A stability control method of a camera system

Abstract

The application discloses a kind of stability regulation methods of camera system, comprising the following steps: S1: the installation position information of camera system is collected in real time by multi-source sensor array, the installation position information includes topographic elevation data, pedestal vibration frequency spectrum and environmental temperature and humidity distribution parameter;S2: three-dimensional stability evaluation model is constructed, and installation position information is input into three-dimensional stability evaluation model to generate stability coefficient matrix;When any dimension parameter in stability coefficient matrix exceeds preset safety threshold, activate self-balancing adjustment mechanism to carry out multi-degree-of-freedom compensation adjustment.The application solves the core problems of insufficient stability, response lag and single protection means of traditional regulation method in complex dynamic environment by fusing environmental perception, intelligent prediction and multi-modal collaborative compensation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of stability regulation, and particularly relates to a stability regulation method of a camera system. BACKGROUND

[0002] The camera system is widely used in the fields of security monitoring, industrial detection, and film shooting. The automatic following camera system realizes the continuous tracking of a moving target by integrating visual recognition, sensors (such as a gyroscope and an accelerometer), and servo control technology. This kind of system relies on the multi-degree-of-freedom motion (such as pitching and rotating) of a holder to keep the position of the target stable in the picture. However, in a complex dynamic environment (such as strong wind, mechanical vibration, and terrain undulation), external disturbances easily lead to the instability of the holder posture, causing problems such as picture shaking, target loss, or imaging blur, which seriously restricts the reliability and application range of the system. Therefore, a stability regulation method needs to be used to intelligently regulate the stability of the camera system.

[0003] The existing stability regulation method adopts a single compensation strategy in the stability regulation process, which leads to poor stability regulation effect and brings certain influence to the use of the stability regulation method. Therefore, a stability regulation method of a camera system is proposed. SUMMARY

[0004] The technical problem to be solved by the present application is how to solve the problem of the existing stability regulation method, which adopts a single compensation strategy in the stability regulation process, leading to poor stability regulation effect. The present application provides a stability regulation method of a camera system.

[0005] The present application solves the above technical problems by the following technical scheme. The present application comprises the following steps:

[0006] S1: Real-time acquisition of installation position information of the camera system by a multi-source sensor array, wherein the installation position information comprises terrain elevation data, base vibration frequency spectrum, and environmental temperature and humidity distribution parameters;

[0007] S2: Construction of a three-dimensional stability evaluation model, input of the installation position information into the three-dimensional stability evaluation model to generate a stability coefficient matrix, and activation of a self-balancing adjustment mechanism for multi-degree-of-freedom compensation adjustment when any dimension parameter in the stability coefficient matrix exceeds a preset safety threshold;

[0008] S3: Following monitoring by the camera system, real-time acquisition of frequency domain features of the holder motion trajectory by a dynamic frequency analysis module during the following monitoring process, establishment of a mapping relationship between the motion features and the stability, generation of an adaptive damping adjustment instruction based on the mapping relationship between the motion features and the stability;

[0009] S4: Real-time monitoring of the dynamic change rate of the pan-tilt rotation angle of the camera system by using an inertial measurement unit, triggering the pre-tightening force compensation mechanism and generating a reverse torque compensation parameter when the angle change rate exceeds the preset critical curve;

[0010] S5: Establishing an abnormal state grading response mechanism, starting a first compensation strategy when the system stability margin is lower than the first warning threshold, switching to a safe locking mode and generating a fault positioning report when it is lower than the second emergency threshold.

[0011] Further, the construction process of the three-dimensional stability evaluation model in S2 is as follows:

[0012] The terrain adaptation degree is calculated, the stress distribution under different terrain conditions is calculated by finite element analysis, and the terrain adaptation factor is defined as:

[0013] ;

[0014] α is the terrain adaptation factor, which is used to quantify the adaptation degree of the installation position to the stability of the camera system (0≤α≤1);

[0015] is the material yield strength of the camera system support structure;

[0016] is the actual stress value at the i-th finite element grid node (unit: MPa);

[0017] N is the total number of nodes of the finite element grid division;

[0018] The vibration response prediction is performed, the transfer function method is used to establish the correlation model of the base vibration and the pan-tilt shaking, and the vibration transfer gain is defined as:

[0019] ;

[0020] In the formula, is the vibration transfer gain (unit: dB), which represents the amplification effect of the base vibration transferred to the pan-tilt;

[0021] is the vibration power spectral density of the pan-tilt, is the vibration power spectral density of the base;

[0022] When the adaptation factor α of any dimension is less than 0.8 or the vibration gain >3dB, the self-balancing adjustment mechanism is triggered.

[0023] Further, the analysis process of the dynamic frequency analysis module is as follows:

[0024] The three-axis vibration data is collected by the acceleration sensor group at a sampling rate of 1000 Hz;

[0025] The three-axis vibration acceleration signal a(t) is collected at a sampling rate of 1000 Hz, and the frequency domain feature vector F is constructed by Fourier transform:

[0026] ;

[0027] In the formula: F is a frequency domain feature vector (a complex vector), representing the energy distribution of the vibration signal at different frequencies, T is the sampling window length (unit: second), a(t) is the time domain vibration acceleration signal (unit: m / s²), ωk is the frequency value of the kth frequency point (unit: Hz), j is the imaginary unit, and satisfies ;

[0028] F is input into a pre-trained convolutional neural network (CNN), and the stability score S is output :

[0029] ;

[0030] In the formula, S is the stability score, and the higher the score, the better the stability, Fi is the ith frequency domain feature vector;

[0031] wi is the ith weight parameter in the convolutional neural network, and b is the bias term of the neural network;

[0032] ReLU is a rectified linear unit activation function, defined as ReLU(x)=max(0,x);

[0033] When <S threshold , an adaptive damping adjustment instruction is generated.

[0034] Further, the specific process of S4 is:

[0035] The differential equation of the angle change rate is established: ;

[0036] In the formula, ω is the change rate of the pan-tilt rotation angle, θ is the pan-tilt rotation angle, T is the torque output by the driving motor, Te is the equivalent torque caused by environmental interference (such as wind resistance and mechanical friction), J is the moment of inertia of the pan-tilt rotating part (unit: kg·m²), and K is the system transmission ratio constant;

[0037] When it is detected that Compensate torque :

[0038] ;

[0039] wherein, is a preset expected angular rate of change, is an actually measured angular rate of change;

[0040] Apply the torque to be compensated by the brushless motor set , and adjust the PWM duty cycle of the motor winding in real time:

[0041] ;

[0042] wherein D is the PWM duty cycle of the brushless motor winding, K t is the motor torque constant, I max is the maximum allowable operating current of the motor.

[0043] Further, the specific process of the abnormal state grading response mechanism in S5 is:

[0044] Define stability margin M:

[0045] ;

[0046] wherein S current is the real-time stability score, S max is the maximum score;

[0047] When 30%≤M<50%, execute the first level compensation strategy: increase the proportional gain K p of the PID controller to 1.5 times, and control the auxiliary balance wheel speed ω:

[0048] ;

[0049] wherein, is the speed coefficient (rpm / √rad), is the angular deviation (rad), is the maximum allowable deviation (rad), K p is the initial proportional gain of the PID controller, whose value is determined by system dynamic characteristic calibration or experimental optimization;

[0050] When M<30%, execute the second level compensation strategy, switch to the hydraulic locking mode, control the hydraulic cylinder pressure P, and the process of controlling the hydraulic cylinder pressure P is:

[0051] ;

[0052] wherein P is the locking pressure of the hydraulic cylinder (unit: Pa), is the required locking force (unit: N), is the cross-sectional area of the hydraulic cylinder piston (unit: m²), t is the duration of the locking action (unit: seconds), is the time constant.

[0053] Further, the method further comprises constructing a dynamic deformation coupling digital twin system, and the specific process is as follows:

[0054] Real-time acquisition of strain components of the support structure by distributed optical fiber sensors Construction of a three-dimensional deformation gradient tensor ;

[0055] Solving deformation-stability coupling equations:

[0056] ;

[0057] In the formula, is the initial stability score, S is the real-time stability score, is the deformation sensitivity coefficient, is the trace of the deformation gradient tensor (representing the volumetric deformation rate), is the torque compensation weight;

[0058] When it is predicted that S(t+Δt)<S threshold , multi-modal compensation is performed in advance, that is, the piezoelectric ceramic array is started to apply a reverse deformation field:

[0059] ;

[0060] Wherein, is the piezoelectric compensation gain;

[0061] Synchronous adjustment of the of the brushless motor is: , wherein is the torque coupling coefficient.

[0062] Further, the method further comprises a cloud-edge collaborative deformation compensation network, and the specific process is as follows:

[0063] The deformation gradient tensor collected by the distributed optical fiber sensor is packaged into an encrypted data packet, and the encryption algorithm is:

[0064] ;

[0065] is the encrypted data packet (binary data stream), ​A dynamically generated 256-bit encryption key (hexadecimal string), is the deformation gradient tensor, and || is the data concatenation symbol.

[0066] Solve the optimization problem in the cloud:

[0067] ;

[0068] In the formula, λ and μ are weight coefficients for balancing the optimization objectives of deformation and torque compensation, and are the optimized piezoelectric compensation gains and torque coupling coefficients.

[0069] The optimized β and γ are returned to the edge node to control the piezoelectric ceramic and brushless motor to perform compound compensation.

[0070] Compared with the prior art, the stability regulation method of the camera system has the following advantages: the terrain, vibration and environmental parameters are collected in real time by the multi-source sensor array, the three-dimensional stability evaluation model is combined to realize all-around environmental adaptability adjustment, the stability of the camera system in complex scenes is significantly improved, the finite element analysis, frequency domain feature extraction and convolutional neural network technology are used to realize high-precision stability scoring and abnormal prediction, the system can respond to potential instability risks in time, the self-balancing adjustment, damping optimization and reverse torque compensation are triggered by real-time monitoring of the angle change rate and vibration characteristics, the environmental interference (such as wind resistance and mechanical vibration) is effectively inhibited, the gimbal movement smoothness is ensured, the multi-level warning thresholds are divided based on the stability margin, the PID gain optimization, hydraulic locking and other multi-modal compensation strategies are adopted to realize progressive safety protection from warning to emergency locking, reduce the risk of system out-of-control, the physical deformation data is mapped in real time by the digital twin system, the cloud optimization algorithm and edge execution are combined to realize deformation-torque coupling compensation, enhance the anti-deformation ability and remote control efficiency of the system, integrate encrypted data transmission, multi-modal sensing and intelligent algorithms, break through the traditional single protection mode, provide full-link closed-loop control from data acquisition, analysis to execution, improve the system reliability and maintenance convenience, thereby realizing high stability, fast response and intelligent management of the camera system in complex environments, significantly better than the protection ability and adaptability of the prior art, making the system more worthy of promotion and use. BRIEF DESCRIPTION OF DRAWINGS

[0071] Figure 1 is the flowchart of the present application. DETAILED DESCRIPTION

[0072] The embodiments of the present application will be described in detail below. The embodiments are implemented on the premise of the technical solutions of the present application, and detailed implementation modes and specific operation processes are given, but the protection scope of the present application is not limited to the following embodiments.

[0073] As Figure 1 shown, the embodiment provides a technical solution: a stability regulation method of a camera system, comprising the following steps:

[0074] S1: Real-time acquisition of installation position information of the camera system through a multi-source sensor array, the installation position information including terrain elevation data, base vibration frequency spectrum and environmental temperature and humidity distribution parameters;

[0075] S2: Construct a three-dimensional stability evaluation model, input the installation position information into the three-dimensional stability evaluation model to generate a stability coefficient matrix; when any dimension parameter in the stability coefficient matrix exceeds a preset safety threshold, activate a self-balancing adjustment mechanism for multi-degree-of-freedom compensation adjustment;

[0076] S3: The camera system performs follow-up monitoring, and in the follow-up monitoring process, the dynamic frequency analysis module is used to real-time acquisition of frequency domain characteristics of the gimbal motion trajectory, a mapping relationship between motion characteristics and stability is established, and an adaptive damping adjustment instruction is generated based on the mapping relationship between motion characteristics and stability;

[0077] S4: Real-time monitoring of the dynamic change rate of the gimbal rotation angle of the camera system by using an inertial measurement unit, triggering a pre-tightening force compensation mechanism and generating a reverse torque compensation parameter when detecting that the angle change rate exceeds a preset critical curve;

[0078] S5: Establish an abnormal state grading response mechanism, start a first compensation strategy when the system stability margin is lower than a first warning threshold, switch to a safety locking mode and generate a fault positioning report when it is lower than a second emergency threshold.

[0079] The construction process of the three-dimensional stability evaluation model in S2 is as follows:

[0080] Terrain adaptation degree calculation, stress distribution under different terrain conditions is calculated through finite element analysis, and the terrain adaptation factor is defined as:

[0081] ;

[0082] α is the terrain adaptation factor, which is used to quantify the adaptation degree of the installation position to the stability of the camera system (0≤α≤1);

[0083] is the material yield strength of the camera system support structure;

[0084] is the actual stress value at the i-th finite element grid node (unit: MPa);

[0085] N is the total number of nodes of the finite element grid division;

[0086] The transfer function method is used to establish a correlation model of base vibration and gimbal shaking, and a vibration transmission gain is defined as:

[0087] ;

[0088] wherein, is a vibration transmission gain (unit: dB), representing an amplification effect of base vibration transmission to the gimbal;

[0089] is a vibration power spectral density of the gimbal, is a vibration power spectral density of the base;

[0090] When the adaptation factor a of any dimension is less than 0.8 or the vibration gain Gv is greater than 3 dB, a self-balancing adjustment mechanism is triggered.

[0091] The terrain stress distribution is calculated through finite element analysis, and a terrain adaptation factor a is defined to quantify the adaptation degree of different terrains to the stability of the camera system. The risk of overloading or tilting of the support structure caused by uneven terrain or soft geology can be avoided, and the optimal selection of the system installation position can be ensured.

[0092] When installing the camera system in a complex terrain in the field, the system calculates the a value in real time. If it is detected that a < 0.8 (such as stress concentration in soft soil area), it automatically prompts to adjust the installation point or activate the base reinforcement module to prevent structural deformation.

[0093] The transfer function method is used to establish a correlation model of base vibration and gimbal shaking, and a vibration transmission gain Gv is quantified to predict the amplification effect of vibration on the stability of the gimbal. The interference of external vibration (such as mechanical vibration, vehicle passing) on the camera picture can be significantly reduced.

[0094] In the bridge monitoring scene, when the base vibration is caused by vehicle passing, the system detects that > 3 dB (vibration transmission is too strong), and immediately triggers the self-balancing mechanism for multi-degree-of-freedom compensation to offset the vibration energy and ensure the stability of the gimbal picture.

[0095] Through the double mechanisms of terrain adaptation and vibration prediction, installation risks are avoided from the source and external interference is dynamically suppressed, significantly improving the reliability and anti-interference ability of the camera system in complex environments.

[0096] The analysis process of the dynamic frequency analysis module is as follows:

[0097] Three-axis vibration data is collected by the acceleration sensor group at a sampling rate of 1000 Hz;

[0098] Three-axis vibration acceleration signals a(t) are collected at a sampling rate of 1000 Hz, and a frequency domain feature vector F is constructed through Fourier transform:​

[0099] ;

[0100] wherein: is a frequency domain feature vector (complex vector) representing the energy distribution of the vibration signal at different frequencies, is the sampling window length (unit: seconds), is the time domain vibration acceleration signal (unit: m / s²), is the frequency value of the kth frequency point (unit: Hz), j is the imaginary unit, and satisfies ;

[0101] F is input into a pre-trained convolutional neural network (CNN), and a stability score S is output :

[0102] ;

[0103] wherein, is the stability score, and the higher the score, the better the stability, is the ith frequency domain feature vector;

[0104] wi is the ith weight parameter in the convolutional neural network, and b is the bias term of the neural network;

[0105] ReLU is a rectified linear unit activation function, defined as ReLU(x) = max(0, x);

[0106] When <S threshold , an adaptive damping adjustment instruction is generated, and S threshold is a preset stability score threshold;

[0107] Three-axis vibration data is collected by an acceleration sensor at a high sampling rate of 1000 Hz, and frequency domain features are extracted by Fourier transform to comprehensively analyze the vibration energy distribution and avoid the missed detection problem of traditional time domain analysis.

[0108] For example, in a highway monitoring scene, a heavy truck passing by causes high-frequency vibration (such as instantaneous impact above 200 Hz), and the system quickly identifies the abnormal frequency band energy peak value through the frequency domain feature vector F, accurately locates the vibration source, and prevents the gimbal from causing blurred images due to high-frequency shaking.

[0109] The pre-trained convolutional neural network (CNN) is used to model the frequency domain features non-linearly, output a dynamic stability score, and trigger adaptive damping adjustment based on a threshold, realizing closed-loop control from perception to compensation.

[0110] For example, in typhoon weather, the camera system's gimbal vibrates at low frequencies (e.g., 5-10 Hz) due to strong wind interference. The CNN model determines the stability score by analyzing the frequency domain features. < threshold, a damping adjustment command is immediately generated to increase the gimbal damping coefficient, suppress low-frequency swings, and ensure image stability.

[0111] Through high-frequency vibration analysis and intelligent scoring mechanisms, rapid identification and targeted compensation of complex vibration patterns are achieved, significantly improving the camera system's active anti-vibration capability and image stability in dynamic interference environments.

[0112] The specific process of S4 is:

[0113] Establish the differential equation for the rate of change of angle: ;

[0114] Where, is the rate of change of the pan-tilt rotation angle, θ is the pan-tilt rotation angle, is the torque output by the driving motor, is the equivalent torque caused by environmental interference (such as wind resistance and mechanical friction), J is the moment of inertia of the gimbal's rotating components (unit: kg·m²), and K is the system transmission ratio constant;

[0115] When detected Calculate the compensation torque when :

[0116] ;

[0117] Where, is the preset expected angle change rate, is the actual measured angle change rate;

[0118] is the preset desired angle, is the actual measured angle;

[0119] Apply the required torque via the brushless motor , and adjust the PWM duty cycle of the motor winding in real time:

[0120] ;

[0121] Where D is the PWM duty cycle of the brushless motor winding, K t is the motor torque constant, I max is the maximum allowable operating current of the motor;

[0122] By establishing the angle change rate differential equation, the dynamic change rate of the rotation angle of the fixed camera system pan-tilt is monitored in real time, and the equivalent calculation of the environmental interference torque (such as strong wind, mechanical friction) is combined to realize the rapid detection and compensation of the angle deviation. It can effectively avoid the pan-tilt shaking or picture deviation caused by sudden interference (such as strong wind, equipment vibration).

[0123] As in the outdoor monitoring tracking scene in the coastal area, the strong wind causes the rotation angle change rate of the fixed pan-tilt to suddenly increase, the system immediately calculates the reverse compensation torque, and applies the reverse torque through the brushless motor to make the pan-tilt restore the preset angle within 0.2 seconds, ensuring the stability of the monitoring picture without trailing.

[0124] Through the dynamic adjustment of the PWM duty cycle of the brushless motor group, the compensation torque is accurately converted into the motor driving signal, realizing the compensation action with high response speed and low delay, and ensuring the smoothness of the pan-tilt in the fixed scene.

[0125] In the city traffic intersection monitoring, the ground vibration caused by the passing of heavy trucks causes the pan-tilt angle change rate to be abnormal, the system generates a PWM duty cycle command according to the compensation formula, and the motor quickly outputs a reverse torque to offset the angle deviation caused by vibration, keeping the picture stable and clear.

[0126] Through the triple mechanism of dynamic monitoring, accurate modeling and efficient execution, the rapid identification and active compensation of the rotation abnormality of the fixed camera system pan-tilt are realized, which significantly improves its anti-deviation ability and control accuracy in the interference environment such as strong wind and vibration, and ensures the continuity and reliability of image acquisition.

[0127] The specific process of the abnormal state grading response mechanism in S5 is as follows:

[0128] Define the stability margin M:

[0129] ;

[0130] Where S current is the real-time stability score, S max is the maximum score;

[0131] When 30%≤M<50%, execute the first compensation strategy: increase the proportional gain K p of the PID controller by 1.5 times, and control the auxiliary balance wheel speed ω:

[0132] ;

[0133] In the formula, is the speed coefficient (rpm / √rad), is the angle deviation (rad), is the maximum angle allowed deviation (rad), Kp The initial proportional gain of the PID controller, the value of which is determined by system dynamic characteristics calibration or experimental optimization;

[0134] When M < 30%, a secondary compensation strategy is executed, switching to a hydraulic locking mode, controlling the hydraulic cylinder pressure P, and the process of controlling the hydraulic cylinder pressure P is:

[0135]

[0136] is the required locking force (unit: N), is the cross-sectional area of the hydraulic cylinder piston (unit: m²), and t is the duration of the locking action (unit: seconds), is the time constant;

[0137] Hierarchical response enhances system resilience: By defining stability margin M and dividing multiple levels of early warning thresholds (first-level compensation when 30%≤M<50%, and second-level compensation when M<30%), a progressive protection from early warning to emergency locking is achieved, avoiding the insufficient response or excessive intervention of a single strategy, and enhancing the fault tolerance of the system under continuous abnormal conditions.

[0138] As in the fixed monitoring scene, device vibration causes the stability margin M to drop to 40%, the system automatically increases the PID proportional gain and accelerates the balance wheel rotation, quickly suppressing the angle deviation caused by vibration; if the vibration further intensifies and M drops to 25%, the system immediately switches to the hydraulic locking mode, fixing the PTZ through the hydraulic cylinder pressure, preventing equipment damage.

[0139] Dynamic parameter optimization and emergency locking: The first-level compensation strategy actively corrects the deviation by adjusting the PID gain and balance wheel speed, and the second-level compensation strategy ensures reliable locking in emergency situations through a hydraulic locking pressure curve (such as an exponential progressive pressure), balancing fast response and equipment protection.

[0140] In the security monitoring of areas with frequent abnormal vibrations, initial vibration causes the PTZ to sway (M=35%), and the system automatically increases the PID control gain, with the balance wheel rotating at an adaptive speed to offset the sway; if the vibration continues to escalate (M=20%), the hydraulic cylinder gradually locks the PTZ according to the preset pressure curve, preventing damage to the mechanical structure due to instantaneous impact.

[0141] Through the hierarchical response mechanism and multi-modal compensation strategy, full-link protection from active correction to emergency locking is achieved, significantly improving the stability and safety of the fixed camera system under continuous or sudden abnormal conditions, ensuring the continuity of image acquisition and the reliability of equipment in critical scenarios.

[0142] The method further includes constructing a digital twin system coupled with dynamic deformation, and the specific process is as follows: ​

[0143] Real-time acquisition of strain components of the support structure by distributed optical fiber sensors , construct the three-dimensional deformation gradient tensor

[0144] Solve the deformation-stability coupled equation:

[0145] ;

[0146] In the formula, is the initial stability score, S is the real-time stability score, is the deformation sensitivity coefficient, is the trace of the deformation gradient tensor (representing the volumetric deformation rate), is the torque compensation weight;

[0147] When it is predicted that S(t+Δt)<S threshold , multi-modal compensation is performed in advance, that is, the piezoelectric ceramic array is started to apply a reverse deformation field:

[0148] ;

[0149] Where, is the piezoelectric compensation gain;

[0150] Synchronous adjustment of the of the brushless motor is: , where is the torque coupling coefficient;

[0151] Deformation-stability coupling prediction and active compensation: real-time monitoring of three-dimensional strain of the support structure by distributed optical fiber sensors, combined with digital twin system to predict the influence of deformation on stability, and piezoelectric ceramic is started in advance to apply a reverse deformation field to avoid system instability caused by sudden deformation.

[0152] As in the fixed monitoring scene of large bridges, the bridge body undergoes slight deformation (such as support structure stretching) due to day-night temperature difference, the system predicts through digital twin that the stability score S will be lower than the threshold value within the next 10 minutes, immediately controls the piezoelectric ceramic array to generate a reverse deformation field, offsets the thermal expansion and contraction effect, and ensures that the gimbal angle does not deviate.

[0153] Dynamic coupling of deformation compensation (piezoelectric ceramic) and torque compensation (brushless motor), through parameter optimization to realize the collaborative suppression of deformation and torque, and improve the comprehensive anti-interference performance of the system in complex mechanical environment.

[0154] In the vibration monitoring scene of industrial plants, local deformation and vibration of the support structure are superimposed by the operation of heavy equipment, and the system synchronously adjusts the deformation compensation amount of the piezoelectric ceramic and the compensation torque of the brushless motor, eliminate the influence of composite interference on the stability of the gimbal, and ensure that the high-definition picture is not distorted.

[0155] Through the deformation prediction and multi-modal compensation linkage mechanism of the digital twin system, the deformation of the support structure is intervened and dynamically suppressed in advance, significantly improving the anti-deformation ability and stability of the fixed camera system in long-term load, temperature difference change or composite interference environment, and avoiding image distortion or equipment damage caused by structural deformation.

[0156] The method also includes a cloud-edge collaborative deformation compensation network, and the specific process is:

[0157] The deformation gradient tensor collected by the distributed optical fiber sensor and is packaged as an encrypted data packet, and the encryption algorithm is:

[0158] ;

[0159] is the encrypted data packet (binary data stream), is a dynamically generated 256-bit encryption key (hexadecimal string), is the deformation gradient tensor, and || is the data concatenation symbol.

[0160] The cloud solves the optimization problem:

[0161] ;

[0162] where λ and μ are weight coefficients used to balance the optimization objectives of deformation and torque compensation, and are the optimized piezoelectric compensation gains and torque coupling coefficients;

[0163] The optimized β and γ are returned to the edge node to control the piezoelectric ceramic and brushless motor to perform composite compensation.

[0164] The deformation gradient tensor and the compensation torque are dynamically encrypted by the AES256 encryption algorithm to ensure that sensitive data is tamper-proof and theft-proof during cloud-edge transmission, ensuring the security of the system data chain.

[0165] In the monitoring scene of a transportation hub, for example, if the system detects abnormal support structure deformation data, it will upload the encrypted data packet to the cloud after generating a 256-bit key. Even if the network is attacked, the encrypted data cannot be cracked, and malicious tampering of compensation parameters to cause the gimbal to lose control is avoided.

[0166] The cloud solves the global optimization problem based on the weight coefficients of deformation and torque, and returns the optimal compensation parameters to the edge node to achieve global optimal collaboration of deformation and torque compensation, breaking through the limitations of single-node computing power.

[0167] Example: In security monitoring, multiple camera systems detect that the support structure deformation is coupled with vibration interference. The cloud centrally analyzes all nodes' ∇ε and data, optimizes β and γ parameters, and then issues them to each edge node to ensure consistency and efficiency of the overall area compensation strategy.

[0168] Through the edge node, real-time piezoelectric ceramic compensation and brushless motor torque compensation optimized by the cloud are performed to achieve fast response and precise cancellation under complex interference.

[0169] For example, in the process of monitoring and tracking near a substation, sudden electromagnetic interference causes high-frequency deformation of the support structure. Based on the cloud-optimized β and γ parameters, the edge node synchronously controls the piezoelectric ceramic array to generate a reverse deformation field and adjusts the output compensation torque of the brushless motor. The interference is eliminated within 0.5 seconds, ensuring the stability of the equipment monitoring picture.

[0170] Through the collaborative mechanism of encrypted transmission, cloud global optimization, and edge rapid execution, data security is ensured, and dynamic coordination and precise execution of multi-node compensation strategies are achieved, significantly improving the anti-deformation ability and overall stability of fixed camera systems in complex and multi-interference scenarios, while reducing local computing resource consumption.

[0171] In addition, the terms "first", "second", "third", etc. are used only for descriptive purposes and should not be construed as indicating or implying relative importance or an indicated number of technical features. Therefore, the features defined as "first", "second", etc. can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, such as two, three, etc., unless otherwise specifically limited.

[0172] In the description of the present application, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present description, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the skilled in the art can combine and combine the different embodiments or examples described in the present description and the features of the different embodiments or examples without contradiction.

[0173] Although the embodiments of the present application have been shown and described above, it is understood that the above-described embodiments are exemplary and are not to be construed as limiting the present application, and that changes, modifications, substitutions and variations can be made by those skilled in the art without departing from the scope of the present application.

Claims

1. A method for controlling the stability of a camera system, characterized in that: The following steps are involved: S1: Using a multi-source sensor array to collect real-time installation location information of the camera system, the installation location information includes terrain elevation data, base vibration frequency spectrum, and ambient temperature and humidity distribution parameters; S2: Construct a three-dimensional stability assessment model and input the installation position information into the three-dimensional stability assessment model to generate a stability coefficient matrix. When any dimension parameter in the stability coefficient matrix exceeds a preset safety threshold, activate the self-balancing adjustment mechanism to perform multi-degree-of-freedom compensation adjustment. S3: The camera system performs follow-up monitoring. During the follow-up monitoring process, the dynamic frequency analysis module collects the frequency domain characteristics of the pan / tilt motion trajectory in real time, establishes a mapping relationship between motion characteristics and stability, and generates adaptive damping adjustment instructions based on the mapping relationship between motion characteristics and stability; S4: An inertial measurement unit is used to monitor the dynamic rate of change of the camera system's pan / tilt rotation angle in real time. When the angle change rate is detected to exceed a preset critical curve, the preload compensation mechanism is triggered and the reverse torque compensation parameters are generated. S5: Establish a hierarchical response mechanism for abnormal conditions. When the system stability margin falls below the first warning threshold, the first-level compensation strategy is activated. When it falls below the second emergency threshold, the system switches to a safety lock mode and generates a fault location report. The construction process of the three-dimensional stability evaluation model in S2 is as follows: Perform terrain adaptability calculations, calculate stress distribution under different terrain conditions through finite element analysis, and define terrain adaptability factors , terrain adaptation factor Specifically: ； in, Yield strength of the material supporting the camera system; is the actual stress value at the i-th finite element mesh node; N is the total number of nodes in the finite element mesh; Predict vibration response, use transfer function method to establish the correlation model between base vibration and gimbal shaking, and define vibration transfer gain , specifically: ； Where, is the vibration power spectrum density of the gimbal, is the vibration power spectrum density of the base; When the adaptation factor of any dimension <0.8 or vibration gain When it is >3dB, the self-balancing adjustment mechanism is triggered; Collect triaxial vibration data at a preset sampling rate through an acceleration sensor group; The triaxial vibration acceleration signal a(t) is collected at a preset sampling rate, and the frequency domain feature vector F is constructed through Fourier transform; Input F into the pre-trained convolutional neural network and output the stability score ; when <S threshold When the adaptive damping adjustment instruction is generated, S threshold The corresponding threshold value for the stability score.

2. The method for controlling the stability of a camera system according to claim 1, wherein: The specific process of S4 is: Establish the differential equation for the rate of change of angle: ； Where, is the rate of change of the pan-tilt rotation angle, θ is the pan-tilt rotation angle, is the torque output by the driving motor, is the equivalent torque caused by environmental interference, J is the moment of inertia of the gimbal rotating parts, and K is the system transmission ratio constant; When detected Calculate the compensation torque when : ； Where, is the preset expected angle change rate, is the actual measured angle change rate; Apply the required torque via the brushless motor , and adjust the PWM duty cycle D of the motor winding in real time.

3. The method for controlling the stability of a camera system according to claim 1, wherein: The specific process of the abnormal state hierarchical response mechanism in S5 is as follows: Define the stability margin M; When 30%≤M<50%, execute the first-level compensation strategy: increase the proportional gain K of the PID controller p Increase to 1.5 times and control the auxiliary balance wheel speed ω; When M is less than 30%, the secondary compensation strategy is executed, switching to the hydraulic locking mode to control the hydraulic cylinder pressure P.

4. The method for controlling the stability of a camera system according to claim 2, wherein: The method also includes constructing a dynamic deformation coupled digital twin system, the specific process of which is as follows: Real-time acquisition of the strain components of the supporting structure through distributed optical fiber sensors , constructing a three-dimensional deformation gradient tensor ; Solve the deformation-stability coupled equations: ； S is the real-time stability score, is the deformation sensitivity coefficient, is the trace of the deformation gradient tensor, is the torque compensation weight; When it is predicted that S(t+Δt) within the future Δt time threshold , perform multimodal compensation in advance, that is, start the piezoelectric ceramic array to apply the reverse deformation field:​ ； in, is the piezoelectric compensation gain; Synchronous adjustment of brushless motor for: ,in is the torque coupling coefficient.

5. The method for controlling the stability of a camera system according to claim 4, wherein: The method also includes a cloud-edge collaborative deformation compensation network, the specific process of which is as follows: The deformation gradient tensor collected by the distributed optical fiber sensor and The data is encapsulated into an encrypted data packet and then optimized in the cloud. After the optimization is completed, the optimized β and γ are transmitted back to the edge node to control the piezoelectric ceramics and brushless motor to perform composite compensation.

Images