Target identification basic probability assignment method based on evidence theory

Abstract

The invention relates to a target identification basic probability assignment method based on an evidence theory, and the method comprises the steps: obtaining a target attribute interval value detected by a sensor, and calculating the square distance between the target attribute interval value and different types of attribute feature values; calculating the interval close degree of the target attribute interval value and the attribute characteristic value based on the square distance; normalizing the interval close degrees of various types of targets; and correcting the normalized interval close degree according to the detection reliability of each sensor under the interference condition to obtain the basic probability assignment of each sensor to each type of target. The method solves the problem of basic probability assignment of how to convert a detection result into an evidence focus element when a target identification detection value is an interval value in a complex interference information environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of target recognition, and in particular relates to a basic probability assignment method for target recognition based on evidence theory in a complex interference information environment. Background Art

[0002] As a generalization of probability theory, Dempser-Shafer evidence theory distributes probability values ​​on the power set of sub-proposition sets. When the evidence provided by a piece of information cannot distinguish two propositions, the probability value is directly distributed to the composite proposition composed of the two propositions, expressing the uncertainty of information in a more natural way. The synthesis of multiple evidences is realized through the Dempster combination formula, and the final decision is made based on the calculated confidence and plausibility. Since confidence and plausibility are the lower and upper bounds of probability, respectively, Dempser-Shafer evidence theory generalizes the accurate point probability in probability theory to interval values ​​to better express uncertainty. The actual use results also prove the superiority of this theoretical tool.

[0003] The basic probability assignment based on evidence theory is generally based on relatively complete information. However, in airborne target detection and identification systems, due to limited detection means or confidentiality, the observation of enemy targets is limited and uncertain, so the number of samples available to establish a description model of target attributes is relatively small. Due to the development of radar frequency agility and frequency hopping technologies, as well as the influence of equipment measurement errors and various interference technologies, the radiation source parameters obtained by electronic reconnaissance equipment are not limited to scalar form, but are often more likely to be interval values. The interval number only requires the given lower limit and upper limit data, which is more suitable for describing applications with lack of information and high uncertainty. Therefore, it is necessary to propose a basic probability assignment method for target recognition based on evidence theory. Summary of the invention

[0004] The technical problems to be solved by the present invention are:

[0005] In order to avoid the shortcomings of the prior art, the present invention provides a basic probability assignment method for target recognition based on evidence theory, which is used to solve the problem of how to convert the detection result into the basic probability assignment of evidence focus when the target recognition detection value is an interval value in a complex interference information environment.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A basic probability assignment method for target recognition based on evidence theory, characterized by comprising:

[0008] Obtain target attribute interval values ​​detected by the sensor, and calculate the square distance between the target attribute interval values ​​and the characteristic values ​​of different types of attributes;

[0009] Calculate the interval closeness between the target attribute interval value and the attribute characteristic value based on the square distance;

[0010] Normalize the interval closeness of various types of targets;

[0011] The normalized interval closeness is corrected according to the detection reliability of each sensor under interference conditions to obtain the basic probability assignment of each sensor to each type of target.

[0012] A further technical solution of the present invention is as follows: the calculation of the square distance between the target attribute interval value and the attribute feature values ​​of different types adopts the following formula:

[0013]

[0014] Wherein, D=[d1, d2] is the target attribute interval value detected by the sensor, and e is the attribute characteristic value.

[0015] A further technical solution of the present invention is as follows: the calculation of the interval closeness between the target attribute interval value and the attribute characteristic value based on the square distance adopts the following formula:

[0016]

[0017] Where c is the adjustment coefficient.

[0018] A further technical solution of the present invention is as follows: the interval closeness of various types of targets is normalized using the following formula:

[0019]

[0020] Where n is the number of target types.

[0021] A further technical solution of the present invention is as follows: the normalized interval closeness is corrected according to the detection reliability of each sensor under interference conditions to obtain the basic probability assignment of each sensor to each type of target; the following formula is used:

[0022] m′(A i )=β j m(A i )

[0023] m′(U)=(1-β j )j=1,……,k

[0024] Among them, m′(A i ) is the basic probability assignment of each sensor to the i-th type of target, k is the number of sensors, and m′(U) is the support for the uncertain set.

[0025] A target recognition method based on evidence theory, characterized by comprising:

[0026] The basic probability assignments of each sensor to the same target are fused to obtain the fusion probability of each target, and the support of each sensor to the uncertain set is fused to obtain the fused support of the uncertain set;

[0027] Based on the fusion probability of each target and the support of the fusion to the uncertain set, the decision based on evidence theory is adopted to determine the target type.

[0028] The target type is determined by using a decision based on evidence theory based on the fusion probability of each target, including:

[0029] Select the maximum fusion probability among the fusion probabilities of each target; select the second maximum fusion probability among the fusion probabilities of each target;

[0030] The difference between the maximum fusion probability and the second maximum fusion probability is judged by the first threshold; the support of the fusion to the uncertain set is judged by the second threshold; the maximum fusion probability and the support of the fusion to the uncertain set are judged;

[0031] When the following conditions are met at the same time: the difference between the maximum fusion probability and the second maximum fusion probability is greater than the first threshold; the support of the fusion for the uncertain set is less than the second threshold; the maximum fusion probability is greater than the support of the fusion for the uncertain set; then the target is the target corresponding to the maximum fusion probability.

[0032] A computer system, characterized in that it comprises: one or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned method.

[0033] A computer-readable storage medium, characterized in that it stores computer-executable instructions, which are used to implement the above method when executed.

[0034] A computer program product, characterized by comprising computer executable instructions, wherein the instructions are used to implement the above method when executed.

[0035] The beneficial effects of the present invention are:

[0036] The present invention provides a target recognition basic probability assignment method based on evidence theory. When the target recognition detection value is an interval value in a complex interference information environment, the interval closeness of the attribute interval value and the attribute characteristic value is calculated for each target detection value, and the normalized interval closeness is corrected according to the detection reliability of each sensor under interference conditions to obtain the basic probability assignment of each sensor to each type of target. The method of the present invention solves the problem of how to convert the detection result into the basic probability assignment of the evidence focus when the target recognition detection value is an interval value in a complex interference information environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0038] Figure 1 The present invention provides a flowchart of enriching the basic probability assignment of target recognition based on evidence theory.

[0039] Figure 2 A description of the uncertainty of the evidence. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention 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 invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0041] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0042] First, let’s explain the theory of evidence:

[0043] (1) Identification framework

[0044] Now suppose there is a decision problem. The set of all possible results that we can recognize for this problem is represented by U. Then any proposition we are concerned about corresponds to a subset of U. Matching propositions and subsets can transform relatively abstract logical concepts into relatively intuitive set theory concepts. The disjunction, composition and implication of any two propositions correspond to the union, intersection and inclusion of the corresponding sets respectively, and the negation of any proposition corresponds to the complement of the set corresponding to the proposition. Shafer pointed out: The selection of U depends on our knowledge, our level of cognition, what we know and what we want to know. In order to emphasize this epistemological characteristic of the possibility set U, Shafer called it an identification framework. When a proposition corresponds to a subset of the framework, it is said that the framework can identify the proposition. The selection of U should be rich enough so that any specific set of propositions we consider can correspond to a set class in the power set of U.

[0045] (2) Basic concepts of evidence theory

[0046] Definition 1: Let U be a recognition frame, then function m:2 U →[0,1] satisfies the following conditions

[0047] (1)m(φ)=0

[0048] (2)

[0049] When , m(A) is called the basic probability assignment of A.

[0050] m(A) represents the precise degree of confidence in proposition A and indicates direct support for A.

[0051] Definition 2: Let U be a recognition frame, m: 2 U →[0,1] is the basic probability assignment on U, defining the function

[0052] BEL: 2 U →[0,1]

[0053]

[0054] This function is called the trust function on U.

[0055] represents the sum of the likelihood measures of all subsets of A, that is, the total trust in A, so we know

[0056] BEL(φ)=0,BEL(U)=1

[0057] Definition 3: If a subset of the recognition framework U is A, with m(A)>0, then A is called the focal element of the trust function BEL,

[0058] The union of all focal elements is called the nucleus.

[0059] For the A not known degree available to measure the degree of trust.

[0060] Definition 4: Let U be an identification frame, define PL:2 U →[0,1]

[0061]

[0062] PL is called the plausibility function.

[0063] PL(A) represents the confidence that A is not denied, which is the sum of the basic probability assignments of all sets intersecting with A, and BEL(A)≤PL(A), and PL(A)-BEL(A) represents the degree of ignorance of A. The specified confidence interval (BEL(A), PL(A)) describes the uncertainty of A.

[0064] Definition 5: [BEL(A), PL(A)] is called the confidence interval of focal element A.

[0065] PL(A)-BEL(A) describes the uncertainty of A, which is called the uncertainty of the focal element A. PL(A) corresponds to the upper probability metric defined by Dempster, and BEL(A) corresponds to the lower probability metric.

[0066] The uncertainty of the evidence is described as follows: Figure 1 As shown:

[0067] (3) Combination rules of evidence theory

[0068] Dempster's law of composition is a law that reflects the joint effect of evidence. Given several belief functions based on different evidences on the same recognition framework, if these batches of evidence are not completely conflicting, a belief function can be calculated using the law of composition, and this belief function can be used as the belief function under the joint effect of those batches of evidence. This belief function is called the direct sum of the original belief functions.

[0069] Assume that BEL1 and BEL2 are two trust functions on the same recognition framework U, m1 and m2 are their corresponding basic probability assignments, and the focal elements are A1,…,A k and B1, ..., B r , and set

[0070]

[0071] but:

[0072]

[0073] In the above formula, if K1≠1, then m determines a basic probability assignment; if K1=1, then m1 and m2 are considered contradictory and the basic probability assignment cannot be combined. For the combination of multiple pieces of evidence, the above rules can be used for pairwise synthesis.

[0074] (4) Decision-making based on evidence theory

[0075] How to make decisions after combining evidence with DS rules is a problem closely related to applications. Let U be the recognition framework and m be the basic probability assignment obtained after combining evidence with DS rules.

[0076] 1) Decision-making based on trust function

[0077] According to the m obtained after the combination, the trust function BEL is obtained, and the trust function is our judgment result, which is actually a soft decision.

[0078] 2) Decision-making based on basic probability assignment

[0079] set up satisfy

[0080]

[0081]

[0082] Rule 1: The category to be decided should have the largest basic probability assignment m(A1);

[0083] Rule 2: The difference between the basic probability assignment of the decided category and other categories must be greater than a certain threshold;

[0084] m(A1)-m(A2)>ε1

[0085] Rule 3: The uncertain probability assignment m(X) must be less than a certain threshold;

[0086] m(X)<ε2

[0087] Rule 4: The basic probability assigned to the decision category must be greater than the uncertain probability assigned;

[0088] m(A1)>m(X)

[0089] 2. Operations on interval numbers

[0090] Remember a=[a - ,a + ]={x|0<a - ≤x≤a +}, a is called a positive closed interval number, or interval number for short. - =a + , a degenerates into a positive real number.

[0091] Let a=[a - ,a + ] and b=[b - ,b + ] are two positive closed interval numbers, then

[0092] ①a+b=[a - +b - ,a + +b + ];

[0093] ②a·b=[a - b - ,a + b + ]; in particular, λa=[λa - ,λa + ];

[0094] ③a / b=[a - / b + ,a + / b - ]; in particular, 1 / b = [1 / b + ,1 / b - ];

[0095] ④a=b, if and only if a - =b - , a + =b + .

[0096] Definition 7: Let A = [a1, a2] and B = [b1, b2] be two interval numbers, then the square distance S between the interval numbers A and B is 2 (A,B) is:

[0097]

[0098] Based on the above theoretical knowledge, the present invention provides a basic probability assignment method for target recognition based on evidence theory. When the target recognition detection value in a complex interference information environment is an interval value, the interval closeness of the attribute interval value and the attribute characteristic value of each target detection value is calculated, and the normalized interval closeness is corrected according to the detection reliability of each sensor under interference conditions to obtain the basic probability assignment of each sensor to each type of target.

[0099] Definition: Interval closeness. The interval closeness referred to in the present invention is the closeness between an interval number and a real number. Let A = [a1, a2] be an interval number, and B = b be a real number, then their closeness T(A, B) is defined as:

[0100]

[0101] Among them, c is the adjustment coefficient, and its main function is to adjust the discrete degree of the generated closeness value.

[0102]

[0103] The method comprises the following steps:

[0104] Step 1: Based on the target attribute interval value D=[d1, d2] detected by the sensor, calculate the square distance between the interval value and the characteristic values ​​e of different types of attributes.

[0105]

[0106] Step 2: Calculate the interval closeness between the detected attribute interval value D = [d1, d2] and the attribute characteristic value e.

[0107]

[0108] From the definition of closeness, it can be seen that when the interval number D degenerates into a real number and is equal to e, S(D,e) = 0, T(D,e) = 1. The greater the difference between the interval numbers D and e, the smaller the closeness is because the distance between the interval numbers is inversely proportional to the similarity.

[0109] The value of the adjustment coefficient c is appropriately selected based on experience.

[0110] Step 3: Normalize the interval closeness of various types of targets:

[0111]

[0112] Where n is the number of target types.

[0113] Step 4: According to the detection reliability β of each sensor under interference conditions j , the normalized interval closeness is corrected to obtain the basic probability assignment m′(A i ):

[0114] m′(A i )=β j m(A i )

[0115] m′(U)=(1-β j )(j=1,……,k)

[0116] Where k is the number of sensors and m′(U) is the support for the uncertain set.

[0117] In order to enable those skilled in the art to better understand the present invention, the present invention is described in detail below in conjunction with specific embodiments.

[0118] In the electronic warfare environment, the detection accuracy and range of various detection facilities are generally greatly affected. Considering the uncertainty and fuzziness of target information, the following example illustrates a basic probability assignment method for target recognition based on evidence theory in a complex interference information environment.

[0119] In this basic probability assignment model, due to the complex interferences, the detection values ​​output by each detection facility are not general real values ​​but interval values.

[0120] In order to verify the effectiveness of the basic probability assignment method, the following examples are described.

[0121] Example 1

[0122] Using low frequency S LF , intermediate frequency S MF and high frequency S HF Three types of sensors are used to identify targets. Suppose the three types of targets detected are A1, A2, and A3, and their characteristic attribute parameters are shown in Table 1.

[0123] Table 1 Target characteristic parameter table

[0124] Target <![CDATA[A1]]> <![CDATA[A2]]> <![CDATA[A3]]> <![CDATA[S LF / MHz]]> 12 18 24 <![CDATA[S MF / MHz]]> 100 112 94 <![CDATA[S HF / MHz]]> 310 316 350

[0125] Assume that in an electronic warfare environment, the three types of sensors are interfered to varying degrees, and the output detection values ​​are interval values: S LF =[12,17], S MF =[98,105], S HF =[312,320]. The adjustment coefficient c is 1.

[0126] Table 2 Normalized values ​​of target proximity

[0127] Target A1 A2 A3 SLF / MHz 0.4590 0.3728 0.1682 SMF / MHz 0.5876 0.1767 0.2357 SHF / MHz 0.2893 0.6494 0.0613

[0128] According to the different intensities of interference to each type of sensor, the detection reliability of the three types of sensors, low-frequency SLF, medium-frequency SMF and high-frequency SHF, are set to 0.9, 0.8 and 0.75 respectively.

[0129] Table 3 Basic probability assignments for various targets

[0130] Target A1 A2 A3 Θ SLF / MHz 0.4131 0.3355 0.1514 0.1000 SMF / MHz 0.4701 0.1414 0.1885 0.2000 SHF / MHz 0.2170 0.4871 0.0460 0.2500

[0131] The basic probability assignments of the three types of sensors to the target, namely low-frequency SLF, medium-frequency SMF and high-frequency SHF, are fused to obtain the final fusion result.

[0132] Table 4 Results after fusion

[0133] Target A1 A2 A3 Θ After fusion 0.5382 0.3618 0.0827 0.0173

[0134] According to the results in Table 4, we can choose ε1 = 0.15 and ε2 = 0.10. Obviously, the decision result can determine that the target is type A1.

[0135] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.

Claims

1. A basic probability assignment method for target recognition based on evidence theory, characterized in that: include: Obtain target attribute interval values ​​detected by the sensor, and calculate the square distance between the target attribute interval values ​​and the characteristic values ​​of different types of attributes; Calculate the interval closeness between the target attribute interval value and the attribute characteristic value based on the square distance; Normalize the interval closeness of various types of targets; The normalized interval closeness is corrected according to the detection reliability of each sensor under interference conditions to obtain the basic probability assignment of each sensor to each type of target.

2. According to claim 1, a target recognition basic probability assignment method based on evidence theory is characterized in that: The calculation of the square distance between the target attribute interval value and the characteristic values ​​of different types of attributes adopts the following formula: Wherein, D=[d1, d2] is the target attribute interval value detected by the sensor, and e is the attribute characteristic value.

3. According to claim 2, a target recognition basic probability assignment method based on evidence theory is characterized in that: The calculation of the interval closeness between the target attribute interval value and the attribute characteristic value based on the square distance adopts the following formula: Where c is the adjustment coefficient.

4. According to claim 3, a target recognition basic probability assignment method based on evidence theory is characterized in that: The interval closeness of various types of targets is normalized using the following formula: Where n is the number of target types.

5. According to claim 4, a target recognition basic probability assignment method based on evidence theory is characterized in that: The normalized interval closeness is corrected according to the detection reliability of each sensor under interference conditions to obtain the basic probability assignment of each sensor to each type of target; Use the following formula: m′(A i )=β j m(A i ) m′(U)=(1-β j )j=1,……,k Among them, m′(A i ) is the basic probability assignment of each sensor to the i-th type of target, k is the number of sensors, and m′(U) is the support for the uncertain set.

6. A target recognition method based on evidence theory, characterized in that: include: The basic probability assignments of each sensor to the same target are fused to obtain the fusion probability of each target, and the support of each sensor to the uncertain set is fused to obtain the fused support of the uncertain set; Based on the fusion probability of each target and the support of the fusion to the uncertain set, the decision based on evidence theory is used to determine the target type.

7. The target recognition method based on evidence theory according to claim 6 is characterized in that: The target type is determined by using a decision based on evidence theory based on the fusion probability of each target, including: Select the maximum fusion probability among the fusion probabilities of each target; select the second maximum fusion probability among the fusion probabilities of each target; The difference between the maximum fusion probability and the second maximum fusion probability is judged with the first threshold; the support of the fusion for the uncertain set is judged with the second threshold; the maximum fusion probability and the support of the fusion for the uncertain set are judged; When the following conditions are met at the same time: the difference between the maximum fusion probability and the second maximum fusion probability is greater than the first threshold; the support of the fusion for the uncertain set is less than the second threshold; the maximum fusion probability is greater than the support of the fusion for the uncertain set; then the target is the target corresponding to the maximum fusion probability.

8. A computer system, characterized in that include: One or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method of claim 1 or 6.

9. A computer-readable storage medium, characterized in that Computer executable instructions are stored, and when the instructions are executed, they are used to implement the method of claim 1 or 6.

10. A computer program product, characterized in that The method comprises computer executable instructions, which are used to implement the method of claim 1 or 6 when being executed.

Images