Surgical operation box based on dynamic area adjustment and fluid self-adaptive control method thereof

Abstract

The invention discloses a dynamic area adjustment-based surgical operation box and a fluid self-adaptive control method thereof.The dynamic area adjustment-based surgical operation box comprises a box body, a fluid circulating system and a control system, the box body comprises at least two sets of wall plates capable of relatively moving, and the wall plates are connected through a telescopic guide mechanism to form a variable-volume cavity; an area adjustment driving assembly is arranged between the wall plates and comprises a linear actuator and a displacement sensor; the fluid circulation system comprises a fluid pump communicated with the variable volume cavity, a pressure regulating valve and a multi-parameter sensor group; the multi-parameter sensor group comprises a pressure sensor, a flow sensor and a temperature sensor; the control system comprises a main control module and a self-adaptive algorithm module, the main control module is connected with the area adjustment driving assembly and the fluid circulation system, and the self-adaptive algorithm module is configured to establish a dynamic balance model, output a cooperative control signal through a resolving model and preferentially trigger a wallboard displacement compensation mechanism when it is detected that pressure fluctuation exceeds a threshold value.

Description

Technical Field

[0001] The present invention relates to the technical field of medical surgical instruments, and in particular to a surgical operating box based on dynamic area regulation and a fluid adaptive control method thereof. Background Art

[0002] During surgical procedures, the stability and dynamic response of surgical chambers (such as the inflatable chambers used in laparoscopic surgery) directly impact surgical safety and operational efficiency. Traditional surgical chambers often use a fixed-volume design and rely on a single fluid pump speed adjustment to maintain pressure, which presents the following issues: (1) Delayed response to pressure fluctuations: When surgical instruments are inserted or tissue is displaced, the traditional cavity will experience sudden pressure changes due to its fixed volume. It is difficult to quickly compensate by adjusting the fluid pump alone, which can easily cause tissue damage or gas embolism risks.

[0003] (2) Insufficient multi-parameter coupling control: Existing technologies lack dynamic modeling for the coordinated control of parameters such as temperature, flow, and pressure, and ignore the nonlinear relationship between fluid viscosity and cavity deformation, resulting in insufficient control accuracy.

[0004] (3) Poor robustness in extreme scenarios: When a sudden pressure shock occurs (such as a blood vessel rupture), traditional methods lack a reverse volume compensation mechanism and are unable to quickly suppress abnormal pressure fluctuations.

[0005] (4) Insufficient personalized adaptability: The anatomical structures and tissue mechanical properties of different patients vary significantly. The existing system cannot adapt to dynamic deformation constraints in real time, which can easily lead to excessive deformation or tissue damage.

[0006] Existing technologies propose adjusting the fluid pump speed through PID algorithms, but this fails to address the dynamic coupling between cavity volume and fluid parameters. Existing technologies also propose using pre-programmed pressure curve control, but this fails to integrate real-time intraoperative data and is unsuitable for complex surgical scenarios. Therefore, a surgical operating box and control method that can dynamically adjust cavity volume, coordinate multi-parameter control, and provide emergency compensation capabilities is urgently needed. Summary of the Invention

[0007] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a surgical operating box based on dynamic area adjustment and a fluid adaptive control method thereof, which is used to solve the technical problems in the existing technology of fixed volume of the surgical box, lag in pressure regulation, and insufficient multi-parameter coordinated control.

[0008] In order to solve the above problems, the technical solutions adopted by the present invention are as follows: A surgical operating box based on dynamic area adjustment comprises a box body, a fluid circulation system and a control system. The box body comprises at least two sets of relatively movable wall panels, which are connected by a telescopic guide mechanism to form a variable volume cavity. An area adjustment drive assembly is provided between the wall panels, and the area adjustment drive assembly includes a linear actuator and a displacement sensor; The fluid circulation system comprises a fluid pump, a pressure regulating valve and a multi-parameter sensor group connected to the variable volume cavity, wherein the multi-parameter sensor group comprises: a pressure sensor, a flow sensor and a temperature sensor; The control system includes a main control module and an adaptive algorithm module. The main control module is connected to the area adjustment drive assembly and the fluid circulation system. The adaptive algorithm module is configured as follows: Establish a dynamic balance model based on real-time monitoring data from a multi-parameter sensor group; Outputting a coordinated control signal through a solution model to synchronously adjust the displacement of the linear actuator and the speed of the fluid pump; When the pressure fluctuation is detected to exceed the threshold, the wall plate displacement compensation mechanism is triggered first.

[0009] A method for adaptive fluid control of a surgical box based on dynamic area regulation comprises the following steps: Real-time acquisition of monitoring data from a multi-parameter sensor group, including cavity internal pressure, fluid flow, and temperature parameters; Establishing a dynamic balance model based on the monitoring data, the dynamic balance model includes a nonlinear mapping relationship between cavity volume change and fluid pressure, and training model parameters through historical data; Solving the dynamic balance model to generate coordinated control instructions includes: predicting a target displacement of the linear actuator based on a current pressure deviation, and dynamically calculating an optimized speed of the fluid pump based on flow demand and temperature parameters; The linear actuator's displacement compensation and the fluid pump's speed regulation are performed synchronously. When the pressure fluctuation exceeds the threshold, the wall plate displacement compensation mechanism is triggered first. The cavity volume is adjusted through the inverse solution model to quickly suppress the pressure fluctuation. After the compensation phase, multi-parameter feedback closed-loop control is started to iteratively optimize the model parameters based on the real-time updated sensor data.

[0010] Preferably, when establishing a dynamic equilibrium model based on the monitoring data, the method includes: During the surgical preparation phase, a calibration fluid is injected, and the wall plate is driven by a linear actuator to perform multi-band reciprocating motion. The pressure-displacement response data is collected to construct an initial parameter matrix. A nonlinear partial differential equation is established, which includes a quadratic term representing the displacement acceleration of the wall panel and an exponential decay term representing the viscous resistance of the fluid. Using the real-time feedback flow pulsation spectrum data to perform online parameter identification on the exponential decay term using a variable weight factor; The nonlinear mapping relationship is decomposed into a volume change dominant term and a fluid disturbance compensation term, and each term is solved by alternate iterative methods.

[0011] Preferably, when training model parameters using historical data, the following steps are included: Construct a multimodal historical dataset containing calibrated motion data, sudden pressure shock patterns, and corresponding instrument motion trajectory feature vectors for different types of surgical scenarios; An improved incremental Bayesian optimization algorithm is used to perform parameter training through adaptive adjustment; The quantum annealing algorithm is used to optimize the constrained loss function, where the dynamic weight factor It is positively correlated with the intensity of instrument activity during the current surgical phase; A parameter drift compensation mechanism is established to trigger abnormal parameter reset based on the adversarial generative network.

[0012] Preferably, when the improved incremental Bayesian optimization algorithm is used for parameter training, the following steps are included: In the offline stage, a global sensitivity analysis of the nonlinear partial differential equation is performed using a Kriging proxy model to screen out a set of key trainable parameters; In the online stage, the dynamic similarity evaluation between real-time surgical data stream and historical data is used to adaptively adjust the training window length.

[0013] Preferably, predicting the target displacement of the linear actuator includes: A pressure-displacement time series prediction model is constructed, whose input is the real-time pressure deviation sequence within the sliding time window, and the output is the displacement increment in the next three control cycles; Use dual-channel convolutional long short-term memory network to process multi-scale features; Based on the real-time pressure deviation sequence, the absolute value of the current pressure deviation is generated, and a dynamic compensation factor is obtained; Based on the dynamic compensation factor, when the absolute value of the current pressure deviation is greater than the set value, the emergency compensation protocol is activated: Reconstruct the cavity deformation constraint boundary based on preoperative CT / MRI images and generate a feasible displacement domain that meets the preset conditions; Combined with intraoperative ultrasound Doppler blood flow velocity field data, the actuator acceleration curve is inversely optimized through the adjoint equation.

[0014] Preferably, the dynamic calculation of the optimized speed of the fluid pump includes: Construct a multi-objective optimization function of flow rate, temperature and speed; Establishing three-dimensional pulse spectrum mapping relationship; Design an anti-saturation dynamic limiting strategy: When the absolute value of the difference between the real-time temperature monitoring value and the biological tissue tolerance temperature threshold is detected to be greater than the set value, the thermal protection constraint is activated; When the prothrombin time is abnormal, a flow rate change constraint is imposed; Nonlinear model predictive control was implemented, in which a thrombosis risk index was embedded as a soft constraint, and the blood flow impedance boundary condition was updated in real time in combination with intraoperative OCT angiography data.

[0015] Preferably, when adjusting the cavity volume by inverse solution of the model, the method includes: Establish the pressure fluctuation inverse mapping equation; Real-time generation of adversarial compensation signals: The ultrasonic blood flow vector field is input into the pre-trained 3D Flow-GAN network to generate a virtual displacement compensation that matches the current anatomical structure. The actual pressure gradient data and the virtual displacement compensation are then fused using the Lyapunov stability criterion. Deploy a distributed real-time solution architecture; Dynamic reconstruction of the wall panel motion trajectory: When it is detected that the current pressure deviation value is greater than the set value, the emergency reverse solution mode is activated.

[0016] Preferably, the iterative optimization of the model parameters includes: Construct a multi-rate fusion feedback matrix: sample multiple source data streams at different frequencies and fuse them through federated Kalman filtering; Design a two-layer nested optimization engine, including: the inner layer performs online sparse dictionary learning, and the outer layer runs an adversarial reinforcement learning strategy.

[0017] Preferably, when iteratively optimizing the model parameters, the following steps are further included: Implementing anatomical-rheological coupling constraints: Converting intraoperative ultrasound elastography data into a deformation Jacobian matrix , impose constraints , is the dynamic displacement gradient of the wall panel; is the deformation safety threshold, is the Frobenius norm; Automatically shrink the feasible displacement domain when fibrotic tissue areas are detected: , is the displacement safety threshold.

[0018] Compared with the prior art, the present invention has the following beneficial effects: (1) Dynamic volume adjustment improves response speed: By driving the wall plate displacement through a linear actuator and combining it with the fluid pump for coordinated control, the pressure fluctuation suppression time is greatly shortened compared to traditional methods; (2) High precision of multi-sensor collaborative control: Based on the nonlinear partial differential equation, a dynamic balance model is constructed, integrating multi-parameter feedback such as pressure, flow, and temperature to achieve multi-objective optimization control, and the control error is significantly reduced; (3) Emergency compensation mechanism ensures surgical safety: When the pressure fluctuation exceeds the threshold, the wall plate displacement compensation mechanism is triggered first. The inverse solution model and 3D Flow-GAN network are used to generate virtual displacement compensation, which can attenuate the abnormal pressure fluctuation to within the safety threshold in a short time. (4) Personalized deformation constraint to avoid tissue damage: The cavity deformation boundary is reconstructed based on preoperative CT / MRI images, and the displacement safety threshold is dynamically adjusted in combination with intraoperative ultrasound elastic imaging data to control the tissue deformation within the physiological tolerance range; (5) Intelligent learning optimization to adapt to complex scenarios: By optimizing the loss function and adversarial reinforcement learning strategy (ARL) through the quantum annealing algorithm, the system can automatically adapt parameters in different types of surgeries, greatly improving the efficiency of model training; (6) Modular design facilitates clinical application: Distributed real-time solution architecture supports edge computing nodes to process high-resolution data (sampling rate ≥ 1kHz), ensuring stable operation of the system in complex surgical environments.

[0019] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a step diagram of a method for adaptively controlling fluid in a surgical operating box based on dynamic area adjustment according to an embodiment of the present invention; Figure 2 This is a flow chart of establishing a dynamic balance model based on monitoring data according to an embodiment of the present invention; Figure 3 This is a flow chart of adjusting the cavity volume by inverse solution of the model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0022] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0023] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0024] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0025] Example 1: A surgical operating box based on dynamic area adjustment provided by the present invention includes a box body, a fluid circulation system and a control system.

[0026] The box body comprises at least two sets of relatively movable wall panels, which are connected by a telescopic guide mechanism to form a variable volume cavity; An area adjustment drive assembly is provided between the wall panels. The area adjustment drive assembly comprises a linear actuator and a displacement sensor, and is capable of changing the effective working volume of the variable volume cavity in real time.

[0027] The fluid circulation system includes a fluid pump connected to the variable volume cavity, a pressure regulating valve and a multi-parameter sensor group, the multi-parameter sensor group includes: a pressure sensor, a flow sensor and a temperature sensor; The control system includes a main control module and an adaptive algorithm module. The main control module connects the area adjustment drive component and the fluid circulation system. The adaptive algorithm module is configured as follows: Establish a dynamic balance model based on real-time monitoring data from a multi-parameter sensor group; Outputting a coordinated control signal through the solution model to synchronously adjust the displacement of the linear actuator and the speed of the fluid pump; When the pressure fluctuation is detected to exceed the threshold, the wall plate displacement compensation mechanism is triggered first.

[0028] Example 2, see Figure 1 The present invention provides a step diagram of a fluid adaptive control method. Figure 1 A fluid adaptive control method for a surgical box based on dynamic area adjustment is shown, comprising the following steps: Step S1: acquiring monitoring data of a multi-parameter sensor group in real time, including the pressure inside the cavity, fluid flow rate, and temperature parameters; Step S2: establishing a dynamic balance model based on the monitoring data. The dynamic balance model includes a nonlinear mapping relationship between cavity volume change and fluid pressure, and the model parameters are trained using historical data. Step S3: solving the dynamic balance model to generate coordinated control instructions, including: predicting the target displacement of the linear actuator based on the current pressure deviation, and dynamically calculating the optimized speed of the fluid pump based on the flow demand and temperature parameters; Step S4: Synchronously perform displacement compensation of the linear actuator and speed regulation of the fluid pump. When the pressure fluctuation exceeds the threshold, the wall plate displacement compensation mechanism is triggered first, and the cavity volume is adjusted by inverse solution model to quickly suppress the pressure fluctuation. Step S5: After the compensation phase is completed, multi-parameter feedback closed-loop control is started, and the model parameters are iteratively optimized according to the real-time updated sensor data to achieve dynamic matching between the effective working volume of the variable volume cavity and the fluid circulation.

[0029] See Figure 2 The flow chart of establishing a dynamic equilibrium model, in the above step S2, when establishing a dynamic equilibrium model based on monitoring data, includes: During the surgical preparation phase, a calibration fluid is injected, and the wall plate is driven by a linear actuator to perform multi-band reciprocating motion. The pressure-displacement response data is collected to construct an initial parameter matrix. A nonlinear partial differential equation is established to describe the coupling relationship between the cavity volume change rate and the fluid pressure gradient; The real-time feedback flow pulsation spectrum data is used to perform online parameter identification on the exponential decay term using a variable weight factor, where the high-frequency pulsation component corresponds to a weight factor of Regular decay, For real-time detection of pulsation frequency, is the viscosity compensation coefficient; The nonlinear mapping relationship is decomposed into the volume change dominant term and the fluid disturbance compensation term, which are solved by alternating iteration using the implicit Euler method and the explicit Adams-Bashforth method respectively.

[0030] Background: Instrument manipulation or tissue displacement during surgery can cause sudden changes in cavity pressure, requiring a fast system response. Traditional fixed-volume designs rely on single-variable control of fluid pumps, which have a delayed response and cannot compensate for nonlinear disturbances. In addition, the interaction between wall motion (structural dynamics) and fluid viscous resistance (fluid mechanics) must be modeled using coupled equations to capture the nonlinear effects of dynamic equilibrium. Based on this: In a possible embodiment, the nonlinear partial differential equation includes a quadratic term representing the displacement acceleration of the wall panel and an exponential decay term representing the viscous resistance of the fluid, as shown in the following equation: ; Where, is the dynamic displacement of the wall panel (unit: m); is the spatial coordinate, represents the normalized position along the expansion and contraction direction of the panel ( is the characteristic length of the cavity), is the time variable, is the quadratic coefficient, which characterizes the additional inertia effect caused by the intervention of surgical instruments (unit: ); is the fluid pressure field in the cavity (unit: Pa), ,in is the baseline pressure (the value set during surgery), is the elastic panel stiffness coefficient (Pa·m), is the turbulence coupling coefficient (Pa·m³); is the effective fluid density (unit: kg / m³), is the equivalent viscosity coefficient (unit: Pa·s), , is the baseline viscosity of Newtonian fluid, is the shear thinning critical rate (calibrated by a multi-parameter sensor set during surgery), is the displacement rate of the wall panel, and the dynamic displacement of the wall panel The first time derivative of (unit: m / s); is the viscous decay time constant (unit: s -1 ), , is the reference attenuation coefficient (unit: s -1 ), is the real-time temperature monitoring value, is the reference value of human core temperature, is the temperature sensitivity coefficient, is a natural constant.

[0031] It is necessary to further explain in the embodiment of the present invention that the above nonlinear partial differential equation introduces a quadratic term ( ) and the time-varying viscosity term ( ), which can more accurately predict the transient response under pressure fluctuations, better than traditional PID or linear models. The quadratic term represents the nonlinear inertial effect of the wall acceleration, and the time-varying viscosity term simulates the time-varying characteristics of the fluid viscous resistance, reflecting the dynamic effect of temperature changes on the fluid viscosity. In addition, the above nonlinear partial differential equation combines the real-time monitoring data (pressure, flow, temperature) during the operation, where the pressure gradient term on the right ( ) The pressure gradient that drives the fluid flow and the displacement rate of the wall together determine the dynamic balance of the cavity volume.

[0032] In the above step S2, when training the model parameters using historical data, it includes: Construct a multimodal historical dataset containing calibrated motion data, sudden pressure shock patterns, and corresponding instrument motion trajectory feature vectors for different types of surgical scenarios; An improved incremental Bayesian optimization algorithm is used for parameter training, including: In the offline stage, the quadratic coefficients in the nonlinear partial differential equation are calculated using the Kriging proxy model. , Newtonian fluid baseline viscosity , reference attenuation coefficient Perform global sensitivity analysis to screen out key trainable parameter sets; In the online stage, the training window length is adaptively adjusted by using the dynamic similarity evaluation between the real-time surgical data stream and historical data; Using quantum annealing algorithm to optimize the constrained loss function ; Establish a parameter drift compensation mechanism, when it is detected hour( is the dynamic drift threshold), triggering abnormal parameter reset based on the Generative Adversarial Network (GAN) to generate robust parameter initial values that match the current physiological signals.

[0033] Background: Data distribution varies significantly across surgical stages and individual patient differences, making fixed-window training difficult to adapt to dynamic features. Furthermore, in emergencies such as vascular rupture, sudden data changes require the model to quickly focus on the current state, shortening the training window to reduce interference from historical data. Based on this: In a possible embodiment, the training window length is adaptively adjusted as shown in the following formula: ; Where, is the dynamic training window length, is the basic window length, Adjust the gain factor for the window, is the KL divergence between the current data and the historical database, is the classification confidence of the surgical stage, is the time variable.

[0034] It is necessary to further explain in the embodiment of the present invention that the above adaptive adjustment process is carried out by KL divergence ( ) quantify the distribution difference between the real-time data stream and the historical database. The greater the difference, the greater the window adjustment. The classification confidence of the surgical stage ( ) reflects the certainty of the current scenario. When the confidence is low, the window is enlarged to enhance the stability of the model. When the confidence is high, the window is reduced to improve the response speed. Among them, the basic window length Used to ensure the minimum utilization of historical experience and prevent over-reliance on real-time noise data; window adjustment gain coefficient A hyperparameter that controls how much the window length varies.

[0035] Background description: There are noise and sudden disturbances in parameters such as pressure, flow, and temperature during surgery. Directly minimizing the prediction error can easily lead to overfitting of the model to noise or local extreme values. The above nonlinear partial differential equation contains quadratic coefficients , equivalent viscosity coefficient For coupled parameters, traditional loss functions are difficult to balance the physical constraints between parameters and the data fitting requirements. Based on this: In one possible embodiment, the dynamic weight factor in the constraint loss function is It is positively correlated with the intensity of instrument activity during the current surgical phase, as shown in the following formula: ; Where, is the total loss function, is the dynamic weight factor, For the The predicted pressure value of each sampling point, For the The measured pressure value of each sampling point, is the number of sampling points, is the regularization strength coefficient, is the trace of the inverse Hessian, representing the curvature of the model parameter space; It is necessary to further explain in the embodiment of the present invention that the dynamic weight factor in the above-mentioned constrained loss function is Dynamically adjust the weights according to the intensity of instrument activity during the surgical phase. For example, when high-frequency instruments are being operated, Increase to strengthen the penalty of current pressure deviation to avoid interference from historical data; reduce in the stable perfusion stage , focusing on long-term regularity learning. Regularization term The curvature of the model parameter space is constrained by the trace of the Hessian inverse to prevent the parameter update amplitude from being too large (such as the baseline viscosity of Newtonian fluid Sudden changes in pressure are avoided, ensuring the physical rationality of the fluid dynamics characteristics. Furthermore, the optimized parameters output by the loss function are directly used to solve the dynamic equilibrium model (nonlinear partial differential equations) in real time, driving the coordinated regulation of the wall plate displacement and the fluid pump speed. When pressure fluctuations exceed a threshold, the regularization constraints of the loss function quickly converge to a safe parameter domain, ensuring the stability of the inverse compensation mechanism.

[0036] In the above step S3, predicting the target displacement of the linear actuator based on the current pressure deviation includes: Construct a pressure-displacement time series prediction model, whose input is the real-time pressure deviation sequence within the sliding time window, and the output is the displacement increment of the next three control cycles ; A dual-channel convolutional long short-term memory (DC-LSTM) network is used to process multi-scale features: the high-frequency channel extracts the spatiotemporal correlation characteristics of pressure fluctuations and surgical electrosurgical unit start and stop events, while the low-frequency channel integrates the slowly varying parameter coupling characteristics of temperature drift and historical surgical stage labels. Based on the real-time pressure deviation sequence, the absolute value of the current pressure deviation is generated, and a dynamic compensation factor is obtained; Detection is based on the dynamic compensation factor. When the absolute value of the current pressure deviation is ( is the pressure fluctuation standard deviation), the emergency compensation protocol is activated: Based on preoperative CT / MRI images, the cavity deformation constraint boundary is reconstructed to generate The feasible displacement domain of Combined with intraoperative ultrasound Doppler blood flow velocity field data, the actuator acceleration curve is inversely optimized through the adjoint equation; in, is the cavity deformation vector field driven by the linear actuator, It is the maximum allowable volume change rate of the cavity based on preoperative CT / MRI reconstruction.

[0037] Background: Traditional surgical chambers rely on pressure monitoring within a fixed time window and are unable to dynamically adapt to sudden pressure changes (such as rapid instrument insertion or blood vessel rupture), resulting in compensation delays. Pressure fluctuations include both high-frequency transient shocks (such as the start and stop of an electrosurgical unit) and low-frequency trend changes (such as tissue perfusion pressure drift). A sliding window mechanism is required to separate pressure deviation characteristics at different time scales. Based on this: In a possible embodiment, the real-time pressure deviation sequence is as shown in the following formula: ; Where, is the real-time pressure deviation sequence, is the target pressure within the sliding time window, is the historical measured pressure, is the time variable, is the time lag index, ; is the window length parameter, which is determined by the dynamic training window length Decide, ,and , is the sampling time interval, is the minimum window constraint, is the maximum window constraint, It is a floor operation; It is necessary to further explain in the embodiment of the present invention that the above-mentioned real-time pressure deviation sequence is obtained by sliding the time window (window length By dynamic training window length Decision) intercepts the real-time pressure deviation sequence and balances the weight of real-time data and historical data. For example, when a sudden pressure shock occurs ( Increase, shrink), window length Automatically shorten and focus on the current disturbance mode. Each sampling point in the above real-time pressure deviation sequence is defined as the target pressure ( ) and actual pressure ( ) difference, through the lag index of historical pressure data ( ) constructs a time series to capture the dynamic evolution of pressure deviation.

[0038] Background description: Intraoperative pressure fluctuations (such as vascular rupture, rapid instrument insertion) are transient and nonlinear. Traditional linear threshold detection is prone to response delays or false triggering (such as noise interference). The standard deviation of pressure fluctuations in different surgical stages (such as tissue cutting and hemostasis) is ) significantly, the compensation trigger threshold needs to be dynamically adjusted to adapt to scene changes. Based on this: In a possible embodiment, the dynamic compensation factor is as shown in the following formula: ; Where, is the dynamic compensation factor, is the time variable, is the absolute value of the current pressure deviation, is the standard deviation of pressure fluctuation in the current surgical stage, Correction threshold for instrument contact force feedback, is a natural constant.

[0039] It is necessary to further explain in the embodiment of the present invention that the above dynamic compensation factor is obtained by an exponential function ( ) Set the current pressure deviation absolute value Mapping to the [0,1] interval allows for a smooth transition of the compensation factor and avoids sudden changes in control caused by hard thresholds. In addition, the pressure fluctuation standard deviation Reflects the pressure fluctuation intensity of the current surgical stage (such as when high-frequency electrosurgery is in operation). Increase), used for normalization ; Instrument contact force feedback correction threshold Dynamic adjustment is made through instrument contact force and tissue hardness (such as ultrasound elastography data). For example, when the contact force is greater than 5N, Reduce by 20%-30% to trigger compensation earlier.

[0040] In the above step S3, the dynamic calculation of the optimized speed of the fluid pump based on the flow demand and temperature parameters includes: Construct a multi-objective optimization function of flow rate, temperature and speed; Establishing three-dimensional pulse spectrum mapping relationship: constructing through particle image velocimetry (PIV) experiment in the offline stage ( , , , ) four-dimensional lookup table and compress it into a reduced-order model in the form of a tensor product; an online radial basis function (RBF) neural network is used to fill the gaps in the pulse spectrum during the operation, and the input layer fuses the ultrasound Doppler velocity profile and the infrared thermal imager temperature field data; Design anti-saturation dynamic limiting strategy: When it is detected Activate thermal protection constraints when: ,in For thermal protection, the speed is limited. is the rated speed; When the prothrombin time (PT) is abnormal, a flow rate change constraint is imposed: ,in is the absolute value of the speed change rate, is the turbulence suppression factor, is the temperature-dependent equivalent viscosity, is the fluid density; Nonlinear model predictive control (NMPC) was performed, in which the thrombosis risk index was embedded as a soft constraint, and the blood flow impedance boundary condition was updated in real time in combination with intraoperative OCT angiography data.

[0041] Background description: The flow demand is positively correlated with the fluid pump speed, but increasing the speed will intensify the fluid shear thermal effect, resulting in temperature ( ) increases and may exceed the tolerance threshold of biological tissues ( ), a dynamic balance needs to be struck between flow accuracy and temperature safety. Equivalent viscosity coefficient Due to the nonlinear influence of temperature (e.g. blood viscosity decreases with increasing temperature), the traditional single-variable control model cannot accurately describe the coupling relationship between flow rate, temperature and speed. In a possible embodiment, the flow-temperature-speed multi-objective optimization function is as shown in the following formula: ; Where, is the target flow demand, for Temperature-dependent pump efficiency correction factor at time t, is the fluid pump speed, is the current pressure deviation value, for The temperature-dependent equivalent viscosity coefficient at time , is the regularization weight coefficient, is a natural constant, is the temperature sensitivity index coefficient, is the real-time temperature monitoring value, The temperature threshold that biological tissue can tolerate.

[0042] It should be further explained in the embodiment of the present invention that the above multi-objective optimization function achieves the coordinated optimization of flow tracking and thermal damage prevention by minimizing the combined loss of the flow error term and the temperature risk term. Flow error term ( ) characterizes the deviation between actual flow and target demand, is the temperature-dependent pump efficiency correction factor (e.g. pump efficiency attenuation at high temperatures).

[0043] Temperature risk item ( ) The exponential function amplifies the overheating risk. > When , this term increases sharply, forcing the optimization result to shift toward the low temperature safety region.

[0044] See Figure 3 The flow chart for adjusting the cavity volume, in the above step S4, when adjusting the cavity volume by inversely solving the model to quickly suppress pressure fluctuations, includes: Establish the pressure fluctuation inverse mapping equation; Generate adversarial compensation signals in real time: Input the ultrasonic blood flow vector field into the pre-trained 3D Flow-GAN network to generate virtual displacement compensation that matches the current anatomical structure and fuse it with the actual pressure gradient data; Deploy a distributed real-time solution architecture: perform finite element reverse solution at the edge computing node, process high-resolution pressure fluctuation data (sampling rate ≥ 1kHz), and execute global constraints on cavity deformation based on topology optimization to ensure ;in, is the maximum modulus length of the dynamic displacement gradient of the panel, is the displacement safety threshold; Dynamic reconstruction of the wall panel motion trajectory: when the current pressure deviation value is detected When the emergency inverse solution mode is activated, the anisotropic constraint boundary of the biological tissue is constructed based on the preoperative DTI fiber tracking data, the Monte Carlo annealing algorithm is used to solve the optimal displacement sequence under the time-varying constraint, and the acceleration of the linear actuator is limited to ( is the acceleration due to gravity).

[0045] Background description: Traditional fluid pump regulation relies on a forward pressure-flow model, which has an inherent delay in responding to sudden pressure shocks such as vascular rupture (such as pressure wave transmission time), making it difficult to meet clinical safety threshold requirements; directly adjusting the cavity volume forward may exceed the deformation safety boundary reconstructed by preoperative imaging (such as fibrotic tissue area), requiring an inverse solution to the displacement compensation that meets the deformation constraints. Based on this: In a possible embodiment, the pressure fluctuation inverse mapping equation is as shown below: ; Where, is the inverse mapping residual function, Control volume for the cavity; is the sensitivity of pressure to displacement, which is calculated in real time by automatic differentiation (AD) based on nonlinear partial differential equations. is the fluid pressure field in the cavity, is the dynamic displacement of the wall panel, is the displacement change, is the fluid stress dynamic coupling coefficient, is the fluid stress tensor, is the stress-displacement divergence term, is the regularization weight coefficient, is the H1 norm constraint.

[0046] It is necessary to further explain in the embodiment of the present invention that the pressure fluctuation inverse mapping equation controls the sensitivity of the pressure in the volume to the displacement by integrating ( ) and displacement change ( ), combined with the fluid stress divergence term ( ) and H1 norm regularization term to construct the inverse mapping residual function , solve the displacement compensation that minimizes the residual . The fluid stress dynamic coupling coefficient ( ) is used to balance the fluid stress field ( ) and the displacement field ( ) is calibrated during surgery through a multi-parameter sensor group (such as the correlation between blood flow shear force and wall acceleration); the H1 norm regularization term is used to constrain the smoothness of the displacement compensation amount to avoid tissue damage caused by high-frequency oscillation deformation.

[0047] Background: The virtual displacement compensation generated by 3D Flow-GAN is based on preoperative imaging and prior knowledge of hemodynamics. However, actual anatomical deformation during surgery (such as tissue adhesion or instrument intervention) may lead to prediction errors, requiring the integration of real-time pressure gradient data to correct the compensation. In addition, rapid displacement of the wall panel may induce cavity pressure oscillations (such as resonance effects), requiring stability criteria to suppress the dynamic divergence of the compensation action. Based on this: In a possible embodiment, the actual pressure gradient data and the virtual displacement compensation are fused by the Lyapunov stability criterion. The fusion process is shown in the following formula: ; Where, is the final displacement compensation amount, is the actual pressure gradient field, is the displacement change, is the fusion weight coefficient, is the virtual displacement compensation, is the pressure gradient-displacement coupling term, Tracking items for virtual compensation; It is necessary to further explain in the embodiment of the present invention that the above fusion process minimizes the pressure gradient-displacement coupling term ( ) and the virtual compensation tracking item ( ), construct the quadratic Lyapunov function to ensure the final displacement compensation after fusion ( ) to meet the real-time pressure suppression requirements and the anatomical adaptability of the virtual model. In addition, the fusion weight coefficient Adaptive adjustment based on the frequency of pressure fluctuations during surgery. For example, it increases when there is a high-frequency pressure shock (such as electrosurgery interference). value, strengthening the role of real-time pressure gradient data; reducing values, focusing on the anatomical constraints of the virtual model.

[0048] In the above step S5, when iteratively optimizing the model parameters according to the real-time updated sensor data, it includes: Construct a multi-rate fusion feedback matrix: 1-2kHz high-frequency sampling is implemented for pressure sensors, 100-200Hz medium-frequency sampling is used for flow and temperature parameters, and the instrument motion trajectory is synchronized at a low frequency of 30-50Hz. A federated Kalman filter is used to fuse multi-source data streams, in which the pressure channel embeds the inverse solution residual output of the pressure fluctuation inverse mapping equation as a process noise covariance correction term. Design a two-layer nested optimization engine, including: The inner layer performs online sparse dictionary learning: extracting dominant eigenmodes from real-time data streams and constructing a dynamic subspace constraint set; The outer layer runs an adversarial reinforcement learning (ARL) strategy, including: The generator network simulates virtual pressure-flow perturbations for extreme surgical scenarios (e.g., portal vein burst); The discriminator network combines intraoperative fluorescence angiography data to identify the true disturbance pattern and iteratively optimize the robustness of the control strategy; Implement anatomical-rheological coupling constraints, including: Convert intraoperative ultrasound elastography data into local tissue deformation, that is, into deformation Jacobian matrix (used to describe the deformation characteristics of tissue after being subjected to force), impose constraints , is the dynamic displacement gradient of the wall panel; is the deformation safety threshold, which is set dynamically according to the tissue type; is the Frobenius norm; Automatically shrink the feasible displacement domain when fibrotic tissue areas are detected: , is the displacement safety threshold.

[0049] Background: Intraoperative multi-sensor data (pressure, flow, temperature, and instrument motion trajectory) is input into the system at varying sampling rates. Directly processing this raw, high-dimensional data results in computational delays, making it difficult to meet real-time control requirements. Furthermore, the data stream contains a significant amount of redundant information (such as low-frequency pressure drift during steady-state perfusion) and noise (high-frequency interference from electrosurgical scalpels). Extracting dominant eigenmodes is crucial to reduce model complexity.

[0050] In a possible embodiment, the dynamic subspace constraint set is as shown below: ; Where, is the dynamic subspace constraint set, is the dynamic basis vector, is the subspace dimension constraint, is the number of dynamic basis vectors, is a linearly generated subspace, is the time variable.

[0051] It is necessary to further explain in the embodiment of the present invention that the above dynamic subspace constraint set extracts dynamic basis vectors from the real-time data stream through online sparse dictionary learning. , filter out the following 5 dominant characteristic modes that are strongly related to the current surgical scene ( ),For example: High-frequency pressure shock mode: corresponds to electrosurgical operation or blood vessel rupture events.

[0052] Low-frequency deformation trend mode: reflects slow tissue displacement or changes in perfusion pressure.

[0053] In addition, the above-mentioned dynamic subspace constraint set projects real-time data into a low-dimensional dynamic subspace, reducing the dimension of optimization variables and accelerating the convergence speed of quantum annealing algorithm and adversarial reinforcement learning (ARL).

[0054] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A surgical operating box based on dynamic area adjustment, comprising a box body, a fluid circulation system and a control system, characterized in that: The box body comprises at least two sets of relatively movable wall panels, which are connected by a telescopic guide mechanism to form a variable volume cavity; An area adjustment drive assembly is provided between the wall panels, and the area adjustment drive assembly includes a linear actuator and a displacement sensor; The fluid circulation system comprises a fluid pump, a pressure regulating valve and a multi-parameter sensor group connected to the variable volume cavity, wherein the multi-parameter sensor group comprises: a pressure sensor, a flow sensor and a temperature sensor; The control system includes a main control module and an adaptive algorithm module. The main control module is connected to the area adjustment drive assembly and the fluid circulation system. The adaptive algorithm module is configured as follows: Establish a dynamic balance model based on real-time monitoring data from a multi-parameter sensor group; Outputting a coordinated control signal through a solution model to synchronously adjust the displacement of the linear actuator and the speed of the fluid pump; When the pressure fluctuation is detected to exceed the threshold, the wall plate displacement compensation mechanism is triggered first.

2. A method for adaptively controlling fluid in a surgical operating box according to claim 1, characterized in that: The following steps are involved: Real-time acquisition of monitoring data from a multi-parameter sensor group, including cavity internal pressure, fluid flow, and temperature parameters; Establishing a dynamic balance model based on the monitoring data, the dynamic balance model includes a nonlinear mapping relationship between cavity volume change and fluid pressure, and training model parameters through historical data; Solving the dynamic balance model to generate coordinated control instructions includes: predicting a target displacement of the linear actuator based on a current pressure deviation, and dynamically calculating an optimized speed of the fluid pump based on flow demand and temperature parameters; The linear actuator's displacement compensation and the fluid pump's speed regulation are performed synchronously. When the pressure fluctuation exceeds the threshold, the wall plate displacement compensation mechanism is triggered first. The cavity volume is adjusted through the inverse solution model to quickly suppress the pressure fluctuation. After the compensation phase, multi-parameter feedback closed-loop control is started to iteratively optimize the model parameters based on the real-time updated sensor data.

3. The fluid adaptive control method according to claim 2, characterized in that: When establishing a dynamic equilibrium model based on the monitoring data, it includes: During the surgical preparation phase, a calibration fluid is injected, and the wall plate is driven by a linear actuator to perform multi-band reciprocating motion. The pressure-displacement response data is collected to construct an initial parameter matrix. A nonlinear partial differential equation is established, which includes a quadratic term representing the displacement acceleration of the wall panel and an exponential decay term representing the viscous resistance of the fluid. Using the real-time feedback flow pulsation spectrum data to perform online parameter identification on the exponential decay term using a variable weight factor; The nonlinear mapping relationship is decomposed into a volume change dominant term and a fluid disturbance compensation term, and each term is solved by alternate iterative methods.

4. The fluid adaptive control method according to claim 3, characterized in that: When training model parameters using historical data, this includes: Construct a multimodal historical dataset containing calibrated motion data, sudden pressure shock patterns, and corresponding instrument motion trajectory feature vectors for different types of surgical scenarios; An improved incremental Bayesian optimization algorithm is used to perform parameter training through adaptive adjustment; The quantum annealing algorithm is used to optimize the constrained loss function, where the dynamic weight factor It is positively correlated with the intensity of instrument activity during the current surgical phase; A parameter drift compensation mechanism is established to trigger abnormal parameter reset based on the adversarial generative network.

5. The fluid adaptive control method according to claim 4, characterized in that: When using the improved incremental Bayesian optimization algorithm for parameter training, it includes: In the offline stage, a global sensitivity analysis of the nonlinear partial differential equation is performed using a Kriging proxy model to screen out a set of key trainable parameters; In the online stage, the dynamic similarity evaluation between real-time surgical data stream and historical data is used to adaptively adjust the training window length.

6. The fluid adaptive control method according to claim 2, characterized in that: When predicting the target displacement of a linear actuator, include: A pressure-displacement time series prediction model is constructed, whose input is the real-time pressure deviation sequence within the sliding time window, and the output is the displacement increment in the next three control cycles; Use dual-channel convolutional long short-term memory network to process multi-scale features; Based on the real-time pressure deviation sequence, the absolute value of the current pressure deviation is generated, and a dynamic compensation factor is obtained; Based on the dynamic compensation factor, when the absolute value of the current pressure deviation is greater than the set value, the emergency compensation protocol is activated: Reconstruct the cavity deformation constraint boundary based on preoperative CT / MRI images and generate a feasible displacement domain that meets the preset conditions; Combined with intraoperative ultrasound Doppler blood flow velocity field data, the actuator acceleration curve is inversely optimized through the adjoint equation.

7. The fluid adaptive control method according to claim 2, characterized in that: Dynamic calculation of the optimal speed of a fluid pump includes: Construct a multi-objective optimization function of flow rate, temperature and speed; Establishing three-dimensional pulse spectrum mapping relationship; Design an anti-saturation dynamic limiting strategy: When the absolute value of the difference between the real-time temperature monitoring value and the biological tissue tolerance temperature threshold is detected to be greater than the set value, the thermal protection constraint is activated; When the prothrombin time is abnormal, a flow rate change constraint is imposed; Nonlinear model predictive control was implemented, in which a thrombosis risk index was embedded as a soft constraint, and the blood flow impedance boundary condition was updated in real time in combination with intraoperative OCT angiography data.

8. The fluid adaptive control method according to claim 2, characterized in that: When adjusting the cavity volume by inverse solving the model, it includes: Establish the pressure fluctuation inverse mapping equation; Real-time generation of adversarial compensation signals: The ultrasonic blood flow vector field is input into the pre-trained 3D Flow-GAN network to generate a virtual displacement compensation that matches the current anatomical structure. The actual pressure gradient data and the virtual displacement compensation are then fused using the Lyapunov stability criterion. Deploy a distributed real-time solution architecture; Dynamic reconstruction of the wall panel motion trajectory: When it is detected that the current pressure deviation value is greater than the set value, the emergency reverse solution mode is activated.

9. The fluid adaptive control method according to claim 8, characterized in that: When iteratively optimizing model parameters, including: Construct a multi-rate fusion feedback matrix: sample multiple source data streams at different frequencies and fuse them through federated Kalman filtering; Design a two-layer nested optimization engine, including: the inner layer performs online sparse dictionary learning, and the outer layer runs an adversarial reinforcement learning strategy.

10. The fluid adaptive control method according to claim 9, characterized in that: When iteratively optimizing model parameters, it also includes: Implementing anatomical-rheological coupling constraints: Converting intraoperative ultrasound elastography data into a deformation Jacobian matrix , impose constraints , is the dynamic displacement gradient of the wall panel; is the deformation safety threshold, is the Frobenius norm; Automatically shrink the feasible displacement domain when fibrotic tissue areas are detected: , is the displacement safety threshold.

Images