is possible to consider further grouping based on the characteristic information of the sensors. After the sensor characteristics are determined, the K-means algorithm is used to group the sensors. Given a sample set D = {x1, x2, ..., x m}, the corresponding cluster set C = {c1, c, …, c k}, the squared error of the K-means clustering algorithm minimization formula is as follows.

[0088]

[0089] in, is cluster C i The smaller the value of E, the higher the similarity of samples in the cluster.

[0090] The means algorithm adopts a greedy strategy to minimize equation (5) and approximates the solution through iterative optimization. First, the mean vector is initialized, and then the current clustering division and the mean vector are iteratively updated; if the clustering result remains unchanged after iterative update, the current clustering division result is returned.

[0091] 3. Solve the correction factor using AHP-TOPSIS method

[0092] After determining the identification framework of the two-dimensional evidence theory, it is necessary to use two-dimensional evidence to correct the one-dimensional evidence. The AHP-TOPSIS method is used to solve the correction factor. The sensor characteristic information index established above generates evidence correction factors based on indicators that reflect the reliability of sensor decisions. This requires assigning weights to different indicators.

[0093] This application combines the AHP and TOPSIS methods to obtain the correction factor of sensor fusion decision. Assuming there are i sensors and j evaluation indicators, the initial decision matrix X is constructed as follows:

[0094]

[0095] Among them, x ij Represents the value of evaluation index j of sensor i. The specific steps of applying the AHP-TOPSIS method are as follows:

[0096] Step 1. Normalize the initial decision matrix. Since the dimensions of each indicator are different, in order to eliminate the influence of different dimensions, the initial decision matrix is ​​linearly normalized to obtain the normalized decision matrix V = {v ij} m×n :

[0097]

[0098] Step 2. Determine the weight W j The analytic hierarchy process (AHP) is used. The AHP is used to determine the weights of the sensor characteristic information indicators. First, a hierarchical structure is established with the target layer at the top, which is the indicator of sensor reliability. The object layer at the bottom represents all sensors in the sample. The criterion layer in the middle is defined by the sensor specific characteristic information indicators.

[0099] Secondly, we need to construct a judgment matrix. The values ​​of the elements in the judgment matrix reflect the sensor's understanding of the relative importance of each element. This understanding is usually based on a fuzzy judgment of objective reality. We use a 1-9 relative importance scale (Table 1) to compare the importance of each factor pairwise and analyze the relationship between the basic factors in the evaluation system.

[0100] Table 1. Relative importance scale

[0101]

[0102] Step 3. Determine the ideal solution v + and negative ideal solution v - :

[0103]

[0104] Among them J + , J - They represent profitability and cost indicators respectively (the larger the profitability indicator value, the better the evaluation object; the smaller the cost indicator value, the better the evaluation object).

[0105] Step 4. Calculate the distance between the evaluation object and the ideal solution and the negative ideal solution respectively. Suppose the weighted Euclidean distance between the evaluation object and the ideal solution and the negative ideal solution are and

[0106]

[0107]

[0108] Step 5. Calculate the closeness T between each evaluation object and the ideal solution i .

[0109]

[0110] 0≤T i ≤1, T i The closer it is to 1, the better the evaluation object is, that is, the higher the reliability of the sensor is. Therefore, T is used. i The value of is taken as the correction factor α.

[0111] 4. Integrating evidence within and between groups

[0112] After the sensors are grouped, considering that the sensors in the group are highly similar and the evidence is likely to be correlated, the weighted average method is used to fuse the evidence in the group. The fusion of group m is as follows:

[0113]

[0114] Where R m is the number of evidences for group m, ω m，n is the weight of evidence n for group m, ω m，n It can be calculated by the Pignistic probability distance between evidence within a group.

[0115]

[0116] in represents the sum of the Pignistic similarities between evidence n in group m and all evidence in group m, represents the sum of the Pignistic similarities of all evidences in group m. a With E b The formula for calculating the sum of the evidence similarity is:

[0117]

[0118] The calculation method is as follows:

[0119]

[0120] After the above fusion process, each group of evidence is transformed into an evidence body, and j new evidence bodies E′={E′1, E′2, …, E′ j The final result of the fusion of each group of evidence is a piece of evidence containing the original evidence source information. After the evidence within the group is fused, the Dempster rule is used to fuse the evidence between groups:

[0121]

[0122] 5. Determination rules

[0123] Assume that there are n singleton subsets A1, A2, …, An in the recognition framework Θ, if they satisfy:

[0124]

[0125] Among them, U represents an uncertain proposition, ε1 and ε2 are two judgment thresholds. In the entire fusion result, A1 is output as the final fusion result.

[0126] The key points of this application are:

[0127] (1) The unmanned driving environment perception system based on multiple sensors is adopted. Through the reasonable arrangement of the multi-sensor fusion strategy on the vehicle body, it can make up for the shortcomings of false detection and missed detection of a single sensor and meet the perception needs of unmanned vehicles in the tunnel environment.

[0128] (2) At the decision-making level, the decision-making information fusion model based on the two-dimensional evidence theory can control the entire process from data collection to decision-making, redistribute the basic probability weights of the evidence obtained at the target level, prevent error interference, and significantly improve the accuracy of real-time decision-making and reduce the false detection rate.

[0129] In one embodiment of the present application, Figure 3 , and also provides a block diagram of a recognition system based on multi-sensor information fusion, the system comprising:

[0130] A construction module, used to construct a recognition framework; wherein the recognition framework includes a one-dimensional recognition framework and a two-dimensional recognition framework, wherein the one-dimensional recognition framework is used to describe the detection result given by the sensor, and the two-dimensional recognition framework is used to describe the characteristic information of the sensor;

[0131] A grouping module, used for grouping sensors according to characteristic information of the sensors;

[0132] A correction module is used to obtain the parameters of the sensor by using the hierarchical analysis method and the ideal point approximation method, and to correct the detection results given by the sensor according to the parameters;

[0133] The fusion module is used to use the DS evidence theory fusion rules to fuse the evidence between different sensor groups to obtain the target recognition result.

[0134] For the specific definition of the identification system based on multi-sensor information fusion, please refer to the definition of the identification method based on multi-sensor information fusion mentioned above, which will not be repeated here. Each module in the above-mentioned identification system based on multi-sensor information fusion can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0135] In one embodiment, an electronic device is provided, which may be a computer. The electronic device includes a processor, a memory and a network interface connected via a system bus. The processor of the device is used to provide computing and control capabilities. The memory of the device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used for identification data based on multi-sensor information fusion. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a recognition method based on multi-sensor information fusion is implemented.

[0136] In one embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned identification method based on multi-sensor information fusion are implemented.

[0137] In one embodiment of the present application, a computer program product is provided, including a computer program / instruction, which implements the steps of the above-mentioned identification method based on multi-sensor information fusion when executed by a processor.

[0138] The computer-readable storage medium and computer program product provided in this embodiment have similar implementation principles and technical effects to those of the above method embodiments, and are not described in detail here.

[0139] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in M ​​forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (SyMchliMk) DRAM (SLDRAM), memory bus (RaMbus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0140] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0141] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A recognition method based on multi-sensor information fusion, characterized in that: The method comprises: Constructing a recognition framework; wherein the recognition framework includes a one-dimensional recognition framework and a two-dimensional recognition framework, wherein the one-dimensional recognition framework is used to describe the detection result given by the sensor, and the two-dimensional recognition framework is used to describe the characteristic information of the sensor; Grouping sensors according to their characteristic information; The analytic hierarchy process and ideal point approximation method are used to obtain the parameters of the sensor, and the detection results given by the sensor are corrected according to the parameters; The DS evidence theory fusion rules are used to fuse the evidence between different sensor groups to obtain the target recognition result.

2. The identification method according to claim 1, characterized in that: Sensors are grouped according to their characteristic information, including: The K-means algorithm is used to group sensors. Given a sample set D = {x1, x2, …, x m }, the corresponding cluster set C = {c1,c,…,c k }, the square error of the K-means clustering algorithm minimization formula is as follows in, is cluster C i The smaller the value of E, the higher the similarity of samples in the cluster.

3. The identification method according to claim 2, characterized in that: Grouping sensors according to their characteristic information also includes: The greedy strategy is used to minimize the equation and the approximate solution is approached through iterative optimization. The mean vector is first initialized, and then the current clustering division and the mean vector are iteratively updated; if the clustering result remains unchanged after the iterative update, the current clustering division result is returned.

4. The identification method according to claim 1, characterized in that: The analytic hierarchy process and ideal point approximation method are used to obtain the parameters of the sensor, and the detection results given by the sensor are corrected according to the parameters, including: Normalize the initial decision matrix; Since the dimensions of each indicator are different, in order to eliminate the influence between different dimensions, the initial decision matrix is ​​linearly normalized to obtain the normalized decision matrix V = {v ij } m×n , specifically: The weights of sensor characteristic information indicators are determined by using the hierarchical analysis method, and a judgment matrix is ​​constructed based on the relative importance scale. First, a hierarchical structure is established, with the top layer being the target layer, the bottom object layer representing all sensors in the example, and the middle criterion layer representing the specific characteristic information indicator definition of the sensor. Determine the ideal solution v + and negative ideal solution v - ; Among them, J + ,J - Represent profitability index and cost index respectively, i and j represent different factors respectively; The distances between the evaluation object and the ideal solution and the negative ideal solution are calculated respectively, and the closeness of each evaluation object to the ideal solution is calculated.

5. The identification method according to claim 1, characterized in that: The distance between the evaluation object and the ideal solution and the negative ideal solution is calculated respectively, and the closeness between each evaluation object and the ideal solution is calculated, including: Determine the weighted Euclidean distances between the evaluation object and the ideal solution and the negative ideal solution respectively. and Calculate the closeness T between each evaluation object and the ideal solution i As correction factor: Where 0≤T i ≤1.

6. The identification method according to claim 1, characterized in that: The DS evidence theory fusion rules are used to fuse the evidence between different sensor groups to obtain the target recognition results, including: The weighted average method is used to fuse the evidence within the group. The fusion of group m includes: Among them, R m is the number of evidences for group m, ω m,n is the weight of evidence n for group m, ω m,n It is calculated by the Pignistic probability distance between the evidence within the group; in, represents the sum of the Pignistic similarities between evidence n in group m and all evidence in group m, represents the sum of the Pignistic similarities of all evidences in group m; where E a With E b The formula for calculating the sum of the evidence similarity is: The calculation method is as follows: After the above fusion process, each set of evidence is transformed into an evidence body, and j new evidence bodies E'={E'1,E'2,…,E' j }; The final result of the fusion of each set of evidence is a piece of evidence containing the original evidence source information; After integrating the evidence within the group, Dempster's rule is used to integrate the evidence between groups: And the target recognition result is determined from all fusion results according to the preset judgment threshold.

7. A recognition system based on multi-sensor information fusion, characterized in that: The system comprises: A construction module, used to construct a recognition framework; wherein the recognition framework includes a one-dimensional recognition framework and a two-dimensional recognition framework, wherein the one-dimensional recognition framework is used to describe the detection result given by the sensor, and the two-dimensional recognition framework is used to describe the characteristic information of the sensor; A grouping module, used for grouping sensors according to characteristic information of the sensors; A correction module is used to obtain the parameters of the sensor by using the hierarchical analysis method and the ideal point approximation method, and to correct the detection results given by the sensor according to the parameters; The fusion module is used to use the DS evidence theory fusion rules to fuse the evidence between different sensor groups to obtain the target recognition result.

8. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the recognition method based on multi-sensor information fusion as claimed in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the identification method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the identification method according to any one of claims 1 to 6 is implemented.

Images