Multi-source communication data fusion method and system and storage medium

Abstract

The invention relates to the technical field of multi-source data fusion, in particular to a multi-source communication data fusion method and system and a storage medium, and the method comprises the steps: obtaining various data of communication equipment in real time; acquiring a monitoring deviation index of any two types of data in the current period according to data difference characteristics between any two types of data, and clustering all types of data; according to the data deviation degree between each type of data and all other types of data in the cluster in which the data is located in the current period, the deviation degree of each type of data relative to historical data, and the synchronization degree of the mutation time of each type of data, obtaining a cheating factor of each type of data in the current period; therefore, the initial credibility of various types of data in the current period is optimized, and then all types of data in the current period are fused. According to the method, the data credibility of each type of data is analyzed, the initial credibility of each type of data is optimized, and the accuracy of multi-source communication data fusion is improved.

Description

Technical Field

[0001] This application relates to the field of multi-source data fusion technology, specifically to a multi-source interconnected data fusion method, system, and storage medium. Background Technology

[0002] In modern information-based warfare, information sources are becoming increasingly diverse, and the amount of information is gradually increasing. Data sensing and processing from a single sensor can no longer meet the needs of integrated electronic systems in future combat environments. Multi-sensor information fusion has become a crucial objective for combat platforms to achieve complex information sensing and processing. Therefore, constructing a method for fusing multi-source communication data is of great significance for improving the efficiency of complex information sensing and processing and enhancing the authenticity of data.

[0003] Due to the complex training environment, both sides typically employ communication jamming and jamming techniques, making communication data collected by sensors often deceptive. Therefore, it's necessary to fuse multi-sensor data to improve its authenticity. Compared to other data fusion algorithms, Dempster-Shafer (DS) evidence theory effectively analyzes incomplete and uncertain information, thus enabling more efficient multi-source data fusion.

[0004] However, when existing technologies fuse multi-source communication data using DS evidence theory, they do not fully consider that strong electromagnetic interference during adversarial exercises may cause synchronous mutations in multiple types of data. This could lead to a decrease in the trust level of undisturbed normal data due to deviations from the group's change trend. Consequently, the trust level autonomously obtained by DS evidence theory for various types of data may be inaccurate, thus affecting the accuracy of subsequent multi-source data fusion. Summary of the Invention

[0005] To address the aforementioned technical problems, the purpose of this application is to provide a method, system, and storage medium for multi-source interconnected data fusion, the specific technical solution of which is as follows: In a first aspect, embodiments of this application provide a method for fusing multi-source communication data, the method comprising the following steps: Real-time acquisition of various data from communication equipment in combat simulation exercise scenarios; The entire data collection time is divided into several periods; based on the dispersion and average level of the data difference characteristics between any two types of data in the current period and the previous preset number of historical periods, the monitoring deviation index of any two types of data in the current period is obtained. Cluster all data types based on the monitoring deviation index of any two data types within the current period; obtain the distrust index of each data type in the current period based on the average level of the monitoring deviation index between each data type and all other data types in its cluster, as well as the degree of deviation of each data type from the historical data trend. The system obtains the mutation moments of various data types within the current period and clusters all mutation moments. Based on the number of elements in the clusters where the mutation moments of various data types within the current period are located, and the non-trust index, it obtains the deception factor of various data types within the current period. This optimizes the initial trust level of various data types within the current period, and then merges all data types within the current period.

[0006] Preferably, the method for obtaining the monitoring deviation index of any two types of data in the current period is as follows: Calculate the absolute difference in the number of data collections between any two types of data in the current period, and the DTW distance; The sum of the normalized value of the absolute difference and the normalized value of the DTW distance is recorded as the first deviation of any two types of data in the current period; Calculate the variance and mean of the first deviation of any two types of data in the current period and the number of historical periods preceding it; The monitoring deviation index of any two types of data in the current period is positively correlated with the variance and the mean.

[0007] Preferably, the specific process of clustering all types of data is as follows: By using the categories of various data as nodes and the monitoring deviation index between any two categories of data as the weight of the edge between the corresponding two nodes, a weighted undirected graph for the current period is constructed. The weighted undirected graph of the current period is used as input to a graph clustering algorithm to divide all data categories of the current period into multiple clusters.

[0008] Preferably, the method for obtaining the distrust index of various types of data in the current period is as follows: The average of the monitoring deviation index between each type of data and all other types of data in the current period is calculated. The prediction model is trained by using various types of data from the previous preset number of historical periods as input, and then the trained prediction model is used to predict various types of data in the current period. The mean of the residuals between all actual data and their corresponding predicted data in the current period for various types of data; The distrust index of each type of data in the current period is positively correlated with the two means mentioned above.

[0009] Preferably, the method for obtaining the abrupt change moments in various types of data within the current period is as follows: Retrieve all mutation points in various types of data for the current period; The absolute difference between each mutation point and the previous element is recorded as the second difference of each mutation point. The moment of the mutation point with the largest second difference among all types of data in the current period is recorded as the mutation moment of all types of data in the current period.

[0010] Preferably, the method for obtaining the deception factor of each type of data in the current period is as follows: The ratio of the total number of elements in the cluster where the abrupt change of various types of data occurs within the current period to the maximum total number of elements in all clusters; The deception factor of each type of data in the current period is positively correlated with the ratio and the distrust index.

[0011] Preferably, the calculation formula for optimizing the initial trust level of various types of data in the current period is as follows: In the formula, The optimized trust level for the i-th type of data in the current period; The initial level of trust in the autonomous acquisition of the i-th type of data in the current period is defined by the DS evidence theory. Let be the deception factor for the i-th type of data in the current period; This is a preset constant, and its value range is [0.005, 0.01].

[0012] Preferably, the specific process of fusing all types of data in the current period is as follows: using the Softmax function to map the optimized confidence level of all types of data in the current period to a probability value; taking each type of data in the current period as a piece of evidence, taking all the evidence in the current period as the input of the DS evidence theory, and taking the probability value mapped by the optimized confidence level of each type of data in the current period as the confidence level of each piece of evidence, thereby obtaining the multi-source data fusion result in the current period.

[0013] Secondly, embodiments of this application provide a multi-source communication data fusion system, which includes: a data acquisition module, a category deviation assessment module, a trust optimization module, and a multi-source data fusion module.

[0014] The data acquisition module is used to acquire various types of data from communication equipment in real time during combat simulation exercises. The category deviation assessment module is used to obtain the monitoring deviation index of any two categories of data in the current period based on the data difference characteristics between any two categories of data in the current period. The trust optimization module is used to obtain the deception factor of each type of data in the current period based on the degree of deviation between each type of data and other types of data in its cluster, the degree of deviation of each type of data relative to historical data, and the degree of synchronization of the mutation time of each type of data, so as to optimize the initial trust of each type of data in the current period. The multi-source data fusion module is used to fuse all types of data in the current period.

[0015] Thirdly, embodiments of this application also provide a multi-source communication data fusion storage medium. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described multi-source communication data fusion methods.

[0016] As can be seen from the above embodiments, the multi-source communication data fusion method, system, and storage medium provided in this application have at least the following beneficial effects: This application addresses the problem that existing technologies do not adequately consider non-causal instantaneous synchronous mutations caused by electromagnetic interference, which can lead to deviations of normal data from the population's changing trends and thus reduce trust. By analyzing the data difference characteristics between any two types of data within the current period, a monitoring deviation index is constructed for any two types of data in the current period. This allows for the assessment of whether there is an inherent correlation between different types of data based on their changing characteristics, thereby classifying all types of data. Furthermore, by analyzing the degree of deviation between each type of data and other types of data in its cluster, as well as the degree of deviation of each type of data relative to historical data, a non-trust index is constructed, enabling a preliminary assessment of the trustworthiness of each type of data. Finally, by combining the synchronization degree of mutation times of each type of data, a deception factor is constructed, allowing for a precise assessment of the trustworthiness of each type of data in conjunction with population synchronization behavior. This optimizes the trustworthiness of various types of data in the DS evidence theory and improves the accuracy of multi-source data fusion. Attached Figure Description

[0017] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the steps of a multi-source communication data fusion method provided in one embodiment of this application; Figure 2 This is a schematic diagram of the structure of a multi-source communication data fusion system provided in one embodiment of this application. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a multi-source interconnected data fusion method, system, and storage medium proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise specified and limited, terms such as “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a circuit structure, article, or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the article or device that includes said element. Furthermore, the term “and / or” as used herein includes any and all combinations of one or more of the associated listed items. All technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0021] The following description, in conjunction with the accompanying drawings, details a specific scheme for a multi-source interconnected data fusion method, system, and storage medium provided in this application.

[0022] Please see Figure 1 The diagram illustrates a flowchart of a multi-source communication data fusion method according to an embodiment of this application, which includes the following steps: Step 1: Acquire various data from communication equipment in real time during combat simulation exercises.

[0023] This application is applied to a combat simulation exercise scenario. During the exercise, various data from communication equipment are collected in real time. The specific data collection methods are as follows: This application deploys network quality probes at the nodes of the communication link of the communication equipment to monitor and collect data on the signal-to-noise ratio, error rate, latency, and packet loss rate of the communication link in real time; it collects signal strength and spectrum occupancy data of the communication equipment in the preset communication frequency band using a spectrum analyzer; it obtains the location coordinates of each communication device in real time using a GPS sensor, and records the Euclidean distance between the location coordinates at the current time and the previous time as the moving distance of each communication device at the current time; and it collects call duration and call interruption rate data of fixed and mobile communications in real time through a program-controlled exchange.

[0024] All data is collected in real time, and the entire data collection time is divided into several cycles according to a preset duration. In this embodiment, the preset duration is set to 10 minutes.

[0025] Because the collected data is heterogeneous with significant differences in dimensions, maximum value normalization was used to normalize various data types from the communication link, communication frequency band, mobile distance, call duration, and call interruption rate to eliminate the influence of these dimensions. Maximum value normalization is a well-known technique, and its specific process will not be elaborated upon here.

[0026] Step 2: Divide the entire data collection time into several periods; based on the dispersion and average level of the data difference characteristics between any two types of data in the current period and the preset number of historical periods before it, obtain the monitoring deviation index of any two types of data in the current period.

[0027] It should be noted that in a physical battlefield, the responses of different sensors to the same real event typically exhibit a causal delay in time (e.g., detecting the signal before calculating the trajectory). Strong electromagnetic interference often causes non-causal, microsecond-level instantaneous synchronization jumps in various levels of the communication link. Therefore, the 'synchronization mutation' in this application specifically refers to this type of non-physically governed instantaneous synchronization characteristic, used to identify electromagnetic interference.

[0028] Because sensor data can be affected by interference from other devices, leading to errors in the monitoring data, data fusion techniques are needed to verify the authenticity of the collected data by utilizing the similarity or complementarity between data points, thereby improving the accuracy and reliability of subsequent decision-making. Furthermore, since the collected data comes from multiple sources, fusing and analyzing each source individually would not only affect fusion efficiency but may also result in the fusion of different data types, leading to errors. Therefore, to improve fusion efficiency and analytical accuracy, it is first necessary to perform preliminary classification of the multi-source data based on the characteristics of each data type and the relationships between them.

[0029] Let's take the i-th and j-th types of data in the current period as examples for analysis.

[0030] In the monitoring of communication links, although the methods for collecting different data items differ, if the monitoring targets the same objective, or if it involves monitoring data from the same source from different objectives (e.g., signal strength and spectrum occupancy data corresponding to the same communication frequency band; or signal strength data from two different communication frequency bands), the collection frequency will usually be relatively consistent and not significantly different. Furthermore, when the monitored objective changes, the monitoring data for that objective will typically change in tandem. For example, when the network quality deteriorates, the signal-to-noise ratio and bit error rate data will also change synchronously. Therefore, the consistency of the data collection frequency and the synchronicity of dynamic evolution can be used to initially analyze whether the i-th and j-th types of data have internal correlation characteristics.

[0031] Calculate the absolute difference between the number of data collections for the i-th and j-th data types in the current period, and denote it as the first difference between the i-th and j-th data types. The first difference reflects the degree of difference in the collection frequency between the i-th and j-th data types; the smaller the value, the more consistent the collection mechanisms of the two data types are, and the more likely it is that they are conducting multi-dimensional monitoring of the same target.

[0032] Calculate the DTW distance between the i-th type of collected data and the j-th type of collected data in the current period, and denote it as the first distance between the i-th and j-th types of data. The first distance can reflect whether the inherent change patterns of the i-th and j-th types of data have a strong correlation; the smaller the value, the more similar the change trends and fluctuation patterns of the two types of data are, and the more likely they are monitoring the same target or are data from the same source for different targets.

[0033] It should be noted that, considering that the number of data collected from the i-th class and the j-th class may not be the same, the DTW distance is used here to analyze the correlation of the data.

[0034] Since the numerical difference between the first difference and the first distance may be too large, resulting in the weak effect of the other parameter, the first difference and the first distance between any two classes of data in the current period are calculated, and the minimum-maximum normalization method is used to normalize all the first differences and all the first distances respectively to obtain the normalized first difference and the first distance between the i-th class and the j-th class of data.

[0035] The sum of the normalized value of the first difference between the i-th and j-th data in the current period and the normalized value of the first distance is recorded as the first deviation between the i-th and j-th data in the current period. The first deviation reflects the degree of deviation between the two types of data in terms of the consistency of the acquisition mechanism and the degree of deviation in response synchronization in the current period; the smaller the value, the greater the possibility that the two types of data are multi-dimensional monitoring data of the same target and are affected by the same change, and the more likely they are to have the basic conditions for fusion analysis.

[0036] Furthermore, considering that data changes generated during normal combat maneuvers in combat exercises typically exhibit temporal causality, but when strong interference exists such as electromagnetic countermeasures, intensive equipment start-ups and shutdowns, and frequent switching of maneuver nodes, various types of data collected within a single cycle are easily affected by interference, resulting in non-causal synchronous mutations. Since historical cycles usually encompass multiple operating conditions, their statistical patterns better reflect the inherent intrinsic relationships between data; therefore, further analysis can be conducted by combining historical cycles.

[0037] In a preferred embodiment, the monitoring deviation index of any two types of data in the current period is obtained based on the dispersion and average level of the data difference characteristics between the two types of data in the current period and the previous preset number of historical periods. The method for obtaining the monitoring deviation index of any two types of data in the current period is as follows: calculate the variance and mean of all first deviations between any two types of data in the current period and the previous T historical periods, respectively. The monitoring deviation index of any two types of data in the current period is positively correlated with both the variance and the mean. Here, T is the first preset number, which is 15 in this embodiment. The positive correlation means that the dependent variable increases (decreases) as the independent variable increases (decreases).

[0038] Preferably, in this embodiment, the product of the variance and the mean of the i-th and j-th data is recorded as the monitoring deviation index of the i-th and j-th data in the current period.

[0039] The monitoring deviation index combines the degree of difference between two types of data and their historical trends to comprehensively reflect whether the two types of data are correlated. The smaller the value, the more stable the deviation is in the long-term trend, indicating that the two types of data are more likely to originate from the same monitoring target or be affected by the same combat event, and are more suitable for fusion analysis.

[0040] Step 3: Cluster all data types based on the monitoring deviation index of any two data types within the current period; obtain the distrust index of each data type in the current period based on the average level of the monitoring deviation index between each data type and all other data types in its cluster, as well as the degree of deviation of each data type from the historical data trend.

[0041] Furthermore, considering that electromagnetic suppression often leads to non-causal, instantaneous synchronous abrupt changes in indicators at various levels of the communication link, including high-error data in the fusion calculation would weaken the reliability of the overall fusion result. Therefore, it is necessary to perform credibility analysis on various types of data in the current period to generate credibility weights, thereby more reasonably optimizing the trust function of the DS theory and improving the reliability of subsequent data fusion results.

[0042] Based on the calculation principle of the monitoring deviation index, the monitoring deviation index between any two types of data in the current period is calculated; then, the categories of each type of data are used as nodes, and the monitoring deviation index between any two types of data is used as the weight of the edge between the corresponding two nodes, thereby constructing a weighted undirected graph for the current period; finally, the weighted undirected graph for the current period is used as the input of a graph clustering algorithm to cluster all nodes, and the output of the clustering algorithm is multiple clusters.

[0043] Graph clustering algorithms are not limited to Markov graph clustering or spectral clustering; graph clustering is a well-known technique and will not be elaborated upon here. Compared to other clustering algorithms, graph clustering can more naturally capture the correlation between data categories based on the degree of monitoring deviation between them. This helps to more reasonably identify data categories originating from the same monitoring target or affected by the same event, thereby enhancing the accuracy and reliability of trust function optimization during data fusion.

[0044] Let's take the i-th data type as an example for analysis. First, calculate the mean of the monitoring deviation index between the i-th data type and all other data types in its cluster in the current period, and record it as the first mean of the i-th data type in the current period. Then, select U (U is a second preset number, 10 in this embodiment) undisturbed periods from the historical periods and record them as historical stable periods. Calculate the first mean of the i-th data type in each historical stable period, and calculate the maximum value of the first mean of the i-th data type in all historical stable periods and the preset minimum constant. The sum of the values; the ratio between the first mean of the i-th type of data in the current period and the sum is recorded as the relative deviation of the i-th type of data in the current period. It should be noted that a preset minimum constant is added. The purpose is to prevent the denominator from being zero when calculating the relative deviation. In this embodiment... The value is 0.0001. The relative deviation reflects whether the synchronization consistency between the i-th type of data and other data in the same cluster has significantly deteriorated in the current period compared with historical stable data. The larger the value, the greater the deviation of the i-th type of data from the cooperative change pattern in the current period, the more likely it has been affected by a large error, and the lower the reliability.

[0045] However, considering that if only the i-th data in the cluster is real data, while the other data categories are subject to significant external interference, the relative deviation of the i-th data in the current period may be large. In such cases, the relative deviation will have a large error, so further analysis is needed.

[0046] The i-th type of data from the previous T historical periods of the current period is used as input to train the prediction model, and the i-th type of data in the current period is predicted based on the trained prediction model. Then, the mean of the residuals between all real collected data and their corresponding predicted data of the i-th type of data in the current period is calculated, and denoted as the prediction residual of the i-th type of data in the current period.

[0047] Calculate the prediction residuals of all other classes in the cluster containing the i-th data type in the current period using the same method; then calculate the maximum value among the prediction residuals of all other classes in the cluster containing the i-th data type and the preset minimum constant. The sum of the predicted residuals of the i-th data in the current period is used as the ratio of the predicted residuals to the sum. The residual ratio reflects the relative severity of the prediction bias of the i-th data compared to the prediction bias of other data in the same cluster. The larger the value, the greater the trend difference between the i-th data and historical data compared to other data in the same cluster, and thus the greater the possibility that the i-th data has been severely disturbed in the current period.

[0048] The neural networks that can be used to build prediction models include, but are not limited to, BP neural networks, LSTM networks, and GRU networks. Activation functions used for model training include, but are not limited to, the Sigmoid function, the tanh function, and the ReLU function. Optimization algorithms include, but are not limited to, SGD and Adam. Loss functions include, but are not limited to, MSE, RMSE, and MAE. Training prediction models is a well-known technique and will not be elaborated upon here.

[0049] As a preferred implementation, based on the average level of the monitoring deviation index between each type of data and all other types of data in its cluster within the current period, and the degree of deviation of each type of data from its historical data trend, a distrust index for each type of data in the current period is obtained. This index is used to characterize the degree of untrustworthiness of each type of data in the current period. The method for obtaining the distrust index of each type of data in the current period is as follows: calculating the first mean of each type of data in the current period; calculating the prediction residual of each type of data in the current period; the distrust index of each type of data in the current period is positively correlated with both the first mean and the prediction residual.

[0050] Preferably, in this embodiment, the distrust index of the i-th type of data in the current period is denoted as... Its specific expression is: In the formula, Let be the distrust index of the i-th type of data in the current period; , , respectively, represent the relative deviation and residual ratio of the i-th type of data in the current period.

[0051] The distrust index is used to reflect the distrust level of the i-th type of monitoring data in the current period. The larger the value, the more it indicates that in the current period, the i-th type of data not only deviates significantly from other types of data in the same cluster in terms of response coordination, but also deviates significantly from historical behavior patterns in its own evolution. This means that the i-th type of data is more likely to be affected by external interference in the current period, resulting in a large error. Therefore, the trust level of the i-th type of data in the current period should be lower to avoid affecting the reliability of data fusion.

[0052] Step 4: Obtain the mutation moments of various data types in the current period and cluster all mutation moments; based on the number of elements in the clusters where the mutation moments of various data types in the current period are located, and the non-trust index, obtain the deception factor of various data types in the current period, thereby optimizing the initial trust level of various data types in the current period, and then merging all data types in the current period.

[0053] Furthermore, considering that the electromagnetic shielding and jamming equipment used in combat exercises typically has a wide range of effects, its impact is often not limited to a single sensor or data type. Therefore, analyzing whether significant abnormal changes occurred simultaneously in multiple data categories within the same timeframe can reflect whether the monitoring data was subjected to widespread electromagnetic interference from the exercise equipment, thereby further assessing the reliability of the data.

[0054] Because the acquired data undergoes a significant and immediate change after being subjected to electromagnetic interference, and the degree of this change is significantly greater than the normal variation in the data, we will continue to analyze the data of type i as an example. A mutation point detection algorithm is used to obtain all mutation points within type i of the current period; then, the absolute difference between each mutation point and the previous element is calculated, denoted as the second difference for each mutation point. The second difference reflects the drastic degree of the instantaneous change at each mutation point; the larger the value, the more drastic the data jump at that mutation point, and the more likely it is caused by strong external electromagnetic interference.

[0055] The time of the mutation point with the largest second difference in the i-th data category within the current period is recorded as the mutation time of the i-th data category in the current period. The mutation times of all data categories in the current period are calculated in the same way, and all mutation times are used as input to the clustering algorithm for clustering. The mutation point detection algorithm is not limited to PELT or Pettitt; the clustering algorithm is not limited to K-means, DPC, or DBSCAN. This embodiment uses the K-means algorithm and obtains the optimal number of clusters through the elbow rule. It should be noted that if no mutation point is detected within a certain data category, that data category will not participate in clustering.

[0056] The number of internal elements in each cluster is denoted as the synchronization count of that cluster. The ratio of the synchronization count of the cluster to which the i-th data belongs at the moment of mutation to the maximum value among all clusters' synchronization counts is denoted as the relative proportion of the i-th data in the current period. The relative proportion of the i-th data reflects the number of data categories that experienced abnormal mutations synchronously with the i-th data. The larger the value, the more synchronized the mutation of the i-th data is, and the more it conforms to the typical characteristics of strong electromagnetic interference—wide range and rapid response—and thus the lower the reliability of the data. Conversely, the smaller the value, the more likely the mutation of the i-th data is an isolated or local phenomenon, and the more likely it is to be a normal operating condition fluctuation, with relatively higher reliability.

[0057] As a preferred implementation, based on the number of elements in the cluster where the mutation time of each type of data occurs within the current period, and the distrust index of each type of data, a deception factor for each type of data in the current period is obtained to characterize the degree of untrustworthiness of each type of data in the current period. The method for obtaining the deception factor of each type of data in the current period is as follows: Statistically calculate the relative proportion of similar changes for each type of data in the current period; Statistically calculate the distrust index of each type of data in the current period; The deception factor of each type of data in the current period is positively correlated with both the relative proportion of similar changes and the distrust index.

[0058] Preferably, in this embodiment, the sum of the relative proportion of the same variation of the i-th type of data in the current period and the non-trust index is denoted as the deception factor of the i-th type of data in the current period. The larger the value, the more likely the i-th type of data is to be affected by strong electromagnetic interference in the current period, and the more its own change trend deviates from the historical change trend. This reflects that the i-th type of data is more likely to deviate from the real data and has a stronger deceptive nature. Therefore, when fusing data in the current period, the trust level of the i-th type of data should be lower to avoid affecting the reliability of subsequent data fusion and communication decisions.

[0059] Furthermore, when using the DS evidence theory to perform multi-source data fusion on various types of data in the current period, the trust level of various types of data in the current period is optimized based on the deception factor of each type of data in the current period. The specific optimization expression is as follows: In the formula, The optimized trust level for the i-th type of data in the current period; The initial confidence level is autonomously obtained by the algorithm for the i-th type of data in the current period, based on the DS evidence theory. The initial confidence level ranges from 0 to 1 and is obtained autonomously by the algorithm. Let be the deception factor for the i-th type of data in the current period; This is a preset constant, used not only to avoid a denominator of 0, but also to limit the calculation to a state without deception. The upper limit of the maximum trust gain under the condition that, i.e., for The magnification does not exceed Its value range is [0.005, 0.01], and in this embodiment, it is 0.008.

[0060] Following the optimization method for the trust level of the i-th type of data in the current period, the optimized trust levels of each type of data in the current period are obtained. Then, the Softmax function is used to map all optimized trust levels to probability values. Each type of data in the current period is treated as one piece of evidence, and all evidence in the current period is used as input to the DS evidence theory. The probability values ​​mapped by the optimized trust levels of each type of data in the current period are used as the trust levels of the corresponding pieces of evidence, thus obtaining the multi-source data fusion result for the current period. The multi-source data fusion using the DS evidence theory is a well-known technique, and the specific process will not be elaborated further.

[0061] By optimizing the trust level using the above methods, we can effectively suppress the misleading influence of interfered data categories on global judgments, thereby improving the fusion effect of multi-source communication data in complex electromagnetic environments and providing reliability for subsequent decisions.

[0062] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of a multi-source communication data fusion system provided in an embodiment of this application. In this embodiment, the modules included in the system are used to execute the steps in the embodiment corresponding to a multi-source communication data fusion method. See also... Figure 2 The multi-source interconnected data fusion system includes: a data acquisition module, a category deviation assessment module, a trust optimization module, and a multi-source data fusion module.

[0063] The data acquisition module is used to acquire various types of data from communication equipment in real time during combat simulation exercises. The category deviation assessment module is used to obtain the monitoring deviation index of any two categories of data in the current period based on the data difference characteristics between any two categories of data in the current period. The trust optimization module is used to obtain the deception factor of each type of data in the current period based on the degree of deviation between each type of data and other types of data in its cluster, the degree of deviation of each type of data relative to historical data, and the degree of synchronization of the mutation time of each type of data, so as to optimize the initial trust of each type of data in the current period. The multi-source data fusion module is used to fuse all types of data in the current period.

[0064] Based on the same inventive concept as the above method, this application embodiment also provides a multi-source communication data fusion storage medium, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the multi-source communication data fusion method described above.

[0065] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0066] It should be noted that, unless otherwise specified and limited, terms such as “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a circuit structure, article, or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such article or device. Without further limitations, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the article or device that includes said element. Furthermore, the term “and / or” as used herein includes any and all combinations of one or more of the associated listed items.

[0067] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not invented herein.

[0068] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for fusing multi-source communication data, characterized in that, The method includes the following steps: Real-time acquisition of various data from communication equipment in combat simulation exercise scenarios; The entire data collection time is divided into several periods; based on the dispersion and average level of the data difference characteristics between any two types of data in the current period and the previous preset number of historical periods, the monitoring deviation index of any two types of data in the current period is obtained. Cluster all data types based on the monitoring deviation index of any two data types within the current period; obtain the distrust index of each data type in the current period based on the average level of the monitoring deviation index between each data type and all other data types in its cluster, as well as the degree of deviation of each data type from the historical data trend. The system obtains the mutation moments of various data types within the current period and clusters all mutation moments. Based on the number of elements in the clusters where the mutation moments of various data types within the current period are located, and the non-trust index, it obtains the deception factor of various data types within the current period. This optimizes the initial trust level of various data types within the current period, and then merges all data types within the current period.

2. The multi-source communication data fusion method as described in claim 1, characterized in that, The method for obtaining the monitoring deviation index between any two types of data in the current period is as follows: Calculate the absolute difference in the number of data collections between any two types of data in the current period, and the DTW distance; The sum of the normalized value of the absolute difference and the normalized value of the DTW distance is recorded as the first deviation of any two types of data in the current period; Calculate the variance and mean of the first deviation of any two types of data in the current period and the number of historical periods preceding it; The monitoring deviation index of any two types of data in the current period is positively correlated with the variance and the mean.

3. The multi-source interconnected data fusion method as described in claim 1, characterized in that, The specific process of clustering all types of data is as follows: By using the categories of various data as nodes and the monitoring deviation index between any two categories of data as the weight of the edge between the corresponding two nodes, a weighted undirected graph for the current period is constructed. The weighted undirected graph of the current period is used as input to a graph clustering algorithm to divide all data categories of the current period into multiple clusters.

4. The multi-source communication data fusion method as described in claim 1, characterized in that, The methods for obtaining the distrust index of various data in the current period are as follows: The average of the monitoring deviation index between each type of data and all other types of data in the current period is calculated. The prediction model is trained by using various types of data from the previous preset number of historical periods as input, and then the trained prediction model is used to predict various types of data in the current period. The mean of the residuals between all actual data and their corresponding predicted data in the current period for various types of data; The distrust index of each type of data in the current period is positively correlated with the two means mentioned above.

5. The multi-source communication data fusion method as described in claim 1, characterized in that, The method for obtaining the abrupt change moments in various types of data within the current period is as follows: Retrieve all mutation points in various types of data for the current period; The absolute difference between each mutation point and the previous element is recorded as the second difference of each mutation point. The moment of the mutation point with the largest second difference among all types of data in the current period is recorded as the mutation moment of all types of data in the current period.

6. The multi-source communication data fusion method as described in claim 1, characterized in that, The method for obtaining the deception factor of the various data types in the current period is as follows: The ratio of the total number of elements in the cluster where the abrupt change of various types of data occurs within the current period to the maximum total number of elements in all clusters; The deception factor of each type of data in the current period is positively correlated with the ratio and the distrust index.

7. The multi-source interconnected data fusion method as described in claim 1, characterized in that, The calculation formula for optimizing the initial trust level of various types of data in the current period is as follows: In the formula, The optimized trust level for the i-th type of data in the current period; The initial level of trust in the autonomous acquisition of the i-th type of data in the current period is defined by the DS evidence theory. Let be the deception factor for the i-th type of data in the current period; This is a preset constant, and its value range is [0.005, 0.01].

8. The multi-source communication data fusion method as described in claim 1, characterized in that, The specific process of fusing all types of data in the current period is as follows: the Softmax function is used to map the optimized confidence level of all types of data in the current period to a probability value; each type of data in the current period is taken as a piece of evidence, all the evidence in the current period is taken as the input of the DS evidence theory, and the probability value mapped by the optimized confidence level of each type of data in the current period is taken as the confidence level of each piece of evidence, so as to obtain the multi-source data fusion result in the current period.

9. A multi-source interconnected data fusion system, characterized in that, The method for multi-source communication data fusion as described in any one of claims 1-8, wherein the multi-source communication data fusion system comprises: The data acquisition module is used to acquire various types of data from communication equipment in real time during combat simulation exercises. The category deviation assessment module is used to obtain the monitoring deviation index of any two categories of data in the current period based on the data difference characteristics between any two categories of data in the current period. The trust optimization module is used to obtain the deception factor of each type of data in the current period based on the degree of deviation between each type of data and other types of data in its cluster, the degree of deviation of each type of data relative to historical data, and the degree of synchronization of the mutation time of each type of data, so as to optimize the initial trust of each type of data in the current period. The multi-source data fusion module is used to fuse all types of data in the current period.

10. A multi-source interconnected data fusion storage medium, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements a multi-source communication data fusion method as described in any one of claims 1-8.

Images