Trusted evidence reasoning method, device and equipment for multi-source data fusion

Abstract

The invention relates to a credible evidence reasoning method, device and equipment for multi-source data fusion, a multi-source decision information system is constructed, in the multi-source decision information system, a support matrix is designed to quantify correlation among multi-source data attributes, and a relationship is evaluated by combining concepts of similarity and distance. Firstly, multi-source data of different scales are converted into confidence distribution, probability significance is given to the confidence distribution, and therefore generated evidence can be applied to multi-scale data fusion. Then, a credible evidence reasoning rule is constructed, and in the rule, the credibility of the evidence is determined by the weight and the reliability of the evidence; by integrating multi-source data samples of different scales, confidence distribution of results is derived. Therefore, the problem of difficulty in determining confidence distribution of evidences is solved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device and equipment for credible evidence reasoning based on multi-source data fusion. Background Art

[0002] In the contemporary field of artificial intelligence (AI) and data, the integration of multi-source information is a key technology to improve the depth and accuracy of analysis, thereby improving the accuracy of AI processing tasks. The proliferation of data-generating devices, platforms, and systems has prompted the development of powerful multi-source data fusion (MSDF) methods, which aim to integrate different data streams into a coherent and information-rich whole. Despite significant progress in this field, ensuring the credibility of the fused data remains a key issue.

[0003] From the perspective of methodological requirements and the uncertainty of results, MSDF techniques are generally divided into model-based methods and statistical methods. The former relies on mathematical structures to integrate information from different sources to improve the authenticity of data and the quality of decision-making. Such methods include Kalman filtering, extended Kalman filtering, and maximum likelihood estimation. The latter focuses on applying statistical principles to integrate multi-source data, including principal component analysis, Bayesian reasoning, and cluster analysis. In addition, the combination of model-based and statistical methods has promoted the development of artificial intelligence, including fuzzy logic, rough sets, Dempster-Shafer theory (DST), neural networks, and machine learning. The emergence of machine learning and data mining has accelerated the evolution of MSDF, enabling it to cover more complex and automated data integration strategies. However, the "black box" nature of these models complicates the reliability of the fusion process and the accuracy of the resulting data. In addition, these models require a large amount of labeled data, which may be scarce or biased.

[0004] The DST-based information fusion method is a prominent intelligent framework that allows the representation of uncertainty and the combination of evidence from independent sources without the need for precise probability assignment. DST has been widely used in information fusion, multi-attribute decision-making, pattern recognition, and other fields, and has been further developed into evidential reasoning (ER). Zhang et al. designed a structural health assessment method based on ER rules (CL Zhang, ZJ Zhou, GY Hu, LH Yang, and SWTang, “Health assessment of the wharf based on evidential reasoning ruleconsidering optimal sensor placement,” Measurement, vol. 186, pp. 110184,2021), which improves the accuracy of multi-sensor data fusion by considering the optimal arrangement of sensors. Ren et al. developed a data fusion technology based on ER rules (ML Ren, P. He, and JJ Zhou, “Decision fusion of two sensors object classification based on the evidential reasoning rule,” Expert Syst. Appl. vol. 210, pp. 118620, 2022), taking into account the differences in classification decisions, and verified it on the Nuscenes and Waymo datasets. Liu et al. proposed an ozone prediction model based on the ER method (XQ Liu, YJ Zhang, JP Wang, H. Huang, and H. Yin, “Multi-source and multivariate ozone prediction based on fuzzy cognitive maps and evidential reasoning theory,” Applied Soft Comput. vol. 119, pp. 108600,2022.), which improved the prediction accuracy of multi-source and multivariate time series data.Wu et al. proposed a tunnel collapse risk assessment method based on the ER method (B. Wu, WX Qiu, GW Meng, Y. Nong, and JSHuang, “A multi-source information fusion evaluation method for the tunnelingcollapse disaster based on the artificial intelligence deformationprediction,” Arabian J. Sci. Eng. vol. 47, no. 4, pp. 5053-5071, 2022.). Huang et al. proposed a multi-source transfer learning method based on the ER method (LQ Huang, JF Fan, WBZhao, and Y. You, “A new multi-source transfer learning method based on two-stage weighted fusion,” Knowl. Based Syst. vol. 262, pp. 110233, 2023.), which significantly improved the classification accuracy. Sun and Guo proposed a dynamic assessment technology for fire danger level based on particle swarm optimization (B. Sun, and T. Guo, “Evidential reasoning and lightweight multi-sourceheterogeneous data fusion-driven fire danger level dynamic assessmenttechnique,” ​​Process Safety. Environ. Protect., vol. 185, pp. 350-366, 2024.), which can process multi-source heterogeneous data.

[0005] Although existing research has significantly advanced the development of evidence theory, it still has obvious defects, such as "counterintuitive" problems and combinatorial explosion problems in large-scale data fusion. In particular, when ER is applied to multi-source and multi-scale data fusion, it faces three challenges: (1) the difficulty of determining the confidence distribution of evidence; (2) the dilemma of achieving credible reasoning; and (3) the bottleneck of determining the reliability of evidence. Summary of the invention

[0006] Based on this, it is necessary to provide a credible evidence reasoning method, device and equipment for multi-source data fusion to address the above technical problems.

[0007] A credible evidence reasoning method for multi-source data fusion, the method comprising: Constructing a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; Dividing the multi-source decision information system into multiple data matrices according to attributes, and judging the association relationship between single-source decision information systems according to the decision feature set, when there is an association relationship, calculating the total similarity between the elements in the data matrix and other similar elements, and when there is no association, calculating the total distance between the elements in the data matrix and other dissimilar elements; Determine the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generating evidence for a single-source decision information system based on the first confidence level; According to the evidence, an identification framework is established, and according to the weight and reliability of the evidence, the credibility of the evidence is calculated, and according to the credibility, a second confidence level and a basic probability mass distribution of the evidence are obtained; A combined confidence between any two pieces of evidence is established according to the basic probability mass distribution and the second confidence, and the confidence distribution of multiple pieces of evidence in the multi-source decision information system is obtained according to the combined confidence.

[0008] In one embodiment, the method further includes: dividing the multi-source decision information system into a plurality of data matrices according to attributes, wherein the data matrices are represented as follows: ; in, Representation attributes The corresponding data matrix is, Each row of represents a sample, Each column of represents a single-source decision information system; The elements in the data matrix are normalized to be expressed as: ; The normalized data matrix is ​​obtained by using the normalized elements to represent the data matrix.

[0009] In one embodiment, the further step further includes: determining, based on the decision feature set, the association relationship between the single-source decision information systems as follows: ; in, represents a data sample, represents the decision features in the decision feature set, , ; When there is an association relationship, the total similarity between the element in the data matrix and other similar elements is calculated as: ; When there is no association, the total distance between an element in the data matrix and other dissimilar elements is calculated as: .

[0010] In one embodiment, it further includes: judging the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance: ; The first confidence of the data sample in terms of attribute relative to the single-source decision information system is calculated based on the support: .

[0011] In one embodiment, the present invention further includes: generating evidence of a single-source decision information system according to the first confidence level as follows: ; in, Represents data samples In Properties The evidence provided above and .

[0012] In one embodiment, the invention further includes: establishing an identification framework based on the evidence. ; It represents a single-source decision information system; According to the weight and reliability of the evidence, the credibility of the evidence is calculated as: ; in, and Evidence The weight and reliability of For evidence Credibility; According to the confidence level, the second confidence level is calculated as: ; Among them, evidence The basic probability distribution of is: ; ; in, and The evidence is Assigned to and The probability mass of is a power set, represented by: ; According to the basic probability, we get evidence The basic probability mass distribution of is: ; in, .

[0013] In one embodiment, the method further includes: establishing a combined confidence between any two pieces of evidence according to the basic probability mass distribution and the second confidence: ; in, , , , ; and Respectively express evidence Assigned to and The probability mass of and Assigned to and The unnormalized combined probability mass of and Assigned to and The normalized combined probability mass, After the combination is assigned to confidence level; According to the combined confidence, the confidence distribution of multiple pieces of evidence in the multi-source decision information system is obtained as follows: ; is the joint credibility of all the evidence.

[0014] A credible evidence reasoning device for multi-source data fusion, the device comprising: A multi-source decision system building module is used to build a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; An evidence generation module is used to divide the multi-source decision information system into multiple data matrices according to attributes, and to judge the association relationship between single-source decision information systems according to the decision feature set; when there is an association relationship, calculate the total similarity between the elements in the data matrix and other similar elements; when there is no association, calculate the total distance between the elements in the data matrix and other dissimilar elements; judge the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generate evidence of the single-source decision information system according to the first confidence; A basic probability mass distribution calculation module is used to establish an identification framework based on the evidence, and calculate the credibility of the evidence based on the weight and reliability of the evidence, and obtain a second confidence level and a basic probability mass distribution of the evidence based on the credibility; A fusion module is used to establish a combined confidence between any two pieces of evidence according to the basic probability mass distribution and the second confidence, and obtain the confidence distribution of multiple pieces of evidence in the multi-source decision information system according to the combined confidence.

[0015] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented: Constructing a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; Dividing the multi-source decision information system into multiple data matrices according to attributes, and judging the association relationship between single-source decision information systems according to the decision feature set, when there is an association relationship, calculating the total similarity between the elements in the data matrix and other similar elements, and when there is no association, calculating the total distance between the elements in the data matrix and other dissimilar elements; Determine the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generating evidence for a single-source decision information system based on the first confidence level; According to the evidence, an identification framework is established, and according to the weight and reliability of the evidence, the credibility of the evidence is calculated, and according to the credibility, a second confidence level and a basic probability mass distribution of the evidence are obtained; A combined confidence between any two pieces of evidence is established according to the basic probability mass distribution and the second confidence, and the confidence distribution of multiple pieces of evidence in the multi-source decision information system is obtained according to the combined confidence.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps: Constructing a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; Dividing the multi-source decision information system into multiple data matrices according to attributes, and judging the association relationship between single-source decision information systems according to the decision feature set, when there is an association relationship, calculating the total similarity between the elements in the data matrix and other similar elements, and when there is no association, calculating the total distance between the elements in the data matrix and other dissimilar elements; Determine the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generating evidence for a single-source decision information system based on the first confidence level; According to the evidence, an identification framework is established, and according to the weight and reliability of the evidence, the credibility of the evidence is calculated, and according to the credibility, a second confidence level and a basic probability mass distribution of the evidence are obtained; A combined confidence between any two pieces of evidence is established according to the basic probability mass distribution and the second confidence, and the confidence distribution of multiple pieces of evidence in the multi-source decision information system is obtained according to the combined confidence.

[0017] The above-mentioned credible evidence reasoning method, device and equipment for multi-source data fusion constructs a multi-source decision information system, in which a support matrix is ​​designed to quantify the correlation between multi-source data attributes, and the concepts of similarity and distance are combined to evaluate the relationship. First, multi-source data of different scales are converted into confidence distributions and given probabilistic meanings, so that the generated evidence can be applied to multi-scale data fusion. Then, a credible evidence reasoning rule is constructed, in which the credibility of evidence is determined by its weight and reliability. By integrating multi-source data samples of different scales, the confidence distribution of the results is derived. Thereby solving the problem of difficulty in determining the confidence distribution of evidence. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of a credible evidence reasoning method for multi-source data fusion in one embodiment; Figure 2 It is a structural block diagram of a credible evidence reasoning device for multi-source data fusion in one embodiment; Figure 3 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0019] 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.

[0020] In one embodiment, Figure 1 As shown, a credible evidence reasoning method for multi-source data fusion is provided, including the following steps: Step 102: construct a multi-source decision information system.

[0021] In this step, the multi-source decision information system is composed of multiple independent single-source decision information systems. The single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data used for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices.

[0022] Step 104, divide the multi-source decision information system into multiple data matrices according to attributes, and judge the association relationship between single-source decision information systems according to the decision feature set. When there is an association relationship, calculate the total similarity between the elements in the data matrix and other similar elements. When there is no association, calculate the total distance between the elements in the data matrix and other dissimilar elements.

[0023] In this step, a support matrix is ​​designed to quantify the correlation between attributes of multi-source data, combining the concepts of similarity and distance to evaluate the relationship. First, multi-source data of different scales are converted into confidence distributions to give them probabilistic meaning.

[0024] Step 106, judging the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculating the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support.

[0025] Step 108, generating evidence of a single-source decision information system based on the confidence level.

[0026] Step 110, establish an identification framework based on the evidence, and calculate the credibility of the evidence based on the weight and reliability of the evidence, and obtain the second confidence level and basic probability mass distribution of the evidence based on the credibility.

[0027] Step 112, establishing a combined confidence between any two pieces of evidence according to the basic probability mass distribution and the second confidence, and obtaining the confidence distribution of multiple pieces of evidence in the multi-source decision information system according to the combined confidence.

[0028] In the above-mentioned credible evidence reasoning method of multi-source data fusion, a multi-source decision information system is constructed. In this multi-source decision information system, a support matrix is ​​designed to quantify the correlation between multi-source data attributes, and the concepts of similarity and distance are combined to evaluate the relationship. First, multi-source data of different scales are converted into confidence distributions and given probabilistic meanings, so that the generated evidence can be applied to multi-scale data fusion. Then, a credible evidence reasoning rule is constructed, in which the credibility of evidence is determined by its weight and reliability. By integrating multi-source data samples of different scales, the confidence distribution of the results is derived. This solves the problem of difficulty in determining the confidence distribution of evidence.

[0029] In one embodiment, the multi-source decision information system is composed of multiple independent single-source decision information systems, which can be described as: ; in, represents a single-source decision information system; Include data samples, ; Include properties, ; is the data matrix consisting of all data samples, ; express middle No. A data matrix, . is a mapping function that establishes , and The relationship between. is the decision feature set, .

[0030] Table 1 shows a general MSDIS. Data samples that share the same attributes between different data sources may be considered to be at different scales. From the attribute perspective, Can also be divided into A data matrix.

[0031] Table 1 A generic MSDIS

[0032] In one embodiment, the multi-source decision information system is divided into multiple data matrices according to attributes. , data matrix It can be expressed as , the dimension is , in , the data matrix is ​​represented as: ; in, Representation attributes The corresponding data matrix is, Each row of represents a sample, Each column represents a single-source decision information system; the elements in the data matrix are normalized as follows: ; Use the normalized elements to represent the data matrix to get the normalized data matrix .

[0033] Table 1 shows the association between different data samples and different decision features. In order to quantify the relationship between different data sources, in one embodiment, the association between single-source decision information systems is determined based on the decision feature set: ; in, represents a data sample, represents the decision features in the decision feature set, , .

[0034] When there is an association relationship, the total similarity between the element in the data matrix and other similar elements is calculated as: ; When there is no association, the total distance between an element in the data matrix and other dissimilar elements is calculated as: .

[0035] In another embodiment, the data matrix Elements in The support for the information system is defined as the sum of the total similarity between the element and similar elements and the total distance between dissimilar elements, which is described as follows: ; Through normalization, we can get the support matrix , which contains each element in Support in By normalization For each row of In Properties Relative to the data source The first confidence level of is: .

[0036] In one embodiment, according to the first confidence level, the evidence for generating a single-source decision information system is: ; in, Represents data samples In Properties The evidence provided above and It is worth noting that confidence is only assigned to a single data source, and mixed data sources are not within the scope of this application.

[0037] In one embodiment, the ER rule was originally proposed based on the Bayesian rule, the DS rule and the ER algorithm. It is built on the so-called Framework of Discrimination (FoD), which is described as follows: ; in, It is A proposition, and , in the context of MSDIS, Equivalent to a single-source decision information system.

[0038] Once evidence is obtained from different data sources, whether and to what extent to trust the evidence remains an open question. Therefore, before making inferences, some consensus needs to be reached: 1) Evidence To some extent, it is credible, which can be measured by the credibility index ( ) to measure.

[0039] 2) Evidence Weight contribute to its unreliability; in particular, It can characterize the degree to which evidence provides accurate judgments or assertions.

[0040] 3) Evidence Reliability Contributes to its unreliability; here, Defined as the degree to which evidence accurately reflects valid conclusions, results, or assessments.

[0041] In one embodiment, the credibility of the evidence is calculated based on the weight and reliability of the evidence: .

[0042] The basic principle of the credibility formula is that when the reliability of evidence reaches the maximum value, the evidence is considered to be completely credible. On the contrary, when the reliability of evidence is lower than 1, its credibility increases with the weight of the evidence. increased with the increase of .

[0043] Therefore, the first confidence can be rewritten as the second confidence, described as: ; in, and The evidence is Assigned to and The probability mass of . is a power set, containing of The subsets are as follows: ; In the above analysis, the evidence The relevant untrustworthiness is assigned to the power set and is calculated as ( ), the basic principle behind it is that reflects global uncertainty or unknown factors, which can indirectly reveal The degree of skepticism about the proposition.

[0044] Therefore, the evidence The basic probability mass distribution of can be expressed as: ; in, .

[0045] In one embodiment, based on the description of the above identification framework, there are two pieces of evidence and , whose weight and reliability are and . and The combination follows the following process: ; ; ; ; ; In the above formula, and Respectively express evidence Assigned to and The probability mass of . and Assigned to and The unnormalized combined probability mass of is generated by the orthogonal sum operator. and Assigned to and Normalized combined probability mass. After the combination is assigned to The confidence level, , obviously, .

[0046] If you have The confidence distribution of the combined result can be generated by iteratively applying the above calculation process, as follows: ; in, is the joint credibility of all the evidence. It's essentially a new piece of evidence.

[0047] property The optimal data source or data scale can be determined by the following formula: .

[0048] The data samples of the optimal data sources corresponding to each attribute can be selected from MSDIS and the final integrated system can be constructed. In theory, this system ensures a more comprehensive consideration of the unique characteristics of each data source and improves the credibility and reliability of the information system by integrating different data sources and data scales and combining the selection of high-quality data.

[0049] The following takes the communication guarantee capability evaluation of the communication guarantee UAV as an example to illustrate the application process of the invented method. The communication coverage radius (km), maneuvering speed (km / h), and load capacity (KG) of the UAV are selected as evaluation indicators, and three different sensors are used for data collection. The three sensors collect the communication coverage radius, maneuvering speed, and load capacity respectively. When establishing a multi-source decision information system, the single-source decision information system generates sample data from a single sensor, and the processed simulation data is shown in Table 2.

[0050] Table 2 UAV communication capability evaluation test data

[0051] In Table 2, the three attributes Corresponding to 3 evaluation indicators, communication coverage radius, maneuvering speed and load capacity, 3 data sources Corresponding to the sensors with three different precisions, the three data samples are recorded as , the three decision features {1, 2, 3} represent the data accuracy as “high”, “medium” and “low” respectively. For example, the data matrix is ​​as follows: ; This gives the normalized data matrix As shown below: ; As can be seen from Table 2, the three groups of samples are all different samples. The total similarity and total distance can be calculated as follows: ; ; therefore, The support degree of is 1.67; by analogy, the support matrix can be obtained as follows: ; This allows us to determine the properties The confidence level relative to each data source over all data samples. For example: ; Therefore, the basic credibility matrix can be obtained: ; In the above formula, Each row of reflects the evidence for a data sample. Assume , and The weights are Then, the reliability of each piece of evidence can be calculated as follows: ; ; Therefore, the basic probability distribution is as follows: ; The above formula shows that the credibility of each piece of evidence is as high as 0.5456, 0.5452 and 0.5456 respectively. By using the evidence reasoning rule, the fusion result can be described as follows: ; It can be clearly seen that the credibility of the fusion result is significantly improved compared with any single initial evidence. This also shows that the ER rule can effectively improve the credibility of the reasoning result by integrating evidence from different data sources.

[0052] Similarly, for the attribute and , the fusion results are as follows: ; According to the above results, , and The optimal data sources are , and Therefore, the final fusion is shown in Table 3.

[0053] Table 3 Final fusion system

[0054] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0055] In one embodiment, Figure 2 As shown, a credible evidence reasoning device for multi-source data fusion is provided, including: a multi-source decision system construction module 202, an evidence generation module 204, a basic probability mass distribution calculation module 206 and a fusion module 208, wherein: The multi-source decision system construction module 202 is used to construct a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; The evidence generation module 204 is used to divide the multi-source decision information system into multiple data matrices according to the attributes, and to judge the association relationship between the single-source decision information systems according to the decision feature set; when there is an association relationship, calculate the total similarity between the elements in the data matrix and other similar elements; when there is no association, calculate the total distance between the elements in the data matrix and other dissimilar elements; judge the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generate evidence of the single-source decision information system according to the first confidence; A basic probability mass distribution calculation module 206 is used to establish an identification framework based on the evidence, and calculate the credibility of the evidence according to the weight and reliability of the evidence, and obtain a second confidence level and a basic probability mass distribution of the evidence according to the credibility; The fusion module 208 is used to establish a combined confidence between any two pieces of evidence according to the basic probability mass distribution and the second confidence, and obtain the confidence distribution of multiple pieces of evidence in the multi-source decision information system according to the combined confidence.

[0056] For the specific definition of the trusted evidence reasoning device for multi-source data fusion, please refer to the definition of the trusted evidence reasoning method for multi-source data fusion mentioned above, which will not be repeated here. Each module in the above-mentioned trusted evidence reasoning device for multi-source data 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.

[0057] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a credible evidence reasoning method for multi-source data fusion is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0058] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0059] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in the above embodiment when executing the computer program.

[0060] In one embodiment, 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 method in the above embodiment are implemented.

[0061] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant 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 many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0062] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above 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.

[0063] 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 understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications 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 present application shall be subject to the attached claims.

Claims

1. A credible evidence reasoning method for multi-source data fusion, characterized in that: The method comprises: Constructing a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; Dividing the multi-source decision information system into multiple data matrices according to attributes, and judging the association relationship between single-source decision information systems according to the decision feature set, when there is an association relationship, calculating the total similarity between the elements in the data matrix and other similar elements, and when there is no association, calculating the total distance between the elements in the data matrix and other dissimilar elements; Determine the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generating evidence for a single-source decision information system based on the first confidence level; According to the evidence, an identification framework is established, and according to the weight and reliability of the evidence, the credibility of the evidence is calculated, and according to the credibility, a second confidence level and a basic probability mass distribution of the evidence are obtained; A combined confidence between any two pieces of evidence is established according to the basic probability mass distribution and the second confidence, and the confidence distribution of multiple pieces of evidence in the multi-source decision information system is obtained according to the combined confidence.

2. The method according to claim 1, characterized in that The multi-source decision information system is divided into a plurality of data matrices according to attributes, including: The multi-source decision information system is divided into a plurality of data matrices according to attributes, and the data matrices are represented as follows: in, Representation attributes The corresponding data matrix is, Each row of represents a sample, Each column of represents a single-source decision information system; The elements in the data matrix are normalized to be expressed as: The normalized data matrix is ​​obtained by using the normalized elements to represent the data matrix.

3. The method according to claim 2, characterized in that The association relationship between the single-source decision information systems is determined according to the decision feature set. When an association relationship exists, the total similarity between the element in the data matrix and other similar elements is calculated. When no association exists, the total distance between the element in the data matrix and other dissimilar elements is calculated, including: According to the decision feature set, the association relationship between single-source decision information systems is determined as follows: in, represents a data sample, represents the decision features in the decision feature set, , ; When there is an association relationship, the total similarity between the element in the data matrix and other similar elements is calculated as: ； When there is no association, the total distance between an element in the data matrix and other dissimilar elements is calculated as: 。 4. The method according to claim 3, characterized in that: According to the total similarity and the total distance, the support of the elements in the data matrix to the data matrix is ​​determined, and according to the support, the first confidence of the data sample in the attribute relative to the single-source decision information system is calculated, including: According to the total similarity and the total distance, the support of the elements in the data matrix to the data matrix is ​​determined as: The first confidence of the data sample in terms of attribute relative to the single-source decision information system is calculated based on the support: 。 5. The method according to claim 4, characterized in that Based on the first confidence level, evidence of a single-source decision-making information system is generated, including: According to the first confidence level, the evidence for generating a single-source decision information system is: in, Represents data samples In Properties The evidence provided above and .

6. The method according to claim 5, characterized in that Based on the evidence, an identification framework is established, and based on the weight and reliability of the evidence, the credibility of the evidence is calculated. Based on the credibility, a second confidence level of the evidence and a basic probability mass distribution are obtained, including: Based on the evidence, establish an identification framework ; It represents a single-source decision information system; According to the weight and reliability of the evidence, the credibility of the evidence is calculated as: in, and Evidence The weight and reliability of For evidence Credibility; According to the confidence level, the second confidence level is calculated as: Among them, evidence The basic probability distribution of is: in, and The evidence is Assigned to and The probability mass of is a power set, represented by: ； According to the basic probability, we get evidence The basic probability mass distribution of is: in, .

7. The method according to claim 6, characterized in that Establishing a combined confidence between any two pieces of evidence according to the basic probability mass distribution and the second confidence, and obtaining the confidence distribution of multiple pieces of evidence in the multi-source decision information system according to the combined confidence, including: The combined confidence between any two pieces of evidence established according to the basic probability mass distribution and the second confidence is: in, , , , ; and Respectively express evidence Assigned to and The probability mass of and Assigned to and The unnormalized combined probability mass of and Assigned to and The normalized combined probability mass, is assigned to confidence level; According to the combined confidence, the confidence distribution of multiple pieces of evidence in the multi-source decision information system is obtained as follows: ； is the joint credibility of all the evidence.

8. A credible evidence reasoning device for multi-source data fusion, characterized in that: The device comprises: A multi-source decision system building module is used to build a multi-source decision information system; the multi-source decision information system is composed of multiple independent single-source decision information systems, the single-source decision information system generates data for artificial intelligence training, and the multi-source decision information system is used to integrate the data for artificial intelligence training; the single-source decision information system is composed of data samples, attributes, data matrices, mapping functions and decision feature sets; the mapping function establishes a mapping relationship between data samples, attributes and data matrices; An evidence generation module is used to divide the multi-source decision information system into multiple data matrices according to attributes, and to judge the association relationship between single-source decision information systems according to the decision feature set; when there is an association relationship, calculate the total similarity between the elements in the data matrix and other similar elements; when there is no association, calculate the total distance between the elements in the data matrix and other dissimilar elements; judge the support of the elements in the data matrix to the data matrix according to the total similarity and the total distance, and calculate the first confidence of the data sample in terms of attributes relative to the single-source decision information system according to the support; generate evidence of the single-source decision information system according to the first confidence; A basic probability mass distribution calculation module is used to establish an identification framework based on the evidence, and calculate the credibility of the evidence according to the weight and reliability of the evidence, and obtain a second confidence level and a basic probability mass distribution of the evidence according to the credibility; A fusion module is used to establish a combined confidence between any two pieces of evidence according to the basic probability mass distribution and the second confidence, and obtain the confidence distribution of multiple pieces of evidence in the multi-source decision information system according to the combined confidence.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

Images