A shooting gate control method based on live wire motion trajectory prediction

Abstract

The application relates to a shooting gate control method based on a fire line motion trajectory prediction, which comprises the following steps: acquiring fire line motion trajectory data in an armored vehicle dynamic shooting process through a stable gyroscope of a fire control system, constructing a data set for long short-term memory neural network training, then transplanting the trained model into the fire control system to realize real-time prediction of the fire line motion trajectory, and designing a prediction gate, using the predicted angle value of the fire line as the condition for judging whether the fire line enters the gate, so that the probability that the fire line is still in the shooting gate when the projectile is shot out of the muzzle under dynamic shooting is improved.

Description

Technical Field

[0001] This invention belongs to the field of weapon control technology, specifically relating to a firing gate control method based on fire trajectory prediction. Background Technology

[0002] The firing determination mechanism of the fire control system of armored vehicles is the firing coincidence gate technology, which issues a firing signal as soon as the fire line enters the firing gate. During the dynamic firing process of armored vehicles, the fire line is indirectly affected by the vibration of the armored vehicle itself and the excitation of the road surface, and cannot remain completely stable at the given firing position. In addition, there is a firing delay between the fire control system issuing a firing signal and the projectile leaving the muzzle. Therefore, the existing firing gate can only guarantee that the fire line is within the firing gate when the firing signal is issued, but cannot guarantee that the fire line is still within the firing gate when the projectile leaves the muzzle, which restricts the improvement of the firing accuracy of the fire control system. Summary of the Invention

[0003] This invention provides a shooting gate control method based on fire trajectory prediction. The technical problem to be solved is: after the firing delay time, it is impossible to determine whether the fire line is inside the shooting gate.

[0004] To address the above technical problems, this invention provides a shooting gate control method based on fire trajectory prediction, characterized by the following steps:

[0005] S1. Train a long short-term memory neural network and implant it into the fire control system to predict the trajectory of the fire line;

[0006] S2. Design a prediction gate. Predict the trajectory of the fire line in the next time period through S1. Use the difference between the predicted angle and the angle of fire calculated by the fire control system to determine whether to give a firing signal.

[0007] S1 specifically refers to:

[0008] First, the movement trajectory data of the fire line during the dynamic firing process of the armored vehicle is obtained through the stabilized gyroscope of the fire control system;

[0009] Secondly, the acquired motion trajectory data is used to construct a dataset for training the long short-term memory neural network; the fire line motion data before the shooting delay is used as input, and the fire line motion data after the shooting delay is used as output.

[0010] Finally, the trained long short-term memory neural network model is implanted into the weapon control system to predict the trajectory of the fire line in real time.

[0011] Beneficial effects: Without increasing the redundancy of the system structure, this invention predicts the trajectory of the fire line and the moment of entry into the firing gate, and then gives the firing signal in advance, so that when the projectile leaves the muzzle, the fire line is just inside the firing gate. This effectively solves the problem caused by firing delay and extends the controlled time of the fire line, which is conducive to improving the firing accuracy of the fire control system. Attached Figure Description

[0012] Figure 1 LSTM neural network architecture diagram

[0013] Figure 2 Logic diagram of the collaborative operation of the prediction gate and the shooting overlap gate. Detailed Implementation

[0014] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below.

[0015] The present invention proposes a shooting gate control method based on fire trajectory prediction, comprising the following steps:

[0016] S1. Train a Long Short-Term Memory (LSTM) neural network and embed it into the fire control system to predict the trajectory of the fire line, as detailed below:

[0017] First, the movement trajectory data of the fire line during the dynamic firing process of the armored vehicle is obtained through the stabilized gyroscope of the fire control system;

[0018] Secondly, the acquired motion trajectory data is used to construct a dataset for training the long short-term memory neural network; the fire line motion data before the shooting delay is used as input, and the fire line motion data after the shooting delay is used as output.

[0019] Finally, the trained Long Short-Term Memory (LSTM) neural network model is implanted into the weapon control system to predict the trajectory of the fire line in real time. (LSTM neural network, for example...) Figure 1 As shown, its specific model is as follows:

[0020] Wherein: g t Let h(t) represent the temporary memory unit, h(t) represent the output state unit, i(t) represent the input gate, f(t) represent the forget gate, o(t) represent the output gate, and c(t) represent the memory unit; let x(t) be the input time series at the current moment, then the state of each control gate is represented by the following formula:

[0021] g t =tanh(W xc x(t)+W hc h t-1 +bg)

[0022]

[0023] i(t)=σ(W xi x(t)+W hi h(t-1)+b i )

[0024] f(t)=σ(W xf x(t)+W hf h(t-1)+b f )

[0025] o(t)=σ(W xo x(t)+W ho h(t-1)+b0)

[0026] y(t)=h(t)=o(t)·tanh(c(t))

[0027] Where σ represents the sigmoid activation function; W xc W is the weight matrix for the unit state input x(t). hc Input h to the unit state t-1 The weight matrix, W xi The weight matrix W is the input gate input x(t). hi Input h to the input gate t-1 The weight matrix, W xf W is the weight matrix of the forget gate input x(t). hf Input h for the forget gate t-1 The weight matrix, W xo The weight matrix W is the input gate input x(t). ho Input h to the input gate t-1 The weight matrix; b g For the bias of the cell state, b i For the bias of the input gate, b f b0 is the bias for the forget gate, and b0 is the bias for the output gate.

[0028] S2. Design a prediction gate, using the predicted angle value of the fire line as the condition for determining whether the fire line has entered the gate, thereby increasing the probability that the fire line is still within the firing gate when the projectile leaves the muzzle under dynamic firing.

[0029] The trajectory of the fire line in the next time period is predicted by S1, and the difference between the predicted angle and the firing angle calculated by the fire control system is used to determine whether to issue a firing signal; the details are as follows:

[0030] Let Δt be the firing delay between the firing signal from the fire control system and the projectile exiting the muzzle. At time T+Δt, the predicted elevation and horizontal angles of the fire line are α and β, respectively, predicted by S1. The fire control system calculates the lead angles α′ and β′. If the absolute value of the difference between the predicted angles α and β at time T+Δt and the lead angles α′ and β′ calculated by the fire control system is less than a threshold given by the fire control system, then the fire line is considered to have entered the firing gate at time T+Δt, and a firing signal is given at time T, regardless of whether the fire line is inside the firing gate at that time. After time Δt, the fire line is exactly inside the gate, and the projectile exits the muzzle at this time, completing one firing action.

[0031] The basic principle of prediction gates is shown in the following equation:

[0032] |α-α′|≤δ v

[0033] |β-β′|≤δ h

[0034] Where δ v To predict the high and low thresholds of the gate, δ h To predict the horizontal threshold of the gate.

[0035] Predictive gates are designed to improve the firing accuracy of fire control systems and play an auxiliary role in the firing process. Figure 2 This is a logic diagram showing the coordinated operation of the prediction gate and the firing coincidence gate. When the operator selects "Enable Prediction Assistance Function", the signal is "1", and the prediction gate is activated to make a firing decision; when the signal is "0", the system makes a decision according to the original predetermined rules.

[0036] This invention uses an LSTM neural network to predict the trajectory of the fire line in real time; a prediction gate is designed using the predicted value of the fire line trajectory, which works in conjunction with the firing coincidence gate; it can effectively solve the problem that when an armored vehicle fires dynamically, the fire line is at the gate when the firing signal is given, but the fire line may not be at the firing gate when the projectile leaves the muzzle.

[0037] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A shooting gate control method based on fire trajectory prediction, characterized in that, Includes the following steps: S1. Train a long short-term memory neural network and implant it into the fire control system to predict the trajectory of the fire line; specifically: First, the movement trajectory data of the fire line during the dynamic firing process of the armored vehicle is obtained through the stabilized gyroscope of the fire control system; Secondly, the acquired motion trajectory data is used to construct a dataset for training the long short-term memory neural network; the fire line motion data before the shooting delay is used as input, and the fire line motion data after the shooting delay is used as output. Finally, the trained long short-term memory neural network model is implanted into the weapon control system to predict the trajectory of the fire line in real time. S2. Design a prediction gate. Predict the trajectory of the fire line in the next time period through S1. Use the difference between the predicted angle and the angle of fire calculated by the fire control system to determine whether to give a firing signal. set up Provide the fire control system with the firing delay between the firing signal and the projectile leaving the muzzle. The predicted angles of the fire line in terms of elevation and horizontal direction are obtained through S1 prediction at any given time. , The fire control system calculates the lead time for the firing angle as follows: , ,when Time Angle Prediction Value , The lead time of the firing angle calculated by the fire control system , If the absolute value of the difference is less than the threshold given by the fire control system, then the fire line is considered to be... Always enter the room.

2. The shooting gate control method based on fire trajectory prediction according to claim 1, characterized in that, The Long Short-Term Memory neural network model is as follows: set up Given the input time series at the current moment, the state of each control threshold is represented by the following formula: in, Represents a temporary memory unit. This indicates a status output unit. Indicates the input gate. Represents the Gate of Oblivion Indicates the output gate. Represents a memory unit; represent Activation function; Input for cell status The weight matrix, Input for cell status The weight matrix, Input to the input gate The weight matrix, Input to the input gate The weight matrix, Input for the forget gate The weight matrix, Input for the forget gate The weight matrix, Input to the input gate The weight matrix, Input to the input gate The weight matrix; This is the bias for the cell state. This is the bias of the input gate. For the offset of the forget gate, This is the bias of the output gate.

3. The shooting gate control method based on fire trajectory prediction according to claim 1, characterized in that: in To predict the high and low thresholds of the gate, To predict the horizontal threshold of the gate.

4. The shooting gate control method based on fire trajectory prediction according to claim 1, characterized in that, When the predictive assistance function is enabled, the signal is 1, and the predictive gate is activated to determine the firing action; when the signal is 0, the fire control system makes the determination according to the original established rules.

Images