A rapid depth control method for underwater vehicles based on imbalance prediction and compensation

By estimating and mapping the imbalance in real time, and combining theoretical and data-driven model prediction, the timing of water injection and drainage is determined, which solves the problem of control instability during the underwater vehicle's depth increase and achieves rapid and stable depth control.

CN123113952APending Publication Date: 2026-11-17CHINA SHIP SCIENTIFIC RESEARCH CENTER
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611513067.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

As existing underwater vehicles deepen, the rate of deepening decreases due to the gradual change in imbalance, and even when they reach the target depth, the rudder effect is insufficient, making it impossible to achieve stable control.

Method used

By estimating the imbalance in real time, a depth-imbalance mapping relationship is established. The theoretical model and the data-driven model are integrated to predict the target depth imbalance. Based on the depth efficiency analysis, the timing of water inlet and outlet is determined to achieve early compensation for the imbalance.

Benefits of technology

This significantly improves the depth-deployment efficiency of underwater vehicles, avoids control instability caused by excessive imbalance, and achieves rapid and stable depth control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure SMS_3
    Figure SMS_3
  • Figure SMS_8
    Figure SMS_8
  • Figure SMS_9
    Figure SMS_9
Patent Text Reader

Abstract

This invention relates to a rapid depth-changing control method for underwater vehicles based on imbalance prediction and compensation. It utilizes a motion controller for depth and attitude control, enabling the underwater vehicle to travel towards a target depth. During the depth-changing process, disturbance observation technology is used to estimate disturbance values ​​in real time and calculate the imbalance, establishing a mapping relationship between the imbalance and depth values. Imbalance prediction based on theoretical models and data-driven methods are employed, and their trend increments are adaptively fused to obtain the predicted imbalance value for the target depth. Based on the predicted value, the impact of the imbalance on the depth-changing rate is determined, and the timing of the buoyancy adjustment system's injection and dewatering is determined to achieve early compensation for the imbalance. This invention, through the fusion of prediction and depth-changing efficiency analysis, can significantly improve the depth-changing efficiency of underwater vehicles, avoid repeated injection and dewatering of the buoyancy adjustment system, and is suitable for rapid depth-changing control of manned submersibles or unmanned underwater vehicles.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the technical field of underwater vehicle control methods, and in particular to a rapid depth-changing control method for underwater vehicles based on imbalance quantity prediction and compensation. Background Technology

[0002] Underwater vehicles are typically equipped with a motion controller, rudder, and buoyancy control system. The motion controller is used to control the rudder and buoyancy control system to control the depth and attitude of the underwater vehicle.

[0003] As an underwater vehicle travels from its current depth to its target depth, it is affected by factors such as seawater density and volume compression, which can cause equivalent interference to the underwater vehicle. This type of equivalent interference is the main source of interference for conventional underwater vehicles.

[0004] The equivalent disturbance caused by seawater density and volume compression to underwater vehicles can be termed the unbalanced quantity of the underwater vehicle. In existing technologies, underwater vehicles typically achieve dynamic depth-changing through steering. After reaching and stabilizing at the target depth, the steady-state unbalanced quantity is estimated using a linear equilibrium formula, and then eliminated through buoyancy adjustment systems to pump water. However, during depth-changing, the gradually changing unbalanced quantity may reduce the depth-changing rate of the underwater vehicle. Even when switching to zero-pitch descent near the target depth, excessive unbalanced quantity may render the steering effect insufficient to achieve zero-pitch depth control, making stable control of the target depth impossible. Therefore, under the influence of ocean density changes and unbalanced quantities caused by the compression of the underwater vehicle's pressure hull, a faster and more feasible depth-changing control method should be considered when underwater vehicles are attempting to change depth. Summary of the Invention

[0005] To address the shortcomings of existing production technologies, this applicant provides a rapid depth-changing control method for underwater vehicles based on imbalance prediction and compensation. This method estimates the imbalance in real time, establishes a depth-imbalance mapping relationship, integrates theoretical and data-driven models to predict the target depth imbalance, and determines the timing of water injection and drainage based on depth-changing efficiency analysis, thereby achieving advance compensation for the imbalance and significantly improving the depth-changing efficiency of the underwater vehicle.

[0006] The technical solution adopted in this invention is as follows: A rapid depth-changing control method for underwater vehicles based on imbalance quantity prediction and compensation includes the following operational steps: S1. Use the underwater vehicle's motion controller to control depth and attitude, enabling the underwater vehicle to travel from the current depth to the target depth. S2. During the deepening process, the disturbance value d experienced by the underwater vehicle is estimated in real time using disturbance observation technology, and the real-time depth is collected; the corresponding imbalance quantity UB is calculated based on the disturbance value d, and the mapping relationship between the imbalance quantity and the depth value is established. S3. Based on the mapping relationship, a prediction algorithm is used to predict the imbalance of the target depth. The prediction algorithm includes prediction based on theoretical models and prediction based on data-driven methods. The trend increments of the two are adaptively fused to obtain the predicted value of the imbalance of the target depth. S4. Based on the predicted value of the imbalance at the target depth, determine the impact of the imbalance on the rate of deepening, determine the start-up timing of the buoyancy adjustment system for water injection and drainage, and achieve advance compensation for the imbalance.

[0007] Its further technical solution lies in: In step S2, establishing the mapping relationship between the imbalance quantity and the depth value includes: discretizing the depth range to obtain discrete depth values ​​η1, η2, ..., η N When the real-time depth first crosses the discrete depth value η i At that time, record the imbalance amount UB at that depth. i Construct data points (η) i UB i ).

[0008] In step S2, the disturbance observation technique is an extended state observer, which estimates the disturbance d based on the real-time rudder angle, depth and attitude of the underwater vehicle, and calculates the imbalance amount according to UB = d / g, where g is the gravitational acceleration.

[0009] In step S3, the prediction based on the theoretical model includes: calculating the theoretical imbalance of the underwater vehicle at each depth based on the density values ​​at various depths in the current sea area and the compressibility characteristics of the pressure hull of the underwater vehicle, and obtaining the function of the theoretical imbalance as a function of depth. .

[0010] Theoretical imbalance quantity change function Approximate expression:

[0011] Among them, V all V is the total underwater displacement volume of a known underwater vehicle. n It is the known volume of the pressure hull of an underwater vehicle. It is the compression ratio coefficient of the pressure hull per unit depth. η is the density of seawater at a known depth of η, but there is a certain error compared to the actual density of seawater; g is the acceleration due to gravity.

[0012] In step S3, the data-driven prediction utilizes the autoregressive integral moving average model ARIMA(p,r,q), based on the data point set Γ={(η1,UB1), (η2,UB2), ..., (η...} obtained in step S2. i UB i )} Predicted target depth The imbalance quantity prediction model is as follows: .

[0013] The ARIMA model parameters include an autoregressive order p=2, a difference order r=1, and a moving average order q=1.

[0014] In step S3, When the underwater vehicle reaches a discrete value η at a depth each time... i When i can take values ​​of 1, 2, ..., N, a fusion prediction calculation is performed, and the trend increments of the theoretical model and the data-based prediction model are defined as follows:

[0015] Where k takes values ​​from 1, 2, ..., Ni; An adaptive weighting factor is introduced to achieve the fusion of trend increments as follows:

[0016] Where λ(η+k)∈[0,1] is the weighting factor. To satisfy the desired effect of a high proportion of the predicted trend in the early stage and a high proportion of the baseline trend in the later stage, a weighting function with exponential decay characteristics is constructed as follows:

[0017] Wherein, λ0∈(0,1) is the initial predicted trend weight, and it is recommended to take a value of 0.8~0.9; K is the set weight decay factor; In η i+k The recursive formula for the correction value of the depth imbalance prediction is as follows:

[0018] Therefore, the approximate value of the unevenness in the target depth can be obtained as follows: .

[0019] In step S4, Determine whether the imbalance is beneficial to the deepening efficiency; if the predicted imbalance values ​​at the starting and ending points of deepening satisfy the following formula, it indicates that the imbalance is detrimental to deepening. In this case, a command value of F(η) can be directly sent to the buoyancy adjustment system. N Upon receiving the injection / drainage command, immediately initiate injection / drainage to eliminate the imbalance.

[0020] Otherwise, if the imbalance is beneficial for the underwater vehicle to reach greater depth, then utilizing the effect of the imbalance, the later the injection and evacuation are initiated before the underwater vehicle reaches the target depth, the better. Considering the rate of equalization in deepening, the theoretical time required to eliminate the imbalance through injection and evacuation is approximately calculated as follows:

[0021] Where Vinout is the injection / displacement rate of the buoyancy regulation system; by assuming that the rate of depth change of the underwater vehicle remains constant during a significant depth change, the approximate time for the underwater vehicle to reach the target depth from the current depth can be obtained as follows:

[0022] Where u is the target speed of the underwater vehicle; θ max It is the maximum pitch angle of the target as the underwater vehicle deepens; w max It is the maximum vertical speed of the underwater vehicle.

[0023] The method is applied to rapid depth control of manned submersibles or unmanned underwater vehicles.

[0024] The beneficial effects of this invention are as follows: (1) This invention proposes a systematic method for predicting and compensating for imbalance during the depth change process, which greatly improves the depth change efficiency of underwater vehicles.

[0025] Traditional underwater vehicles typically achieve dynamic depth progression through steering, and after reaching and stabilizing at the target depth, estimate the steady-state imbalance using a linear equilibrium formula. This imbalance is then eliminated by filling or draining water through a buoyancy adjustment system. This "depth progression first, equilibrium later" approach has significant drawbacks: the gradually changing imbalance during depth progression can reduce the rate of depth progression, and even when switching to zero-pitch descent near the target depth, excessive imbalance can render the steering ineffective for achieving zero-pitch depth control, thus failing to achieve stable control at the target depth. By estimating and predicting the imbalance at the target depth in real time during depth progression, and initiating water filling or draining water before reaching the target depth to eliminate the imbalance, parallel control—deepening, prediction, and compensation—is achieved. This significantly improves the depth progression efficiency of the underwater vehicle and avoids the problem of unstable control due to excessive imbalance after reaching the target depth.

[0026] (2) The present invention designs a method for predicting unbalanced quantities by integrating theoretical models and data-driven approaches, thereby achieving a unity of long-term reliable prediction and short-term high-precision prediction.

[0027] This paper integrates imbalance prediction based on theoretical models with data-driven imbalance prediction. Predictions based on real-time estimation data have high accuracy in the short term, but may exhibit significant trend deviations in long-term predictions. Predictions based on theoretical models may deviate somewhat from the actual values, but overall, the long-term trend patterns are more reliable. By using differential operations to obtain the trend increments of both prediction models, an adaptive weighting factor that decays exponentially with the prediction step size is introduced. In the short term, real-time prediction trends are primarily relied upon, while in the long term, theoretical trends from the sea area are adopted. This achieves a unification of reliable long-term and high-precision short-term imbalance prediction. Reliable long-term prediction can avoid magnitude deviations in the prediction of imbalance at the target depth during the initial depth-changing phase of the underwater vehicle, leading to incorrect timing of premature buoyancy injection and dewatering. High-precision short-term prediction can more accurately predict the imbalance at the target depth as the underwater vehicle approaches, facilitating early and accurate elimination of imbalance and avoiding repeated buoyancy injection and dewatering of the buoyancy adjustment system.

[0028] (3) The present invention determines the start-up timing of water injection and drainage based on variable depth efficiency analysis, thereby achieving optimal compensation for unbalanced quantities.

[0029] Based on the direction of the impact of the imbalance on the depth-changing rate, differentiated decisions are made regarding the timing of water injection / discharge initiation: if the equivalent disturbance of the imbalance is beneficial to the depth-changing rate, water injection / discharge is initiated as late as possible to fully utilize the imbalance and reduce the amplitude of rudder movements; if the equivalent disturbance of the imbalance is detrimental to the depth-changing rate, water injection / discharge is initiated as early as possible to accelerate the depth-changing rate. This invention determines the optimal timing for water injection / discharge initiation by comparing the theoretical time required to complete the imbalance elimination with the time it takes for the underwater vehicle to reach the target depth from the current depth. This achieves automatic pre-compensation for imbalance based on imbalance prediction. During the water injection / discharge process, the predicted imbalance value is used to control the amount of water injection / discharge, achieving one water injection / discharge per depth-changing stage, avoiding energy waste and system wear caused by repeated water injection / discharge.

[0030] (4) This invention establishes a mapping relationship between imbalance quantity and depth value, providing a reliable data basis for prediction.

[0031] During the depth-adjusting process, an extended state observer is used to estimate the disturbance values ​​experienced by the underwater vehicle in real time and collect real-time depth data. When the depth first crosses a discrete depth value, the imbalance quantity at that depth is recorded, constructing data points for depth and imbalance quantity, and establishing a mapping relationship between imbalance quantity and depth value. This mapping relationship provides a reliable data foundation for data-driven prediction models, enabling them to accurately capture the actual patterns of imbalance quantity changes with depth, thus improving the accuracy and reliability of predictions.

[0032] (5) The present invention adapts to the uneven amount caused by changes in ocean density and compression of the pressure hull of underwater vehicles, and is suitable for deep-sea variable-depth working conditions.

[0033] As underwater vehicles deepen, they experience gradually varying imbalances due to changes in seawater density and volume compression. This invention establishes a theoretical imbalance prediction model by comprehensively considering density and the compression of the underwater vehicle's pressure hull. This model is adaptable to density variations and pressure hull compression characteristics in different sea areas and at different depths, making it suitable for rapid depth control under deep-sea conditions. It exhibits good environmental adaptability and engineering practicality.

[0034] (6) The method and process are clear, the amount of calculation is moderate, and it is easy to implement in engineering.

[0035] The control method includes steps such as depth and attitude control, establishment of imbalance mapping relationships, fusion prediction, and injection / discharge timing decision-making. The process is clear, the calculation methods for each step are well-defined, and the computational load is moderate, making it easy to implement in existing underwater vehicle control systems. It can be applied to the rapid depth control of manned submersibles or unmanned underwater vehicles, and has broad application prospects. Detailed Implementation

[0036] The specific embodiments of the present invention are described below.

[0037] The specific operation steps of the rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation described in this embodiment are as follows: (1) Depth and attitude control of underwater vehicles: Assume the initial depth of the underwater vehicle is η. s After the user sets the target depth η e η e >η s The underwater vehicle's navigation motion controller uses conventional PID control to calculate the desired rudder angle in real time and send it to the servo motor, thereby enabling the underwater vehicle to navigate towards the target depth.

[0038] (2) Mapping relationship between the imbalance of underwater vehicles and depth values: Step 2.1: Determine the depth range η of the underwater vehicle. s and η e Discretize the depth values ​​between them to obtain discrete depth values ​​η1, η2, ..., η N for

[0039] in, It is the set depth distance from the walking distance.

[0040] Step 2.2: Real-time estimation of underwater vehicle interference. The navigation motion controller estimates the interference d experienced by the underwater vehicle based on its real-time rudder angle, depth, and attitude using a mature extended state observer.

[0041] Step 2.3: Based on the disturbance d, calculate the corresponding imbalance quantity as UB = d / g. g is the acceleration due to gravity.

[0042] Step 2.4: As the underwater vehicle travels deeper, the depth will gradually transition from η1 to η. N During this period, the depth values ​​of the underwater vehicle are collected in real time. When the depth first crosses the discrete value η... i (The value of i ranges from 1, 2, ..., N), then the imbalance at this time is recorded as UB. i Construct depth and imbalance data points (η) at this depth. i UB i+1 ).

[0043] (3) Prediction of imbalance during the underwater vehicle's depth increase process: Step 3.1: Prediction of underwater vehicle imbalance based on theoretical model.

[0044] During the underwater vehicle's ascent, the database contains environmental information about the current sea area, including density values ​​at various depths. Using this information, the theoretical buoyancy imbalance of the underwater vehicle at each depth can be calculated and fitted. Therefore, a function relating the underwater vehicle's buoyancy imbalance to depth can be constructed. Assuming the initial depth of the underwater vehicle is 0, and the buoyancy imbalance is zero at this point, considering both density and the compression of the underwater vehicle's pressure hull, the theoretical buoyancy change caused by external factors during the ascent can be approximated as follows: (1) Among them, V all V is the total underwater displacement volume of a known underwater vehicle. n It is the known volume of the pressure hull of an underwater vehicle. It is the compression ratio coefficient of the pressure hull per unit depth. η is the density of seawater at a known depth of η, but there is a certain error compared to the actual density of seawater; g is the acceleration due to gravity.

[0045] Therefore, according to equation (1), the theoretical imbalance generated by the underwater vehicle before and after the depth change can be predicted as follows: (2) in, and These are the depths before and after the increase in depth.

[0046] Step 3.2: Data-based underwater vehicle imbalance prediction model.

[0047] Based on the actual data point set Γ={(η1, UB1), (η2, UB2), ..., (η... i UB i This paper uses the autoregressive integral moving average model ARIMA(p,r,q) to predict the imbalance of an underwater vehicle at a target depth. Here, p represents the autoregressive order (which can be 2), r is the difference order (which can be 1), and q is the moving average order (which can be 1). Based on the data point set Γ, let the target depth predicted by the autoregressive integral moving average model ARIMA(2,1,1) be denoted. The imbalance quantity prediction model is (3) Step 3.3: Prediction of underwater vehicle imbalance based on incremental fusion.

[0048] Predictions based on real-time estimated data using Equation (3) have high accuracy in the short term, but may show significant trend deviations in long-term predictions. While theoretical imbalance values ​​based on Equation (2) may deviate from actual values, the overall long-term trend predictions should be more reliable. Therefore, this paper integrates the two at the trend level, relying primarily on real-time predicted trends for short-term predictions and adopting theoretical trends from the sea area for long-term predictions.

[0049] This embodiment proposes an adaptive fusion prediction method for trend increments. It utilizes differential operations to obtain trend increments based on both a theoretical imbalance prediction model and a data imbalance prediction model. An adaptive weighting factor related to the prediction length is introduced to fuse the trend increments of the two models, thereby achieving long-term imbalance prediction. First, when the underwater vehicle reaches a discrete value η at a certain depth each time... i When i can take values ​​of 1, 2, ..., N, a fusion prediction calculation is performed. The trend increments based on the theoretical model and the data-based prediction model are defined as follows: (4) Where k takes values ​​from 1, 2, ..., Ni.

[0050] An adaptive weighting factor is introduced to achieve the fusion of trend increments as follows: (5) Where λ(η+k)∈[0,1] is the weighting factor. To satisfy the desired effect of a high proportion of the predicted trend in the early stage and a high proportion of the baseline trend in the later stage, a weighting function with exponential decay characteristics is constructed as follows: (6) Wherein, λ0∈(0,1) is the initial predicted trend weight, and it is recommended to take 0.8~0.9; K is the set weight decay factor.

[0051] In η i+k The recursive formula for the correction value of the depth imbalance prediction is as follows: (7) Therefore, the approximate value of the unevenness in the target depth can be obtained as follows: .

[0052] (4) Pre-compensation for imbalance in underwater vehicles: Imbalance is a disturbance to the underwater vehicle. Since the underwater vehicle's injection and dewatering rate is very low, to improve the underwater vehicle's depth-deployment efficiency, imbalance is analyzed and evaluated during compensation. This allows for the selection of an appropriate injection and dewatering start time, enabling automatic pre-emptive compensation based on imbalance prediction. The main compensation strategy is as follows: (a) If the disturbance equivalent to the unbalanced amount is beneficial to the deepening rate, the injection and drainage should be carried out as late as possible to make full use of the unbalanced amount to reduce the amplitude of the rudder. (b) If the disturbance equivalent to the unbalanced quantity is detrimental to the deepening rate, then water injection and drainage should be carried out as early as possible to accelerate the deepening rate. During the water injection and drainage process, the water injection and drainage volume should be controlled in conjunction with the predicted value of the unbalanced quantity, so as to achieve one water injection and drainage per deepening stage.

[0053] Based on the above principles, the compensation process for imbalances is as follows: First, determine whether the imbalance is beneficial to the deepening efficiency. If the predicted imbalance values ​​at the starting and ending points of the deepening satisfy equation (8), it indicates that the imbalance is not conducive to deepening. In this case, a command value of F(η) can be directly sent to the buoyancy adjustment system. N Upon receiving the injection / drainage command, immediately initiate injection / drainage to eliminate the imbalance.

[0054] (8) Otherwise, if the imbalance is beneficial for the underwater vehicle to reach greater depth, then utilizing the effect of the imbalance, the later the injection and evacuation are initiated before the underwater vehicle reaches the target depth, the better. Considering the rate of equalization in deepening, the theoretical time required to eliminate the imbalance through injection and evacuation is approximately calculated as follows: (9) Where Vinout is the injection / displacement rate of the buoyancy regulation system. By assuming that the rate of depth change of the underwater vehicle remains constant during a significant depth change, the approximate time for the underwater vehicle to reach the target depth from the current depth can be obtained as follows: (10) where u is the set target navigation speed of the underwater vehicle, which generally remains constant throughout the entire depth change process; θ max is the maximum value of the target pitch angle when the underwater vehicle changes depth; w max is the maximum vertical speed of the underwater vehicle.

[0055] Comparing Equation (9) and Equation (10), if tbu<te is satisfied, the process feeds back to step 2.1 for cyclic execution and iterative prediction analysis. Otherwise, the buoyancy adjustment system is used to start water injection and discharge to eliminate the unbalanced quantity.

[0056] This embodiment performs unbalanced quantity prediction during the depth change process, analyzes to obtain an appropriate timing to start water injection and discharge in advance to eliminate the unbalanced quantity, which can greatly improve the depth change efficiency of the underwater vehicle.

[0057] This embodiment is an unbalanced fusion prediction method that integrates a theoretical unbalanced quantity prediction model and a data-based unbalanced quantity prediction model, gives full play to the combined advantages of the two prediction models, and further realizes high-precision short-term prediction and credible long-term prediction of the unbalanced quantity. The long-term credible prediction can avoid the problem that the underwater vehicle has a magnitude deviation in the prediction of the unbalanced quantity at the target depth in the initial depth change stage, which leads to an incorrect timing of advance water injection and discharge; the short-term high-precision prediction can more accurately predict the unbalanced quantity at the target depth as the underwater vehicle gets closer to the target depth, which is conducive to realizing the advance high-precision elimination of the unbalanced quantity at the target depth, thereby equivalently improving the depth change efficiency of the underwater vehicle and avoiding repeated water injection and discharge of the buoyancy adjustment system.

[0058] The above description is an explanation of the present invention, not a limitation to the present invention. The scope defined by the present invention refers to the claims, and any form of modification can be made within the protection scope of the present invention.

Claims

1. A rapid depth-changing control method for underwater vehicles based on imbalance quantity prediction and compensation, characterized in that: The following steps are included: S1. Use the underwater vehicle's motion controller to control depth and attitude, enabling the underwater vehicle to travel from the current depth to the target depth. S2. During the deepening process, the disturbance value d experienced by the underwater vehicle is estimated in real time using disturbance observation technology, and the real-time depth is collected; the corresponding imbalance quantity UB is calculated based on the disturbance value d, and the mapping relationship between the imbalance quantity and the depth value is established. S3. Based on the mapping relationship, a prediction algorithm is used to predict the imbalance of the target depth. The prediction algorithm includes prediction based on theoretical models and prediction based on data-driven methods. The trend increments of the two are adaptively fused to obtain the predicted value of the imbalance of the target depth. S4. Based on the predicted value of the imbalance at the target depth, determine the impact of the imbalance on the rate of deepening, determine the start-up timing of the buoyancy adjustment system for water injection and drainage, and achieve advance compensation for the imbalance.

2. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: In step S2, establishing the mapping relationship between the imbalance quantity and the depth value includes: discretizing the depth range to obtain discrete depth values ​​η1, η2, ..., η N When the real-time depth first crosses the discrete depth value η i At that time, record the imbalance amount UB at that depth. i Construct data points (η) i UB i ).

3. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: In step S2, the disturbance observation technique is an extended state observer, which estimates the disturbance d based on the real-time rudder angle, depth and attitude of the underwater vehicle, and calculates the imbalance amount according to UB = d / g, where g is the gravitational acceleration.

4. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: In step S3, the prediction based on the theoretical model includes: calculating the theoretical imbalance of the underwater vehicle at each depth based on the density values ​​at various depths in the current sea area and the compressibility characteristics of the pressure hull of the underwater vehicle, and obtaining the function of the theoretical imbalance as a function of depth. .

5. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 4, characterized in that: Theoretical imbalance quantity change function Approximate expression: Among them, V all V is the total underwater displacement volume of a known underwater vehicle. n It is the known volume of the pressure hull of an underwater vehicle. It is the compression ratio coefficient of the pressure hull per unit depth. η is the density of seawater at a known depth of η, but there is a certain error compared to the actual density of seawater; g is the acceleration due to gravity.

6. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: In step S3, the data-driven prediction utilizes the autoregressive integral moving average model ARIMA(p,r,q), based on the data point set Γ={(η1,UB1), (η2,UB2), ..., (η...} obtained in step S2. i UB i )} Predicted target depth The imbalance quantity prediction model is as follows: 。 7. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 6, characterized in that: The ARIMA(p,r,q) model uses parameters with autoregression order p=2, difference order r=1, and moving average order q=1.

8. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: In step S3, When the underwater vehicle reaches a discrete value η at a depth each time... i When i can take values ​​of 1, 2, ..., N, a fusion prediction calculation is performed, and the trend increments of the theoretical model and the data-based prediction model are defined as follows: Where k takes values ​​from 1, 2, ..., Ni; An adaptive weighting factor is introduced to achieve the fusion of trend increments as follows: Where λ(η+k)∈[0,1] is the weighting factor. To satisfy the desired effect of a high proportion of the predicted trend in the early stage and a high proportion of the baseline trend in the later stage, a weighting function with exponential decay characteristics is constructed as follows: Wherein, λ0∈(0,1) is the initial predicted trend weight, and it is recommended to take a value of 0.8~0.9; K is the set weight decay factor; In η i+k The recursive formula for the correction value of the depth imbalance prediction is as follows: Therefore, the approximate value of the unevenness in the target depth can be obtained as follows: .

9. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: In step S4, Determine whether the imbalance is beneficial to the deepening efficiency; if the predicted imbalance values ​​at the starting and ending points of deepening satisfy the following formula, it indicates that the imbalance is detrimental to deepening. In this case, a command value of F(η) can be directly sent to the buoyancy adjustment system. N Upon receiving the injection / drainage command, immediately initiate injection / drainage to eliminate the imbalance. Otherwise, if the imbalance is beneficial for the underwater vehicle to reach greater depth, then utilizing the effect of the imbalance, the later the injection and evacuation are initiated before the underwater vehicle reaches the target depth, the better. Considering the rate of equalization in deepening, the theoretical time required to eliminate the imbalance through injection and evacuation is approximately calculated as follows: Where Vinout is the injection / displacement rate of the buoyancy regulation system; by assuming that the rate of depth change of the underwater vehicle remains constant during a significant depth change, the approximate time for the underwater vehicle to reach the target depth from the current depth can be obtained as follows: Where u is the target speed of the underwater vehicle; θ max It is the maximum pitch angle of the target as the underwater vehicle deepens; w max It is the maximum vertical speed of the underwater vehicle.

10. The rapid depth control method for underwater vehicles based on imbalance quantity prediction and compensation as described in claim 1, characterized in that: The method is applied to rapid depth control of manned submersibles or unmanned underwater vehicles.