Aerial target potential threat prediction method and system based on SSA-BP

Abstract

The present invention provides a method and system for predicting the potential threat of aerial targets based on SSA-BP, including collecting information on the target to obtain real-time feature data of the aerial target; predicting the future state data of the aerial target; identifying the emotional information of our navigator at the current moment; constructing a threat data set of the aerial target at both levels that integrates the navigator's emotions and the characteristics of the aerial target, and obtaining a standard target feature-emotion data set after quantification and normalization processing; optimizing the BP neural network structure through the sparrow search algorithm, fusion analysis of the target's future state data and the commander's emotional information, and obtaining the target's threat at the future moment from the situation level of both sides. The present invention constructs an efficient threat prediction method from both levels, the method is simple, the tracking speed is fast, the interactivity is strong, and it is easy to implement, and it can effectively deal with the interference of aerial targets.

Description

Technical Field

[0001] The present invention relates to the field of target detection and recognition, and in particular to a potential threat prediction scheme for aerial targets based on a Sparrow Search Algorithm (SSA) optimized BP (Back Propagation) neural network. Background Art

[0002] Prediction of target potential threat is the most basic link in achieving situation assessment, and it is the potential damage capability of aerial targets relative to our aircraft. Aerial targets such as birds and drones may pose a threat, and the threat to us should be considered comprehensively based on speed, position, height, etc. To address this problem, the main methods currently used include fuzzy analysis, evidence reasoning, support vector machine, component analysis, multi-attribute decision-making, neural network and other methods, and considerable research results have been achieved. Although there are many methods for target threat assessment, most of them are limited to the target level and rarely take our side into consideration. Compared with the situation where the power disparity between the two sides is not large, in the possible situation where the enemy is strong and we are weak, even if the state of the target remains unchanged, its threat to us will increase significantly, that is, "relative imbalance" will occur, resulting in the assessed target threat value not being able to effectively provide a reference for situation assessment.

[0003] In view of this problem, the present invention provides a potential threat prediction method based on SSA-BP neural network fusion of our navigator's emotions and air target status, using radar detectors and other equipment to collect target information within a period of time, obtain the target and the time state information of the target, input the state information within this period of time into ConvLSTM (Convolutional Long Short-Term Memory, ConvLSTM) neural network, obtain the spatiotemporal characteristics of the target and build a related prediction model to predict its future state information. Inside the cabin, the facial expression of our navigator is recorded by a camera, and five categories of anger, sadness, calmness, happiness and excitement are preset. Convolutional neural network (Convolutional Neural Networks, CNN) is used to extract features and build a classification model. The facial expression of the navigator is tracked and analyzed in real time by a camera and input into the convolutional neural network model to obtain the navigator's emotional information. Different emotional information represents different work efficiency, the highest work efficiency when excited, and the most irrational when angry. Different emotional information is assigned different threat weights according to the urgency of the state of the rational target at that time, and then the state information of the target and the emotional information of the navigator are used to construct a related data set. Based on the prior knowledge of only considering the target's characteristic state to obtain its threat level, the target's threat level is further increased or decreased according to the different emotional labels of the navigator. The fused target threat level data set at both levels is obtained. The BP neural network optimized by the sparrow search algorithm is then used to train the data set to obtain the threat level assessment model. Since human emotions are stable for a certain period of time, the navigator's current emotional information and the target's future state information are input into the assessment model to obtain the target's potential threat level value. Summary of the invention

[0004] The present invention provides a solution for predicting the potential threat degree of a target by fusing the emotions of our navigator and the state of an aerial target based on an SSA-BP neural network, so as to obtain the potential threat degree value of the target.

[0005] The technical solution of the present invention provides a method for predicting the potential threat level of aerial targets based on SSA-BP, comprising the following steps:

[0006] Step 1: collect information about the target and obtain real-time characteristic data of the aerial target;

[0007] Step 2, predicting and obtaining the future state data of the aerial target;

[0008] Step 3, identifying the current emotional information of our navigator;

[0009] Step 4: construct a threat dataset of aerial targets at both levels that integrates the navigator's emotions and the characteristics of aerial targets, and obtain a standard target feature-emotion dataset after quantification and normalization.

[0010] Step 5, the sparrow search algorithm optimizes the BP neural network structure, including the following steps:

[0011] Step 5.1, input the standard target feature-emotion data set obtained in step 4 into the BP neural network. The input of the BP neural network is a total of 7 features: target type, target speed, target heading angle, target interference capability, target altitude, target distance and our navigator's emotion. The output of the BP neural network is the threat level of the target at both levels.

[0012] Step 5.2, using the sparrow search algorithm to optimize the BP neural network to obtain the SSA-BP network;

[0013] Step 5.3, input the target state information and the navigator's emotional information into the SSA-BP network, build a potential threat degree prediction model, and output the target threat degree;

[0014] Step 6, input the future state data of the air target obtained in step 2 and the navigator's emotional information obtained in step 3 into the potential threat degree prediction model obtained in step 5.3 to obtain the potential threat degree of the air target at the level of both parties;

[0015] Furthermore, the real-time characteristic data of the aerial target include target type, target speed, target heading angle, target interference capability, target altitude and target distance.

[0016] Moreover, in step 2, the implementation method of predicting the future state data of the aerial target is as follows:

[0017] Step 2.1, the first 14 seconds of the collected time data are used as historical reference data, the 15th second data are used as test data, and the time window is 4 seconds, which are input into the ConvLSTM neural network;

[0018] Step 2.2, obtain the spatiotemporal characteristics and prediction model of the target;

[0019] Step 2.3, input the target feature data at the 12th, 13th, 14th, and 15th seconds into the prediction model obtained in step 2.2 to obtain the target future state data at the 16th second.

[0020] Moreover, step 3 is implemented as follows,

[0021] Step 3.1, data set selection, including classifying the image emotions into five categories: anger, sadness, calm, happiness, and excitement, with corresponding training sets and test sets for each category;

[0022] Step 3.2, training the model and using the trained model to identify the current emotion, including inputting the training set obtained in step 3.1 into the convolutional neural network, using the VGG-19 model for training, and treating the navigator's emotions as corresponding labels as the output of the convolutional neural network; judging the current navigator's emotion based on the encoding output by the trained convolutional neural network.

[0023] Moreover, step 4 is implemented as follows,

[0024] Step 4.1, obtaining the target feature level threat level dataset;

[0025] Step 4.2, modify the threat level of the target feature threat level data set according to the navigator's emotional state, and obtain the target feature-emotion threat level data set after fusion at both levels;

[0026] Step 4.3, quantize and normalize the features of the target feature-emotion threat dataset obtained in step 4.2 to obtain a standard target feature-emotion dataset.

[0027] The present invention provides an aerial target potential threat degree system based on SSA-BP, which is used to implement the aerial target potential threat degree method based on SSA-BP as described above.

[0028] Furthermore, the following modules are included,

[0029] The first module is used to collect information about the target and obtain real-time characteristic data of the aerial target;

[0030] The second module is used to predict the future state data of the air target;

[0031] The third module is used to identify the current emotional information of our navigator;

[0032] The fourth module is used to construct a threat dataset of aerial targets at both levels that integrates the navigator's emotions and the characteristics of aerial targets. After quantification and normalization, a standard target feature-emotion dataset is obtained.

[0033] The fifth module is used to optimize the BP neural network structure using the sparrow search algorithm, and includes the following units:

[0034] The first unit is used to input the standard target feature-emotion data set obtained in the fourth module into the BP neural network. The input of the BP neural network is a total of 7 features, namely, target type, target speed, target heading angle, target interference capability, target altitude, target distance and our navigator's emotion. The output of the BP neural network is the threat level of the target at both levels.

[0035] The second unit is used to optimize the BP neural network using the sparrow search algorithm to obtain the SSA-BP network;

[0036] The third unit is used to input the target status information and the navigator's emotional information into the SSA-BP network, build a potential threat degree prediction model, and output the target threat degree;

[0037] The sixth module is used to input the future state data of the aerial target obtained in the second module and the navigator's emotional information obtained in the third module into the potential threat degree prediction model obtained in the fifth module to obtain the potential threat degree of the aerial target at the level of both parties.

[0038] Alternatively, it includes a processor and a memory, the memory is used to store program instructions, and the processor is used to call the stored instructions in the memory to execute the above-mentioned method for determining the potential threat level of aerial targets based on SSA-BP.

[0039] Alternatively, it comprises a readable storage medium having a computer program stored thereon, and when the computer program is executed, the method for determining the potential threat level of an aerial target based on SSA-BP as described above is implemented.

[0040] On the one hand, the present invention solves the subjective emotional influence of our navigator in threat assessment, and on the other hand, solves the problems existing in the prior art. Compared with other latest research results, the SSA optimization BP method proposed in the present invention is very competitive in terms of search accuracy, convergence, algorithm speed, and stability. SSA has high performance in different search spaces and has a good ability to explore the potential area of ​​global optimality, thereby effectively avoiding the local optimal problem.

[0041] Compared with the prior art, the present invention can effectively evaluate the potential threat of aerial targets from both sides. The design process of the method is relatively simple, and the method is simple, the system is flexible, the robustness is strong, the tracking speed is fast, the interactivity is strong, and it is easy to implement. Through the above scheme, the interference of aerial targets can be effectively dealt with, and the occurrence of underestimation of the enemy caused by the influence of human subjective emotions can be prevented, which can provide an effective reference for the intelligent situation assessment system. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The present invention is a flowchart of a method for predicting potential threat by integrating the SSA-BP neural network with the emotions of our navigator and the state of an aerial target.

[0043] Figure 2 The present invention is a training flow chart of a method for predicting potential threat by integrating the SSA-BP neural network with the emotions of our navigator and the state of an aerial target.

[0044] Figure 3 It is a flow chart of the training phase of optimizing the BP neural network by the sparrow search algorithm according to an embodiment of the present invention.

[0045] Figure 4 This is the network structure of the BP neural network optimized by the sparrow search algorithm in the embodiment of the present invention. DETAILED DESCRIPTION

[0046] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0047] The present invention uses radar detectors and other equipment to collect target information within a period of time, obtain the target and the time state information of the target, input the state information within this period of time into the ConvLSTM (Convolutional Long Short-Term Memory, ConvLSTM) neural network, obtain the spatiotemporal characteristics of the target and construct a related prediction model to predict its future state information. Inside the cabin, a camera is used to record the facial expression of our navigator, and five categories of anger, sadness, calmness, happiness, and excitement are preset. The Convolutional Neural Network (CNN) is used to extract features and build a classification model. The camera is used to track and analyze the facial expression of the navigator in real time and input it into the convolutional neural network model to obtain the navigator's emotional information. Different emotional information represents different work efficiency. When excited, the work efficiency is the highest, and when angry, it is the most irrational. Different emotional information is assigned different threat weights according to the urgency of the state of the rational target at that time, and then the state information of the target and the emotional information of the navigator are used together to construct a related data set. On the basis of the threat degree of the target by only considering the characteristic state of the target in the prior knowledge, the threat degree of the target is further increased or decreased according to the different emotional labels of the navigator. The target threat dataset at both levels is obtained after fusion. The BP neural network optimized by the sparrow search algorithm is then used to train the dataset to obtain the threat assessment model. Since human emotions are stable over a certain period of time, the navigator's current emotional information and the target's future state information are input into the assessment model to obtain the target's potential threat value.

[0048] This patent implements the prediction of potential threat level based on the sparrow search algorithm optimization BP neural network fusion of our navigator's emotions and the state of the air target. The embodiment of the present invention provides a method for predicting potential threat level based on the SSA-BP neural network fusion of our navigator's emotions and the state of the air target, including the following steps:

[0049] Step 1: Use radar detectors and other equipment to collect information about the target and obtain real-time feature data of the aerial target, including target type, target speed, target heading angle, target interference capability, target height, target distance, etc. The target type and target interference capability are automatically identified by the system and are non-digital data, which do not need to be predicted by a neural network. Target types include large, small, and helicopter. Target interference capabilities include strong, medium, weak, and none. Some of the data input into the neural network for training are shown in the following table:

[0050]

[0051]

[0052] Step 2, predicting and obtaining the future state data of the aerial target;

[0053] Step 2.1: The first 14 seconds of the collected time data are used as historical reference data, and the 15th second data are used as test data, with a time window of 4 seconds, and input into the ConvLSTM neural network;

[0054] Step 2.2: Obtain the spatiotemporal characteristics and prediction model of the target;

[0055] Step 2.3: Input the target feature data at the 12th, 13th, 14th and 15th seconds into the prediction model in step 2.2 to obtain the target future state data at the 16th second. The entire working process is shown in the attached figure. Figure 2 As shown in the upper part, first obtain the training time series data, then train the ConvLSTM model, and finally use the model to predict future data.

[0056] Step 3: Identify the current emotional information of our navigator. The actual process of emotion recognition in the cabin is shown in the attached figure. Figure 2 The lower part shows the whole process of data acquisition, model building method, and emotion output;

[0057] Step 3.1: Dataset selection. In the embodiment, the facial expression dataset is derived from the internationally used Fer2013 facial expression dataset and CK+ dataset. Due to the different job identities of the pilots and the relatively convenient and clear characteristics of image collection, the specific implementation can proceed from the actual situation, pre-select the images that meet the conditions in the above datasets, and further divide the emotions into five categories: anger, sadness, calm, happiness, and excitement. There are 135 pictures in each category, 100 of which are training sets and 35 are test sets;

[0058] Step 3.2: Train the model and use the trained model to identify the current emotion. In the embodiment, the training set obtained in step 3.1 is input into the convolutional neural network, and the standard VGG-19 model is used to train it, and the navigator's emotions are regarded as corresponding labels as the output of the convolutional neural network. Encode anger, sadness, calmness, happiness, and excitement as [1 0 0 0 0], [0 1 0 0 0], [0 010 0], [0 0 0 1 0], [0 0 0 0 1], respectively, and the current navigator's emotion can be judged according to the code output by the convolutional neural network.

[0059] Step 4: Construction of fusion data set. Add the navigator's emotion feature to the threat level data set at the target feature level, and further modify the threat level according to different emotions to construct a threat level data set of aerial targets at both the navigator's emotion and the aerial target feature levels.

[0060] Step 4.1: Obtain the target feature-level threat dataset. The target feature-level threat dataset comes from Wang Gaige’s article “Target Threat Estimation Based on Intelligent Algorithms”. There are 75 sets of data in total, 60 sets of training data and 15 sets of test data. Some of the data are shown in the following table:

[0061]

[0062] Step 4.2: Further modify the threat level of the target feature threat level data set according to the navigator's emotions. Obtain the fused target feature-emotion threat level data set at both levels. Different emotions represent different working states. The navigator is the most irrational when angry and the most efficient when excited. The impact of emotional changes on the threat level is a nonlinear and complex process. In specific implementation, it can support users to modify the threat level by analyzing the target's speed, distance and other attributes, combined with the navigator's emotional state and comprehensive analysis. For the 75 data of the target feature threat level data set in step 4.1, add 5 emotions to each of them to become 5 data, a total of 375 data can be obtained, and their threat level is modified. Obtain the target feature-emotion threat level data set after fusion at both levels. Among them, 360 are used as training data and 15 are used as test data. Some of the data are shown in the following table:

[0063]

[0064] Step 4.3: Further quantify the characteristics of the target features at both sides and the emotional threat dataset after fusion. Use the nine-level quantization theory of GAMiller for quantification.

[0065] Target type: quantified into 3, 5, and 8 respectively according to large targets such as fighter-bombers, small targets such as stealth aircraft, cruise missiles, and helicopters;

[0066] Target speed: quantized into 1-9 in equal intervals (200m / s) from 0m / s to 1800m / s;

[0067] Target heading angle: quantized into 9-1 in 4° intervals from 0 to 36°;

[0068] Target interference capability: strong, medium, weak, and none are quantified as 2, 4, 6, and 8 respectively;

[0069] Target height: ultra-low, low, medium, and high are quantified as 2, 4, 6, and 8 respectively;

[0070] Target distance: quantized into 9 to 1 at equal intervals (50 km) from 0 km to 450 km.

[0071] Quantify anger, sadness, calmness, happiness, and excitement as 1, 3, 5, 7, and 9 respectively.

[0072] After quantizing the data set, the training set and test set can be mean normalized. The mean normalization mapping relationship is as follows:

[0073]

[0074] In the formula, x max =max(x),x min =min(x), x represents the value to be normalized under a certain feature, x max represents the maximum value in this set of features, x min represents the minimum value in the set of features, and y represents the normalized value of the mean corresponding to the value. After normalization, all data are normalized to the range of [0,1]. After quantizing and normalizing the data, the standard target feature-emotion data set can be obtained.

[0075] Step 5, the sparrow search algorithm optimizes the BP neural network structure;

[0076] Step 5.1: Determine the BP neural network structure. Input the standard target feature-emotion data set obtained in step 4.3 into the BP neural network. The input is a total of 7 features, including target type, target speed, target heading angle, target interference capability, target altitude, target distance, and our navigator's emotions. The output is the threat level of the target at both levels. The BP neural network topology in this patent is 7-15-1, mainly because for general pattern recognition problems, a 3-layer network (input layer-hidden layer-output layer) can solve the problem well. And the empirical formula for the relationship between the input layer nodes and the hidden layer nodes is:

[0077] n i+1 =2n i +1 (2)

[0078] where n i+1 Indicates the number of nodes in the i+1th layer, n i represents the number of nodes in the i-th layer. So the number of nodes in the hidden layer is set to 15. Initialize the weights and thresholds of the BP neural network. Figure 4 The BP neural network structure of the sparrow search optimization is shown. The input includes the target type, target speed, target heading angle, electronic countermeasure attributes, target distance, target height and the mood of our navigator, and the output is the threat level of the target.

[0079] Step 5.2: The process of optimizing BP neural network by sparrow search algorithm is as follows:

[0080] Each sparrow has only one attribute: location, which represents the location of the food it found. Each sparrow has three possible behaviors: 1. As a finder, continue to search for food, 2. As a follower, follow a finder to find food, 3. Be alert and detect, and give up the food if there is danger.

[0081] The population of n sparrows in the D-dimensional solution space is:

[0082]

[0083] Among them, d represents the dimension of the variable to be optimized, and n is the number of sparrows. Then, the fitness value of each sparrow can be expressed as follows:

[0084] fitness=argmin(mse(TrainError)) (4)

[0085] Among them, TrainError represents the prediction error of the training set, mse is the mean square error function, and argmin() represents the value of the variable when the value in the brackets is minimized. The prediction error calculation formula is as follows:

[0086] TrainError=predict 威胁值 -true 威胁值 (5)

[0087] Among them, predict 威胁值 Indicates the input of this data, that is, the predicted threat value output by the BP neural network, true 威胁值 Indicates the real threat value of the data. The fitness function shows that the network that the present invention ultimately wants to obtain is a network that can obtain better results on the training set.

[0088] Initialize the sparrow population, determine 20 sparrows, the warning value is 0.6, the proportion of discoverers is 0.7, the proportion of sparrows aware of danger is 0.2, and the maximum number of iterations is 30. The lower and upper boundaries of the weight threshold are set to -1 and 1 respectively.

[0089] Because the finder is responsible for finding food for the entire sparrow population and providing foraging directions for all joiners. Therefore, the finder can obtain a larger foraging search range than the joiner. During each iteration, the finder's position update is described as follows:

[0090]

[0091] Where t represents the current iteration number, j = 1, 2, 3, ..., d, d represents the dimension of the problem to be optimized; iter max is a constant, indicating the maximum number of iterations; Indicates the position information of the i-th sparrow in the j-th dimension at the t+1-th iteration; represents the position information of the i-th sparrow in the j-th dimension at the t-th iteration; exp() represents the exponential function with the natural constant e as the base; α∈(0,1] is a random number; R 2 (R 2 ∈[0,1]) and ST(ST∈[0.5,1]) represent the warning value and safety value respectively; Q is a random number that obeys the normal distribution; L represents a 1×d matrix, in which all elements in the matrix are 1.

[0092] When R 2 When <ST, it means that there are no predators around the foraging environment at this time, and the discoverer can perform extensive search operations. 2 ≥ST, which means that some sparrows in the population have discovered the predator and sounded the alarm to other sparrows in the population. At this time, all sparrows need to fly quickly to other safe places to forage.

[0093] As described above, during the foraging process, some followers will always monitor the discoverer. Once they realize that the discoverer has found better food, they will immediately leave their current position to compete for the food. If they win, they can immediately get the food of the discoverer. The position update of the follower is described as follows

[0094]

[0095] in, represents the optimal position occupied by the current discoverer at the t+1th iteration, represents the worst position of the current global at the tth iteration; A represents a 1×d matrix, in which each element is randomly assigned a value of 1 or -1, and AT Indicates the transposition of matrix A, (AA T ) -1 Indicates the inverse of the matrix in the brackets. This indicates that the i-th participant with lower fitness has not obtained food and is in a very hungry state. At this time, it needs to fly to other places to find food to obtain more energy. The position update description of the vigilant is as follows:

[0096]

[0097] in, represents the current global optimal position at the tth iteration; β is the step size control parameter, which is a random number that obeys a normal distribution with a mean of 0 and a variance of 1; K∈[-1,1] is a random number, f i is the fitness value of the current sparrow individual; f g and f w are the current global best and worst fitness values, respectively. ε is a constant to avoid zero in the denominator. For simplicity, when f i >f g This means that the sparrows are at the edge of their population and are extremely vulnerable to predators. best This means that the sparrow in this position is the best position in the population and is also very safe. i =f g This indicates that the sparrows in the middle of the population are aware of the danger and need to get closer to other sparrows to minimize their risk of being preyed upon. K represents the direction in which the sparrows move and is also a step length control parameter.

[0098] Complete the update of the position of the sparrow group and iterate to find the optimal fitness function value, that is, the minimum mean square error of the training set, and obtain the global optimal position of the group, that is, the optimal weight and threshold of the BP neural network. Figure 3 The flowchart of the sparrow search algorithm optimizing the BP neural network in the embodiment is shown, and the specific optimization process is as follows:

[0099] (1) Initialize the sparrow population and construct a BP neural network.

[0100] (2) Calculate the fitness value of the sparrow and determine the individual extreme value and the global optimal extreme value.

[0101] (3) Using equations (6) to (8) in step 5.2, the position of the sparrow is updated (including updating the positions of the discoverer, follower, and sentinel) to obtain the updated value of the sparrow's fitness.

[0102] (4) Repeatedly update the individual extreme values ​​and global extreme values ​​of the sparrows according to the new fitness values.

[0103] (5) After repeated iterations, when the error reaches the expected value or reaches the set maximum number of iterations, the sparrow search algorithm ends, and new neural network weights and thresholds are set based on the optimal result obtained.

[0104] Step 5.3: Assign the obtained optimal weights and thresholds directly to the BP neural network, which is the sparrow search algorithm optimized BP neural network SSA-BP, which can build a potential threat prediction model that integrates our navigator's emotions and the state of the air target. The workflow diagram of the sparrow search optimized BP neural network threat prediction is as shown in the attached figure. Figure 1 As shown, the target status information and the navigator's emotional information are input into SSA-BP, a potential threat degree prediction model is built, and the target threat degree is output.

[0105] Step 6, input the target future state data of the 16th second obtained in step 2 and the navigator's emotional information obtained in step 3 into the potential threat level prediction model obtained in step 5.3 to obtain the potential threat level of the air target at the level of both parties.

[0106] In specific implementation, the method proposed by the technical solution of the present invention can be automatically operated by those skilled in the art using computer software technology. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of the present invention and computer devices running the corresponding computer program, should also be within the scope of protection of the present invention. In some possible embodiments, a system for the potential threat level of aerial targets based on SSA-BP is provided, including the following modules:

[0107] The first module is used to collect information about the target and obtain real-time characteristic data of the aerial target;

[0108] The second module is used to predict the future state data of the air target;

[0109] The third module is used to identify the current emotional information of our navigator;

[0110] The fourth module is used to construct a threat dataset of aerial targets at both levels that integrates the navigator's emotions and the characteristics of aerial targets. After quantification and normalization, a standard target feature-emotion dataset is obtained.

[0111] The fifth module is used to optimize the BP neural network structure using the sparrow search algorithm, and includes the following units:

[0112] The first unit is used to input the standard target feature-emotion data set obtained in the fourth module into the BP neural network. The input of the BP neural network is a total of 7 features, namely, target type, target speed, target heading angle, target interference capability, target altitude, target distance and our navigator's emotion. The output of the BP neural network is the threat level of the target at both levels.

[0113] The second unit is used to optimize the BP neural network using the sparrow search algorithm to obtain the SSA-BP network;

[0114] The third unit is used to input the target status information and the navigator's emotional information into the SSA-BP network, build a potential threat degree prediction model, and output the target threat degree;

[0115] The sixth module is used to input the future state data of the aerial target obtained in the second module and the navigator's emotional information obtained in the third module into the potential threat degree prediction model obtained in the fifth module to obtain the potential threat degree of the aerial target at the level of both parties.

[0116] In some possible embodiments, a system for determining the potential threat level of an aerial target based on SSA-BP is provided, comprising a processor and a memory, wherein the memory is used to store program instructions, and the processor is used to call the stored instructions in the memory to execute a method for determining the potential threat level of an aerial target based on SSA-BP as described above.

[0117] In some possible embodiments, a system for determining the potential threat level of aerial targets based on SSA-BP is provided, comprising a readable storage medium having a computer program stored thereon, and when the computer program is executed, a method for determining the potential threat level of aerial targets based on SSA-BP as described above is implemented.

[0118] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A method for predicting the potential threat level of aerial targets based on SSA-BP, characterized by: The following steps are included: Step 1: collect information about the target and obtain real-time characteristic data of the aerial target; Step 2, predicting and obtaining the future state data of the aerial target; Step 3, identifying the current emotional information of our navigator; Step 4: construct a threat dataset of aerial targets at both levels that integrates the navigator's emotions and the characteristics of aerial targets, and obtain a standard target feature-emotion dataset after quantification and normalization. Step 5, the sparrow search algorithm optimizes the BP neural network structure, including the following steps: Step 5.1, input the standard target feature-emotion data set obtained in step 4 into the BP neural network. The input of the BP neural network is a total of 7 features: target type, target speed, target heading angle, target interference capability, target altitude, target distance and our navigator's emotion. The output of the BP neural network is the threat level of the target at both levels. Step 5.2, using the sparrow search algorithm to optimize the BP neural network to obtain the SSA-BP network; Step 5.3, input the target state information and the navigator's emotional information into the SSA-BP network, build a potential threat degree prediction model, and output the target threat degree; Step 6, input the future state data of the air target obtained in step 2 and the navigator's emotional information obtained in step 3 into the potential threat degree prediction model obtained in step 5.3 to obtain the potential threat degree of the air target at the level of both parties; The real-time characteristic data of the aerial target, including target type, target speed, target heading angle, target jamming capability, target altitude and target distance; In step 2, the method for predicting the future state data of the aerial target is as follows: Step 2.1, the first 14 seconds of the collected time data are used as historical reference data, the 15th second data are used as test data, and the time window is 4 seconds, which are input into the ConvLSTM neural network; Step 2.2, obtain the spatiotemporal characteristics and prediction model of the target; Step 2.3, input the target feature data at the 12th, 13th, 14th, and 15th seconds into the prediction model obtained in step 2.2 to obtain the target future state data at the 16th second; Step 3 is implemented as follows, Step 3.1, data set selection, including classifying the image emotions into five categories: anger, sadness, calm, happiness, and excitement, with corresponding training sets and test sets for each category; Step 3.2, training the model and using the trained model to identify the current emotion, including inputting the training set obtained in step 3.1 into the convolutional neural network, using the VGG-19 model for training, and treating the navigator's emotions as corresponding labels as the output of the convolutional neural network; judging the current navigator's emotion based on the encoding output by the trained convolutional neural network.

2. The method for predicting the potential threat level of aerial targets based on SSA-BP according to claim 1, characterized in that: Step 4 is implemented as follows, Step 4.1, obtaining the target feature level threat level dataset; Step 4.2, modify the threat level of the target feature threat level data set according to the navigator's emotional state, and obtain the target feature-emotion threat level data set after fusion at both levels; Step 4.3, quantize and normalize the features of the target feature-emotion threat dataset obtained in step 4.2 to obtain a standard target feature-emotion dataset.

3. A potential threat level system for aerial targets based on SSA-BP, characterized by: Used to implement a potential threat degree method for aerial targets based on SSA-BP as described in any one of claims 1-2.

4. The SSA-BP-based air target potential threat system according to claim 3 is characterized by: Includes the following modules, The first module is used to collect information about the target and obtain real-time characteristic data of the aerial target; The second module is used to predict the future state data of the air target; The third module is used to identify the current emotional information of our navigator; The fourth module is used to construct a threat dataset of aerial targets at both levels that integrates the navigator's emotions and the characteristics of aerial targets. After quantification and normalization, a standard target feature-emotion dataset is obtained. The fifth module is used to optimize the BP neural network structure using the sparrow search algorithm, and includes the following units: The first unit is used to input the standard target feature-emotion data set obtained in the fourth module into the BP neural network. The input of the BP neural network is a total of 7 features, namely, target type, target speed, target heading angle, target interference capability, target altitude, target distance and our navigator's emotion. The output of the BP neural network is the threat level of the target at both levels. The second unit is used to optimize the BP neural network using the sparrow search algorithm to obtain the SSA-BP network; The third unit is used to input the target status information and the navigator's emotional information into the SSA-BP network, build a potential threat degree prediction model, and output the target threat degree; The sixth module is used to input the future state data of the aerial target obtained in the second module and the navigator's emotional information obtained in the third module into the potential threat degree prediction model obtained in the fifth module to obtain the potential threat degree of the aerial target at the level of both parties.

5. The SSA-BP-based air target potential threat system according to claim 4 is characterized in that: The method comprises a processor and a memory, wherein the memory is used to store program instructions, and the processor is used to call the stored instructions in the memory to execute a method for determining the potential threat level of aerial targets based on SSA-BP as described in any one of claims 1 to 2.

6. The SSA-BP-based air target potential threat system according to claim 4, characterized in that: It comprises a readable storage medium, on which a computer program is stored. When the computer program is executed, a method for determining the potential threat degree of an aerial target based on SSA-BP as described in any one of claims 1 to 2 is implemented.

Images