1. A method for improving a current statistical model based on a neural network is characterized by comprising the following steps:

s1, building an improved current statistical model based on a neural network, connecting the neural network serving as a feedback network with the traditional statistical model, and acquiring the maximum acceleration value according with the target maneuvering condition;

s2, selecting a training set and a label set, and preprocessing the training set and the label set according to a set normalization standard rule;

s3, setting training parameters, and training the neural network by using the training parameters to finally achieve convergence;

and S4, inputting the estimation state and the measurement of the target to be processed into the improved model to obtain the estimation value of the state of the target at the current moment.

2. The neural network-based method for improving the current statistical model according to claim 1, wherein: the estimating of the target current state value in step S1 specifically includes the following steps:

s101, at each tracking moment, carrying out normalization processing on the target state estimation value at the previous moment and the measurement value at the current moment according to a normalization standard;

and S102, the processed numerical value enters a feedback network to obtain the maximum acceleration value which accords with the target maneuvering condition at the current moment.

3. The method of claim 2, wherein the neural network based method comprises: the number of nodes of the input layer of the neural network corresponds to the sum of the state, the measurement dimension, the number of the hidden layer is 1, the number of the hidden layer is 7, the number of the nodes of the output layer is 1, and the network relation is as follows:

in step S103, the current statistical model obtains a state estimation value at the current time through a measurement and filtering algorithm.

4. The neural network-based method for improving the current statistical model according to claim 1, wherein: in step S2, the training set includes a target data set with multiple maneuver levels, and the target data set is segmented to obtain a training set and a label set, where the label set is a target real state at each time.

5. The method of claim 4, wherein the current statistical model is modified based on a neural network, and the method comprises the following steps: the normalization standardization rule is a Min-Max standardization mode.

6. The neural network-based method for improving the current statistical model according to claim 1, wherein: training a back propagation neural network in an end-to-end manner by using MSE as a loss function in the step S3;

wherein the loss function is:

where m is the number of training data.

7. The neural network-based method for improving the current statistical model according to claim 1, wherein: in the step S3, an Adam optimization method is adopted in the neural network training process.

8. An improved current statistical model system based on neural networks, characterized in that: the system comprises:

the model construction module is used for constructing an improved current statistical model based on a neural network, and connecting the neural network serving as a feedback network with a traditional statistical model so as to estimate a target current state value;

the data set preprocessing module is used for selecting a training set and a label set and preprocessing the training set and the label set according to a set normalization standard rule;

the network training module is used for setting training parameters and training the neural network by using the training parameters to finally achieve convergence; and;

and the estimated value calculation module is used for inputting the estimated state and the measurement of the target to be processed into the improved model to obtain the estimated value of the state of the target at the current moment.

9. An improved current statistical model storage medium based on neural networks, characterized by: the storage medium having stored thereon a computer program which, when being executed by a processor, is capable of carrying out the step of improving a current statistical model using a neural network as claimed in any one of claims 1 to 7.

10. An improved current statistical model device based on neural network, characterized in that: the apparatus comprises a memory, a processor, and an algorithmic program stored on the memory and executable on the processor to improve a current statistical model using a neural network, the algorithmic program to improve a current statistical model using a neural network being configured to implement the steps of improving a current statistical model using a neural network of any of claims 1 to 7.

Background

In recent years, target tracking is a core key technology of radar data processing, and real-time state estimation of a target can be performed through measurement information acquired by a radar through track initiation, track prediction and a filtering algorithm to obtain a motion track and motion parameters of the target, so that the target is tracked and radar terminal display is completed. The track prediction and filtering algorithm belongs to a target tracking loop and is important for high-precision estimation of a target state. The existing track prediction method is located after the track is started, and the filtering algorithm is located after the track is predicted, and both parts relate to target motion modeling. The commonly used basic object motion models include a constant speed model, a constant acceleration model, a cooperative turning model, a Singer model, a current statistical model, a Jerk model and the like.

The disclosure number is CN111157983A, which provides a radar target tracking method, wherein parameters of a model are adjusted according to changes of target maneuvering characteristics, so as to achieve real-time adaptive target tracking update, achieve high-precision target tracking, and improve radar identification precision, but the target tracking model in the method mainly comprises a uniform velocity model, a uniform acceleration model, a current statistical model, a turning model, a continuous turning model, and several models are all represented in a bayesian filtering manner. Meanwhile, the current statistical model method also has the problem that parameters cannot be adjusted in a self-adaptive manner, and when a target moves suddenly, the tracking error is greatly increased. The tracking method needs manual adjustment, and the adjusted tracking effect is difficult to achieve real-time optimization, so that the existing current statistical model method has the problems of simple model, low complexity, poor universality, lack of learning capability and the like, and is difficult to solve the problem of high-precision tracking on the whole.

Disclosure of Invention

The present invention is directed to a method, system, storage medium and computer device for improving a current statistical model based on a neural network, so as to solve the problems in the background art.

In order to achieve the purpose, the invention provides the following technical scheme:

a method for improving a current statistical model based on a neural network comprises the following steps:

s1, building an improved current statistical model based on a neural network, connecting the neural network serving as a feedback network with the traditional statistical model, and acquiring the maximum acceleration value according with the target maneuvering condition;

s2, selecting a training set and a label set, and preprocessing the training set and the label set according to a set normalization standard rule;

s3, setting training parameters, and training the neural network by using the training parameters to finally achieve convergence;

and S4, inputting the estimation state and the measurement of the target to be processed into the improved model to obtain the estimation value of the state of the target at the current moment.

Preferably, the estimating of the target current state value in step S1 specifically includes the following steps:

s101, at each tracking moment, carrying out normalization processing on the target state estimation value at the previous moment and the measurement value at the current moment according to a normalization standard;

and S102, the processed numerical value enters a feedback network to obtain the maximum acceleration value which accords with the target maneuvering condition at the current moment.

Preferably, the number of nodes of the input layer of the neural network corresponds to the sum of the state and the measurement dimension, the number of the hidden layer is 1, the number of the hidden layer is 7, the number of the nodes of the output layer is 1, and the network relation is as follows:

in step S103, the current statistical model obtains a state estimation value at the current time through a measurement and filtering algorithm.

Preferably, the training set in step S2 includes a target data set with multiple maneuver levels, and the target data set is segmented to obtain the training set and a label set, where the label set is a target real state at each time.

Preferably, the normalization standardization rule is a Min-Max standardization mode.

Preferably, the step S3 is to train the back propagation neural network in an end-to-end manner by using MSE as a loss function;

wherein the loss function is:

where m is the number of training data.

Preferably, an Adam optimization method is adopted in the neural network training process in step S3.

To achieve the above object, the present invention further provides an improved current statistical model system based on a neural network, wherein the system comprises:

the model construction module is used for constructing an improved current statistical model based on a neural network, and connecting the neural network serving as a feedback network with a traditional statistical model so as to estimate a target current state value;

the data set preprocessing module is used for selecting a training set and a label set and preprocessing the training set and the label set according to a set normalization standard rule;

the network training module is used for setting training parameters and training the neural network by using the training parameters to finally achieve convergence; and;

and the estimated value calculation module is used for inputting the estimated state and the measurement of the target to be processed into the improved model to obtain the estimated value of the state of the target at the current moment.

To achieve the above object, the present invention further provides a storage medium for improving a current statistical model based on a neural network, wherein the storage medium stores a computer program, and the computer program, when executed by a processor, can implement the above steps of improving the current statistical model using the neural network.

In order to achieve the above object, the present invention further provides an apparatus for improving a current statistical model based on a neural network, wherein the apparatus comprises a memory, a processor and an algorithm program stored in the memory and executable on the processor for improving the current statistical model using the neural network, and the algorithm program for improving the current statistical model using the neural network is configured to implement the above step of improving the current statistical model using the neural network.

Compared with the prior art, the invention has the beneficial effects that:

the method for improving the current statistical model is based on learning of a back propagation neural network, enhances the self-adaptive adjustment capability of the traditional current statistical model by means of the nonlinear expression capability of the neural network, wherein the back propagation neural network is used as a feedback network to output the self-adaptive adjustment maximum acceleration according to a target state estimation value and a measurement value, finally outputs the maximum acceleration to the traditional current statistical model for current state estimation, establishes mapping among a target state, the measurement and the maximum acceleration, and greatly reduces the tracking error when the target is maneuvered.

Drawings

FIG. 1 is a flow chart of a method for improving a current statistical model by a neural network according to the present invention;

FIG. 2 is a schematic structural diagram of a neural network improving a current statistical model system according to the present invention;

FIG. 3 is a schematic block diagram of a back propagation neural network of the present invention;

FIG. 4 is a block diagram of the improved current statistical model network structure based on neural network of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example (b):

referring to fig. 1 to 4, the present invention provides a technical solution:

a method for improving a current statistical model based on a neural network comprises the following steps:

and S1, constructing an improved current statistical model based on the neural network, connecting the neural network serving as a feedback network with the traditional statistical model, and acquiring the maximum acceleration value according with the target maneuvering condition.

The estimation of the target current state value specifically comprises the following steps:

s101, at each tracking moment, carrying out normalization processing on the target state estimation value at the previous moment and the measurement value at the current moment according to a normalization standard;

and S102, the processed numerical value enters a feedback network to obtain the maximum acceleration value which accords with the target maneuvering condition at the current moment.

The number of nodes of the input layer of the neural network corresponds to the sum of the state, the measurement dimension, the number of the hidden layer layers is 1, the number of the hidden layer nodes is 7, the number of the nodes of the output layer is 1, and the network relation is as follows:

and S2, selecting the training set and the label set, and preprocessing the training set and the label set according to a set normalization standard rule.

The training set comprises a plurality of target data sets with maneuvering grades, the target data sets are segmented to obtain the training set and a label set, the label set is the target real state at each moment, and the normalization standardization rule is a Min-Max standardization mode.

And S3, setting training parameters, and training the neural network by using the training parameters to finally achieve convergence.

The training parameters comprise learning rate and optimizer parameters, the parameters can be adjusted according to training and testing effects, an optimal parameter value is finally selected, MSE is used as a loss function, and a back propagation neural network is trained in an end-to-end mode;

wherein the loss function is:

and in addition, an Adam optimization method is adopted in the neural network training process, and the adaptive adjustment capability of the traditional current statistical model is enhanced by training a back propagation neural network.

And S4, inputting the estimation state and the measurement of the target to be processed into the improved model to obtain the estimation value of the state of the target at the current moment.

Wherein a to be output_maxInputting the maximum acceleration into a traditional current statistical model, estimating a current state value of the target by the current statistical model according to the maximum acceleration obtained in the step S102, estimating the current state by the current statistical model by outputting the maximum acceleration into the traditional current statistical model on the basis of self-adaptive maximum acceleration, obtaining a state estimation value at the current moment through a prediction and filtering algorithm, establishing mapping among the state, the measurement and the maximum acceleration of the target, and greatly reducing the tracking error when the target maneuvers.

To achieve the above object, the present invention further provides an improved current statistical model system based on a neural network, wherein the system comprises:

the model construction module is used for constructing an improved current statistical model based on a neural network, and connecting the neural network serving as a feedback network with a traditional statistical model so as to estimate a target current state value;

the data set preprocessing module is used for selecting a training set and a label set and preprocessing the training set and the label set according to a set normalization standard rule;

the network training module is used for setting training parameters and training the neural network by using the training parameters to finally achieve convergence; and;

and the estimated value calculation module is used for inputting the estimated state and the measurement of the target to be processed into the improved model to obtain the estimated value of the state of the target at the current moment.

To achieve the above object, the present invention further provides a storage medium for improving a current statistical model based on a neural network, wherein the storage medium stores a computer program, and the computer program, when executed by a processor, can implement the above steps of improving the current statistical model using the neural network.

The storage medium may include: any entity or device capable of carrying computer program code, recording medium, U.S. disk, removable hard disk, magnetic disk, optical disk, computer memory, Read Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution media, and the like. It should be noted that the computer readable medium may contain other components which may be suitably increased or decreased as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, in accordance with legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunications signals.

In order to achieve the above object, the present invention further provides an apparatus for improving a current statistical model based on a neural network, wherein the apparatus comprises a memory, a processor and an algorithm program stored in the memory and executable on the processor for improving the current statistical model using the neural network, and the algorithm program for improving the current statistical model using the neural network is configured to implement the above step of improving the current statistical model using the neural network.

The device may be a desktop computer, an industrial computer, a numerical control device, an industrial robot, a server, or other computing device. Those skilled in the art will appreciate that the device includes a processor and a memory, the description of which is for storing instructions is merely an example of a device and does not constitute a limitation of the device, and may include more or less components, or combine certain components, or different components, e.g., the device may also include input output devices, network access devices, buses, etc.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.