CN-122020159-A - Flying object track prediction method and device and electronic equipment

CN122020159ACN 122020159 ACN122020159 ACN 122020159ACN-122020159-A

Abstract

The invention provides a method, a device and electronic equipment for predicting a flying object track, and relates to the technical field of track prediction, wherein the method comprises the steps of obtaining historical track data of a flying object, and dividing the historical track data into a training set, a verification set and a test set according to a preset proportion; based on a convolution layer and a Kolmogorov-Arnod network architecture, constructing a CKAN model and initializing the model, globally optimizing parameters of a CKAN model by using a goat optimization algorithm, determining an optimal parameter combination of a CKAN model to obtain an optimized CKAN model, training the optimized CKAN model based on a training set, inputting a testing set into the optimized CKAN model after training, and predicting a flying object track to obtain a prediction result. The accuracy of the aircraft trajectory prediction is improved by combining CKAN models and goat optimization algorithms.

Inventors

ZHANG XIAOWEI
NIE YUN
LIU YING
DONG WENTAO
Kong Zining
DONG YUCAI

Assignees

中国电子科技集团公司第十五研究所

Dates

Publication Date: 20260512
Application Date: 20260112

Claims (10)

1. A method of predicting a trajectory of a flying object, comprising: acquiring historical track data of a flying object, and dividing the historical track data into a training set, a verification set and a test set according to a preset proportion; Based on a convolution layer and a Kolmogorov-Arnod network architecture, constructing CKAN a model and initializing the model; The parameters of the CKAN model are globally optimized by using a goat optimization algorithm, and the optimal parameter combination of the CKAN model is determined to obtain an optimized CKAN model; Training the optimized CKAN model based on the training set, inputting the test set into the optimized CKAN model after training, and predicting the track of the flying object to obtain a prediction result.
2. The method of claim 1, wherein the acquiring historical trajectory data of the flying object and dividing the historical trajectory data into a training set, a validation set and a test set according to a preset ratio comprises: acquiring first track data of the flying object under different emission conditions and different environments; cleaning the first track data, removing abnormal data and error data, and obtaining second track data; Normalizing the second track data, and mapping the second track data to a [0,1] interval to obtain third track data; and dividing the third track data into a training set, a verification set and a test set according to a preset proportion.
3. The method of claim 1, wherein constructing CKAN a model and initializing the model based on a convolutional layer and a kolmogorov-arnold network architecture, comprises: Integrating a convolution layer into a Kolmogorov-Arnod network architecture to obtain a CKAN model, wherein the convolution layer is utilized to extract time sequence data characteristics, and a DILATE function is used as a loss function; parameters of the CKAN model in the initial state are determined, including the number and size of filters of the convolutional layers, the number of layers and nodes of the KAN, and the weight of the DILATE loss function.
4. The method of claim 1, wherein the globally optimizing parameters of the CKAN model using a goat optimization algorithm, determining an optimal combination of parameters for the CKAN model, resulting in an optimized CKAN model, comprises: Randomly generating a goat population containing a preset number of goats in an initial state by taking parameters of a CKAN model as individuals in a goat optimization algorithm, wherein each goat in the goat population represents a group of parameter combinations of the CKAN model; the goat optimization algorithm is utilized to iterate parameters for preset times, in each iteration, a current CKAN model is built according to a current parameter combination formed by all current parameters, the current CKAN model is trained on a training set, and a fitness value is determined on a verification set; and sequencing all goats in the goat population according to the sequence from the big to the small of the fitness value, determining the optimal goats, and combining the optimal goats as optimal parameters to obtain an optimized CKAN model.
5. The method of claim 4, wherein the determining a fitness value on the verification set comprises: taking the prediction performance of the current CKAN model on a verification set as an adaptability function of the goat optimization algorithm; an fitness value is calculated on the verification set using the fitness function.
6. The method of claim 1, wherein training the optimized CKAN model based on the training set comprises: and inputting the training set into the optimized CKAN model for training, and adjusting the weight and parameters of the optimized CKAN model to obtain the optimized CKAN model after training.
7. The method of claim 1, wherein after obtaining the prediction result, the method further comprises: Measuring the difference between the prediction result and the real track data of the testing set by adopting a preset index, and evaluating the prediction accuracy of the CKAN model after optimization; Wherein the preset index comprises any one or more of root mean square error, average absolute error and average absolute percentage error.
8. A flying object trajectory prediction device, comprising: The acquisition unit is used for acquiring historical track data of the flying object and dividing the historical track data into a training set, a verification set and a test set according to a preset proportion; the construction unit is used for constructing CKAN models based on the convolution layer and the Kolmogorov-Arnod network architecture and initializing the models; The optimization unit is used for globally optimizing parameters of the CKAN model by using a goat optimization algorithm, determining an optimal parameter combination of the CKAN model, and obtaining an optimized CKAN model; the prediction unit is used for training the optimized CKAN model based on the training set, inputting the testing set into the optimized CKAN model after training, and predicting the track of the flying object to obtain a prediction result.
9. A computer-readable storage medium storing computer instructions for causing a computer to perform the method of predicting a trajectory of a flying object according to any one of claims 1 to 7.
10. An electronic device comprising at least one processor and a memory communicatively coupled to the at least one processor, wherein the memory stores a computer program executable by the at least one processor to cause the at least one processor to perform the method of aircraft trajectory prediction of any one of claims 1 to 7.

Description

Flying object track prediction method and device and electronic equipment Technical Field The disclosure relates to the technical field of track prediction, in particular to a method and a device for predicting a track of a flying object and electronic equipment. Background Maneuvering reactions of the flyer in the middle of the track are considered as one of the most effective methods for dealing with the reverse guiding interception system, however, due to the concealment of the deployment position of the reverse flyer system, and the great flexibility of the interception means and the interception time, the flyer is difficult to accurately determine maneuvering directions, maneuvering times and maneuvering times. In this case, the reaction speed of the anti-aircraft interception system becomes a decisive factor if an effective interception is to be carried out in the active section of the aircraft flight. Therefore, advanced prediction technology is needed to carry out prospective analysis on the flight track of the flying object, and the flight parameters of the flying object in the active section are accurately obtained, so that sufficient reaction time is reserved for the anti-flying object interception system, and the interception success rate is improved. The track prediction method adopted in the related technology mostly depends on a kinematic or dynamic model, and solves the target track information through mathematical modeling. However, in practical application, the flying process of the target flying object is extremely susceptible to complex environmental factors and artificial interference, so that accurate modeling is difficult, and the prediction accuracy of the flying object track is low. Aiming at the problem of lower prediction precision of the track of the flying object in the related technology, no effective technical solution is proposed at present. Disclosure of Invention The main objective of the present disclosure is to provide a method, an apparatus and an electronic device for predicting a flying object trajectory, so as to solve the problem of low accuracy of predicting the flying object trajectory in the related art. To achieve the above object, a first aspect of the present disclosure provides a method for predicting a trajectory of an aircraft, including: acquiring historical track data of a flying object, and dividing the historical track data into a training set, a verification set and a test set according to a preset proportion; Based on a convolution layer and a Kolmogorov-Arnod network architecture, constructing CKAN a model and initializing the model; The parameters of the CKAN model are globally optimized by using a goat optimization algorithm, and the optimal parameter combination of the CKAN model is determined to obtain an optimized CKAN model; Training the optimized CKAN model based on the training set, inputting the test set into the optimized CKAN model after training, and predicting the track of the flying object to obtain a prediction result. Optionally, the acquiring historical track data of the flying object and dividing the historical track data into a training set, a verification set and a test set according to a preset proportion includes: acquiring first track data of the flying object under different emission conditions and different environments; cleaning the first track data, removing abnormal data and error data, and obtaining second track data; Normalizing the second track data, and mapping the second track data to a [0,1] interval to obtain third track data; and dividing the third track data into a training set, a verification set and a test set according to a preset proportion. Optionally, the constructing CKAN a model and initializing the model based on the convolutional layer and the kolmogorov-arnod network architecture includes: Integrating a convolution layer into a Kolmogorov-Arnod network architecture to obtain a CKAN model, wherein the convolution layer is utilized to extract time sequence data characteristics, and a DILATE function is used as a loss function; parameters of the CKAN model in the initial state are determined, including the number and size of filters of the convolutional layers, the number of layers and nodes of the KAN, and the weight of the DILATE loss function. Optionally, the global optimization CKAN of the parameters of the model by using a goat optimization algorithm determines an optimal parameter combination of the CKAN model to obtain an optimized CKAN model, which includes: Randomly generating a goat population containing a preset number of goats in an initial state by taking parameters of a CKAN model as individuals in a goat optimization algorithm, wherein each goat in the goat population represents a group of parameter combinations of the CKAN model; the goat optimization algorithm is utilized to iterate parameters for preset times, in each iteration, a current CKAN model is built according to a current paramete