Photoelectric turret visual axis stability control method based on adaptive disturbance prediction compensation and computer product

Abstract

The invention discloses a photoelectric turret visual axis stability control method based on self-adaptive disturbance prediction compensation and a computer product, and the method comprises the following steps: S1, establishing a gyro speed loop discrete model, calculating the deviation between the output of the model and the angular velocity of a visual axis measured by a gyro, and obtaining a disturbance estimation value through an inverse model and a low-pass filter; s2, constructing a mathematical model of disturbance based on a recursive least square method, and performing multi-step forward prediction on equivalent disturbance by using the model; and S3, carrying out weighted fusion on the predicted equivalent disturbance and the real-time disturbance estimation value, and finally feeding a fused compensation signal forward to the input end of the control system to counteract the influence of the actual disturbance on the system. According to the method, real-time estimation and dynamic prediction are fused, the learning-prediction capacity is added on the basis of perception-compensation of a traditional DOB, the method is suitable for a photoelectric turret control system which is remarkable in periodic disturbance or sensitive to delay, phase lag of the traditional DOB is reduced, and the stable control bandwidth and disturbance rejection capacity of a visual axis are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of photoelectric turret line-of-sight control technology, specifically to a photoelectric turret line-of-sight stabilization control method and computer product based on adaptive disturbance prediction compensation. Background Technology

[0002] Line-of-sight stabilization control in optoelectronic systems is a key technology for ensuring imaging quality and target tracking accuracy of optical payloads in dynamic environments. In applications on various stabilized platforms (such as vehicle-mounted and shipborne systems), external disturbances such as carrier vibration and airflow disturbances can cause the optical axis to deviate from the target, preventing the optoelectronic payload from maintaining effective "staring" within the integration time, thus reducing the performance of target reconnaissance, identification, and tracking. Therefore, suppressing the effects of interference through high-bandwidth, high-precision line-of-sight stabilization control is a necessary condition for the high-performance operation of optoelectronic systems.

[0003] Against this backdrop, traditional Disturbance Observer (DOB) compensation has become an effective means of improving system robustness because it can estimate and compensate for equivalent disturbances online by constructing a disturbance observer. DOB transforms the difference between the output of the actual object and the nominal model into a disturbance estimate, achieving low-frequency disturbance suppression without additional sensors, thus significantly improving the system's robustness.

[0004] However, the shortcomings of traditional DOB (Discretionary Observer) cannot be ignored: First, the observer performance is limited by the accuracy of the nominal model, and model mismatch can lead to estimation errors. Second, to ensure stability, a low-pass filter needs to be added to the observer, but the selection of the cutoff frequency can cause a contradiction between insufficient high-frequency disturbance suppression and phase loss. Increasing the bandwidth of the disturbance observer filter broadens the system's observation range of disturbances, but may introduce stability and noise problems, leading to system oscillations. These shortcomings restrict the further application of DOB in complex optoelectronic systems and urgently require optimization through algorithm improvement or integration with other control strategies. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a photoelectric turret line-of-sight stabilization control method and computer product based on adaptive disturbance prediction compensation. This method adaptively compensates for fluctuations in the line-of-sight angular velocity during maneuvers, thereby improving the anti-interference capability and stability accuracy of the photoelectric turret's line-of-sight system. It is not limited by the frame composition of the photoelectric system or the accuracy of the gyroscope sensing element.

[0006] Specifically, the present invention provides a photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation, comprising:

[0007] S1: Establish a second-order discrete model of the gyroscope velocity loop based on the recursive least squares (RLS) method:

[0008] ,

[0009] Wherein, parameter vector Measurement vector .

[0010] The deviation between the output of the second-order discrete model of the gyroscope velocity loop and the line-of-sight angular velocity measured by the gyroscope is calculated, and the perturbation estimate is obtained by passing the inverse model and a low-pass filter. :

[0011] ,

[0012] in, The output of the identified nominal model is... The output after inverting the nominal model. The angular velocity of the gyroscope's line of sight. It is a low-pass filter;

[0013] S2: Construct a mathematical model of the perturbation based on the recursive least squares method, and predict the equivalent perturbation in the next N steps. :

[0014] ,

[0015] Wherein, parameter vector Measurement vector L is the dimension of the measurement vector;

[0016] S3: Utilizing the angular velocity sensed by the gyroscope, a feedback closed loop for line-of-sight stabilization control is constructed, which converts the predicted and estimated equivalent disturbances. With real-time disturbance Weighted fusion is then performed, and the resulting disturbance prediction compensation signal is then... Feedforward to the control input to offset the impact of actual disturbances on the system and achieve adaptive disturbance prediction and compensation.

[0017] Furthermore, the discrete model identification of the gyro velocity loop in S1 includes:

[0018] S11: Estimate model parameters using a recursive least squares parameter adaptive law, index function:

[0019] ,

[0020] S12: Apply the same control input signal to the input terminals of the actual system and the identification model to construct the model evaluation function; verify the degree of fit between the constructed model and the actual system by real-time statistical mean square error.

[0021] Mean square error calculation for:

[0022] ε represents the prior prediction error;

[0023] Gradually increase the model order and observe the mean square error. The changing trend, select the mean squared error The optimal model order is the order at which the price decreases significantly but has not yet leveled off, thus avoiding underfitting or overfitting and achieving the goal of accurately modeling and predicting time series data.

[0024] Furthermore, the model inverse obtained based on recursive least squares identification may be unstable, leading to exponential growth of disturbance observations, severe oscillations in the control signal, and divergence in the control system. Therefore, it is necessary to construct an approximate inverse of the nominal model based on Zero Phase Error Tracking Control (ZPETC). When taking the approximate inverse of the model, the poles and zeros are considered first, and the original poles are replaced with mirror poles to ensure stability. Secondly, a delay term is introduced to compensate for the relative order difference and avoid non-causal inverses. Finally, gain adjustment is used to sacrifice high-frequency accuracy in exchange for tracking performance in low-frequency bands (such as near the fundamental frequency). Therefore, the discrete model identification of the gyroscope velocity loop in S1 also includes:

[0025] S13: Using a time delay operator By shifting the analysis of the control system from the discrete-time domain to the frequency domain, the transfer function becomes:

[0026] ,

[0027] in: Equivalent to ;

[0028] Zero points, including stable zero points and unstable zeros , As the extreme point, For gain, Pure delay;

[0029] Decompose the zero point into stable zero points. and unstable zeros :

[0030] ,

[0031] ;

[0032] Constructing an approximate inverse of the nominal model based on zero-phase error tracking control:

[0033] ,

[0034] ;

[0035] in, This is the gain compensation coefficient.

[0036] Furthermore, the perturbation model is a fourth-order model, and the perturbation model is as follows:

[0037] ,

[0038] Wherein, parameter vector Measurement vector .

[0039] Furthermore, in S3, the estimated equivalent disturbance will be predicted. With real-time disturbance Weighted fusion yields the disturbance prediction compensation signal. for:

[0040] ,

[0041] in, Weights for predicting confidence levels.

[0042] The present invention also provides a computer product, including computer program instructions that, when executed by a processor, implement the steps of the above-described method.

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0044] (1) The present invention uses the recursive least squares method to establish a discrete model of the gyroscope velocity loop online, which can dynamically update the nominal model parameters and its inverse model expression, effectively overcoming the mismatch problem of the traditional fixed parameter model when the system operating point changes, and providing a reliable model basis for the prediction and compensation of the subsequent disturbance observer.

[0045] (2) It has achieved the characterization of periodic and non-stationary disturbances, providing an accurate model basis for disturbance suppression.

[0046] (3) This invention achieves adaptive prediction and compensation by dynamically adjusting the prediction time domain and compensation weight. It adds the ability of "learning-prediction" on the basis of the traditional DOB's "perception-compensation", which significantly improves the system's anti-interference ability and stability accuracy. Attached Figure Description

[0047] Figure 1 This is a block diagram of the photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation according to the present invention;

[0048] Figure 2The diagram shows the model identification method based on recursive least squares.

[0049] Figure 3 A comparison chart of the line-of-sight stabilization accuracy of the traditional method and the method of the present invention under a 1° 1Hz disturbance on a moving carrier;

[0050] Figure 4 A comparison chart of the line-of-sight stabilization accuracy of the traditional method and the method of the present invention under a 1° 2Hz disturbance on a moving carrier; Detailed Implementation

[0051] The present invention will be further described in detail below through specific embodiments.

[0052] Example 1

[0053] like Figure 1 As shown, this embodiment discloses a photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation, including:

[0054] S1: Establish a second-order discrete model of the gyroscope velocity loop based on the recursive least squares (RLS) method:

[0055] ,

[0056] Wherein, parameter vector Measurement vector .

[0057] The deviation between the output of the second-order discrete model of the gyroscope velocity loop and the line-of-sight angular velocity measured by the gyroscope is calculated, and the perturbation estimate is obtained by passing the inverse model and a low-pass filter. :

[0058] ,

[0059] in, The output of the identified nominal model is... The output after inverting the nominal model. The angular velocity of the gyroscope's line of sight. It is a low-pass filter.

[0060] The specific control block diagram for model identification based on recursive least squares method in step S1 is as follows: Figure 2 As shown, the steps are as follows:

[0061] Based on theoretical prior knowledge, the model order is initially estimated.

[0062] Pseudo-random binary sequences (PRBS) are superimposed on the control inputs of the motor and model, and the outputs of the motor and model are acquired. The PRBS is generated by a shift register via feedback, and the maximum sequence length is 2^32. n -1, where n is the number of bits in the shift register.

[0063] S11: Estimate model parameters using a recursive least squares parameter adaptive law, index function:

[0064] ,

[0065] in, Solving the equation yields the recursive least squares (RLS) parameter adaptation law:

[0066] ,

[0067] ,

[0068] ;

[0069] in, Here is the gain matrix. This represents the prior prediction error.

[0070] S12: Apply the same control input signal to the input terminals of the actual system and the identification model to construct the model evaluation function; verify the degree of fit between the constructed model and the actual system by real-time statistical mean square error; mean square error calculation. for:

[0071] ε represents the prior prediction error;

[0072] Gradually increase the model order and observe the mean square error. The changing trend, select the mean squared error The optimal model order is the order at which the price decreases significantly but has not yet leveled off, thus avoiding underfitting or overfitting and achieving the goal of accurately modeling and predicting time series data.

[0073] It should be noted that the model inverse obtained based on recursive least squares identification may be unstable, leading to an exponential increase in perturbation observations, severe oscillations in the control signal, and divergence in the control system. Therefore, it is necessary to construct an approximate inverse of the nominal model based on Zero Phase Error Tracking Control (ZPETC). When taking the approximate inverse of the model, firstly, poles and zeros are considered, and the original poles are replaced with mirrored poles to ensure stability. Secondly, a delay term is introduced to compensate for the difference in relative order and avoid non-causal inverses. Finally, gain adjustment is used to sacrifice high-frequency accuracy in exchange for tracking performance in the low-frequency range (such as near the fundamental frequency).

[0074] S13: Using a time delay operator By shifting the analysis of the control system from the discrete-time domain to the frequency domain, the transfer function becomes:

[0075] ,

[0076] in: Equivalent to ;

[0077] Zero points, including stable zero points and unstable zeros , As the extreme point, For gain, Pure delay;

[0078] Decompose the zero point into stable zero points. and unstable zeros :

[0079] ,

[0080] ;

[0081] Constructing an approximate inverse of the nominal model based on zero-phase error tracking control:

[0082] ,

[0083] ;

[0084] in, This is the gain compensation coefficient.

[0085] S2: A mathematical model of the perturbation is established online based on the recursive least squares method. The identification process of the perturbation model does not require external excitation signals and the input is zero. By monitoring the dynamic deviation between the system output and the nominal model in real time, a dynamic estimation model of the perturbation is constructed. Taking a fourth-order model as an example, the perturbation model is as follows:

[0086] ,

[0087] Wherein, parameter vector Measurement vector .

[0088] Then, using the line-of-sight angular velocity sensed by the gyroscope, a PI control is employed to form a feedback closed loop for line-of-sight stabilization control. Based on this disturbance dynamic model, multi-step forward prediction is performed, that is, predicting the equivalent disturbance for the next N steps (the number of prediction steps can be determined according to the actual system application, such as 1 step, 2 steps, 3 steps, etc.). :

[0089] ,

[0090] Wherein, parameter vector Measurement vector L is the dimension of the measurement vector.

[0091] S3: Utilizing the angular velocity sensed by the gyroscope, a feedback closed loop for line-of-sight stabilization control is constructed, which converts the predicted and estimated equivalent disturbances. With real-time disturbance Weighted fusion is then performed, and the resulting disturbance prediction compensation signal is then... Feedforward to the control input to offset the impact of actual disturbances on the system and achieve adaptive disturbance prediction and compensation.

[0092] Disturbance prediction compensation signal for:

[0093] ,

[0094] in, Weights for predicting confidence levels.

[0095] By simulating the motion disturbances experienced by a vehicle-mounted optoelectronic turret during maneuvering using a swing platform, quantitative testing and analysis of its control performance were conducted under typical operating conditions. For example... Figure 3-4 As shown, experimental results indicate that compared to the currently used dual-PI control gyroscope velocity closed-loop stabilization control method, the control strategy proposed in this invention improves the stability accuracy of a two-axis, two-frame photoelectric turret from 30.98urad to 13.57urad under a disturbance of 1° / 1Hz, representing a 56.2% improvement; and under a disturbance of 1° / 2Hz, the stability accuracy of the photoelectric turret improves from 104.5urad to 44.68urad, representing a 57.24% improvement.

[0096] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A method for stabilizing the line-of-sight control of a photoelectric turret based on adaptive disturbance prediction compensation, characterized in that, include: S1: Establishing a second-order discrete model of the gyroscope velocity loop based on the recursive least squares method: ， Wherein, parameter vector Measurement vector ; The deviation between the output of the second-order discrete model of the gyroscope velocity loop and the line-of-sight angular velocity measured by the gyroscope is calculated, and the real-time disturbance estimate is obtained by passing the inverse model and a low-pass filter. : ， in, The output of the identified nominal model is... The output after inverting the nominal model. The angular velocity of the gyroscope's line of sight. It is a low-pass filter; S2: Construct a mathematical model of the perturbation based on the recursive least squares method, and predict the equivalent perturbation in the next N steps. : ， Wherein, parameter vector Measurement vector L is the dimension of the measurement vector; S3: Utilizing the angular velocity sensed by the gyroscope, a feedback closed loop for line-of-sight stabilization control is constructed, which converts the predicted and estimated equivalent disturbances. With real-time disturbance Weighted fusion yields disturbance prediction compensation signal Then, the disturbance prediction compensation signal Feedforward to the control input to achieve adaptive disturbance prediction and compensation for photoelectric turret line-of-sight stabilization control.

2. The photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation according to claim 1, characterized in that, Discrete model identification of the gyro velocity loop in S1 includes: S11: Estimate model parameters using a recursive least squares parameter adaptive law, index function: ， S12: Apply the same control input signal to the input terminals of the actual system and the identification model to construct the model evaluation function; verify the degree of fit between the constructed model and the actual system by real-time statistical mean square error. Mean square error calculation for: ε is the prior prediction error; Gradually increase the model order and observe the mean square error. The changing trend, select the mean squared error The order at which the price decreases significantly but has not yet leveled off is taken as the optimal model order.

3. The photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation according to claim 2, characterized in that, Discrete model identification of the gyro velocity loop in S1 also includes: S13: Using a time delay operator By shifting the analysis of the control system from the discrete-time domain to the frequency domain, the transfer function becomes: ， in: Equivalent to ; Zero points, including stable zero points and unstable zeros , As the extreme point, For gain, Pure delay; Decompose the zero point into stable zero points. and unstable zeros : ， ； Constructing an approximate inverse of the nominal model based on zero-phase error tracking control: ， ； in, This is the gain compensation coefficient.

4. The photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation according to claim 1, characterized in that, The perturbation model is a fourth-order model, and the perturbation model is as follows: ， Wherein, parameter vector Measurement vector .

5. The photoelectric turret line-of-sight stabilization control method based on adaptive disturbance prediction compensation according to claim 1, characterized in that, The equivalent disturbance predicted in S3 will be estimated. With real-time disturbance Weighted fusion yields the disturbance prediction compensation signal. for: ， in, Weights for predicting confidence levels.

6. A computer product comprising computer program instructions, characterized in that, When executed by a processor, the computer program instructions implement the steps of the method according to any one of claims 1-5.

Images