Intelligent shock absorption and attitude control method and system for field medical equipment

Abstract

The application provides a field medical equipment intelligent damping and posture control method and system, relates to the technical field of field medical equipment, and comprises the following steps: acquiring pose sensor data and damping force state, constructing a data matrix, generating a damping control instruction by using wavelet transform, principal component analysis and adaptive fuzzy control, calculating control parameters by discrete Fourier transform and least square fitting, and monitoring the damping effect in real time, so that the stability of medical equipment in a field environment can be improved, the interference of vibration on medical operation can be reduced, and the medical treatment effect in a critical situation can be ensured.

Description

Technical Field

[0001] This invention relates to the field of battlefield medical equipment technology, and in particular to a method and system for intelligent shock absorption and attitude control of battlefield medical equipment. Background Technology

[0002] Battlefield medical equipment is a key piece of equipment in military medical rescue. It needs to operate stably under various complex terrains and environmental conditions to provide timely and effective medical treatment for the wounded. In the battlefield environment, uneven terrain, explosion impact, vibration and other factors can seriously affect the stability of medical equipment, especially the accuracy of precision medical instruments and monitoring equipment, thereby affecting the medical treatment effect.

[0003] Traditional battlefield medical equipment typically employs simple mechanical shock absorption structures. While passive shock absorption systems are simple in structure, they cannot adapt to vibrations and impacts of different frequencies, resulting in limited shock absorption effects. Active shock absorption control technology, on the other hand, acquires environmental information through a sensing system and adjusts the parameters of the shock absorption device in real time through a control system to achieve the best shock absorption effect.

[0004] However, existing active vibration damping systems still have insufficient adaptability to complex battlefield environments. In particular, when faced with a mixture of sudden strong impacts and continuous random vibrations, they are unable to quickly identify vibration characteristics and respond effectively. This can lead to unstable attitudes of medical equipment at critical moments, low computational efficiency of control algorithms, and difficulty in achieving real-time control. Especially in resource-constrained battlefield environments, the response delay of the control system is significant, failing to meet the rapid and stable requirements of medical equipment. Furthermore, the lack of adaptive learning capabilities means that the control strategy cannot be automatically adjusted according to different terrain conditions and vibration modes, and the long-term stable operation of medical equipment cannot be guaranteed. Summary of the Invention

[0005] This invention provides a method and system for intelligent shock absorption and attitude control of battlefield medical equipment, which can at least solve some of the problems existing in the prior art.

[0006] A first aspect of this invention provides a method for intelligent shock absorption and attitude control of battlefield medical equipment, comprising:

[0007] Acquire the position sensor data of the battlefield medical equipment and calculate the attitude angle data; collect the pressure data and displacement data on the shock absorber and determine the current stress state.

[0008] The attitude angle data and the current force state are arranged in time sequence to construct a data matrix. Wavelet transform is performed on the data matrix to obtain frequency domain features. Principal component analysis is performed on the frequency domain features to extract feature vectors. The feature vectors are mapped in fuzzy space based on an adaptive fuzzy control algorithm to obtain membership values. Compensation rules are selected according to the membership values ​​and the weight coefficients of the compensation rules are optimized by a genetic algorithm. The optimized weight coefficients and the membership values ​​are weighted and summed to obtain the vibration reduction compensation value and generate vibration reduction control commands.

[0009] The vibration reduction compensation value is subjected to discrete Fourier transform to obtain spectral components. A state feedback matrix is ​​constructed based on the spectral components and convolved with a preset feature compensation matrix to obtain a control gain matrix. The vibration reduction compensation value is fitted with least squares based on the control gain matrix to calculate the current control parameters and attitude adjustment parameters of the vibration damper. The vibration damper and Mecanum wheel of the field medical equipment are controlled according to the current control parameters and attitude adjustment parameters.

[0010] The vibration reduction effect is monitored in real time. If the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude.

[0011] In one alternative implementation,

[0012] Acquiring posture sensor data from the battlefield medical equipment and calculating attitude angle data, collecting pressure and displacement data from the shock absorber and determining the current stress state includes:

[0013] The field medical equipment's posture sensor data and shock absorber sensing data are collected, wherein the posture sensor data includes acceleration data, angular velocity data, and geomagnetic data, and the shock absorber sensing data includes pressure data and displacement data.

[0014] The acceleration data, angular velocity data, and geomagnetic data are combined in time sequence, and the predicted state is calculated through a pre-set state equation. The predicted state is then corrected by combining the actual observation values ​​to obtain the attitude angle data of the battlefield medical equipment.

[0015] The pressure data and displacement data are paired by timestamp, and the current stress state of the shock absorber is determined by the ratio of the pressure change to the displacement change at adjacent times.

[0016] In one alternative implementation,

[0017] A data matrix is ​​constructed by arranging the attitude angle data and the current force state in time sequence. Wavelet transform is performed on the data matrix to obtain frequency domain features. Principal component analysis is then performed on the frequency domain features to extract feature vectors. Based on an adaptive fuzzy control algorithm, these feature vectors are mapped in the fuzzy space to obtain membership values, including:

[0018] The attitude angle data and the current force state are arranged into a data matrix according to the sampling time sequence. After normalization, the attitude angle data and the current force state are sorted by timestamp, so that each row of the data matrix corresponds to the attitude angle data and the current force state at a sampling time.

[0019] The data matrix is ​​subjected to wavelet transform to obtain a wavelet coefficient matrix. The energy spectrum and signal-to-noise ratio of the wavelet coefficient matrix are calculated. The resolution of the scale parameter is adaptively adjusted based on the signal-to-noise ratio. The sliding window size of the translation parameter is dynamically selected according to the energy spectrum. The data matrix is ​​decomposed and reconstructed at multiple scales based on the scale parameter and the sliding window to extract the frequency domain features.

[0020] Calculate the covariance matrix of the frequency domain features, use the singular value decomposition method to perform eigenvalue decomposition on the covariance matrix to obtain an initial feature group, select the initial features in the initial feature group whose cumulative contribution rate reaches a preset contribution threshold to construct an orthogonal projection matrix, and multiply the frequency domain features with the orthogonal projection matrix to obtain the feature vector;

[0021] The feature vectors are clustered using a fuzzy C-means clustering algorithm to obtain cluster centers. An exponential function is used to construct a membership function. Based on the membership function, the feature vectors are mapped to a fuzzy space and initial membership values ​​are calculated. Based on the initial membership values, the cluster centers are iteratively optimized using a pre-set performance index function and gradient descent. The membership values ​​are then recalculated and output based on the optimized cluster centers.

[0022] In one alternative implementation,

[0023] The compensation rule is selected based on the membership value, and the weight coefficients of the compensation rule are optimized using a genetic algorithm. The optimized weight coefficients are then weighted and summed with the membership value to obtain the damping compensation value and generate damping control commands, including:

[0024] A compensation rule base is constructed based on the membership value. The compensation rule base includes multiple compensation rules. Each compensation rule includes a membership value and a corresponding initial compensation value. Based on the membership value and the initial compensation value, the rule activation degree of each compensation rule is calculated using the algebraic product T-norm operator.

[0025] Initialize the rule weight coefficients corresponding to the compensation rule, encode the rule weight coefficients as chromosomes, construct a fitness function based on control error and control increment, use the roulette wheel algorithm to select the rule weight coefficients and perform arithmetic crossover operation to obtain crossover weight coefficients, perform non-uniform mutation operation on the crossover weight coefficients and iteratively optimize according to the fitness function to obtain optimized rule weight coefficients;

[0026] The optimized rule weight coefficient is multiplied by the rule activation degree to obtain the weighted activation degree. The weighted activation degree is multiplied by the initial compensation value and summed to obtain the compensation sum value. The weighted activation degrees are summed to obtain the activation sum value. The compensation sum value is divided by the activation sum value to obtain the compensation control amount.

[0027] The compensation control quantity is compared with the preset compensation upper and lower limits to obtain the amplitude limiting compensation value. The difference between the amplitude limiting compensation value and the compensation control quantity at the previous moment is calculated to obtain the control increment. The control increment is compared with the preset increment upper and lower limits to obtain the limited control increment. The limited control increment is added to the compensation control quantity at the previous moment to obtain the damping compensation value corresponding to the current moment and a damping control command is generated.

[0028] In one alternative implementation,

[0029] The vibration reduction compensation value is subjected to discrete Fourier transform to obtain spectral components. A state feedback matrix is ​​constructed based on the spectral components and convolved with a preset feature compensation matrix to obtain a control gain matrix, including:

[0030] The vibration reduction compensation value is subjected to discrete Fourier transform to obtain the spectral components. The spectral components are decomposed into real and imaginary parts. The amplitude spectrum is calculated based on the square root of the sum of the squares of the real and imaginary parts. The phase spectrum is calculated based on the arctangent function of the quotient of the imaginary and real parts.

[0031] The amplitude spectrum and the phase spectrum are arranged in order of frequency components to construct a state feedback matrix. The state feedback matrix includes an amplitude spectrum sequence in the first row and a phase spectrum sequence in the second row. Multiple preset frequency weight coefficients are multiplied by the amplitude spectrum and phase spectrum of the corresponding frequency components in the state feedback matrix to obtain a weighted state feedback matrix.

[0032] A feature compensation matrix is ​​constructed, which includes multiple amplitude compensation coefficients in the first row and multiple phase compensation coefficients in the second row. The weighted state feedback matrix and the feature compensation matrix are subjected to a two-dimensional convolution operation. The control gain matrix is ​​calculated based on the sum of the products of the elements in the weighted state feedback matrix and the compensation coefficients at the corresponding positions in the feature compensation matrix.

[0033] In one alternative implementation,

[0034] Based on the control gain matrix, the vibration reduction compensation value is fitted using least squares to calculate the current control parameters and attitude adjustment parameters of the shock absorber. Controlling the shock absorber and Mecanum wheel of the field medical equipment according to the current control parameters and attitude adjustment parameters includes:

[0035] Obtain the control gain matrix, multiply the control gain matrix by a preset first control coefficient and add a bias coefficient to obtain the shock absorber current control quantity, calculate the sum of squares of the difference between the shock absorber current control quantity and the preset expected current value and construct a first objective function, and under the constraints that the current change rate does not exceed the preset change rate upper limit and the current amplitude does not exceed the preset amplitude limit, minimize the function value corresponding to the first objective function to solve for the first control coefficient and bias coefficient as current control parameters;

[0036] The attitude control input is obtained by multiplying the control gain matrix by a pre-set first attitude coefficient and superimposing the attitude error by a second attitude coefficient. A Lyapunov function containing the attitude error and the pre-acquired angular velocity is constructed. The derivative of the Lyapunov function is weighted with the sum of the squares of the attitude control input to construct a second objective function. The function value corresponding to the second objective function is minimized to obtain the first attitude coefficient and the second attitude coefficient as attitude adjustment parameters.

[0037] The product of the current control parameter and the attitude error is added to the shock absorber current control quantity to obtain the actual control current of the shock absorber. The product of the attitude adjustment parameter and the difference between the actual control current of the shock absorber and the shock absorber current control quantity is added to the attitude control input to obtain the wheel system control command. The shock absorber and Mecanum wheel are controlled in a coordinated manner according to the actual control current of the shock absorber and the wheel system control command.

[0038] In one alternative implementation,

[0039] Real-time monitoring of the vibration reduction effect; if the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude, including:

[0040] Vibration acceleration signals are collected by an accelerometer installed on the shock absorber, and vibration displacement signals are collected by a displacement sensor. The root mean square value of the vibration acceleration signal and the maximum amplitude of the vibration displacement signal are used as evaluation indicators of the shock absorption effect. When the evaluation indicators exceed the preset shock absorption threshold, parameter optimization is triggered.

[0041] The current attitude is obtained, and the sum of squares of the difference between the current control value of the shock absorber and the expected current value is fitted by least squares to obtain the updated current control parameters. The Lyapunov function constructed by the attitude error and angular velocity is optimized to obtain the updated attitude adjustment parameters.

[0042] The updated current control parameters and the updated attitude adjustment parameters are applied to the coordinated control of the shock absorber and the Mecanum wheel, respectively, until the evaluation index is less than the preset shock absorption threshold.

[0043] A second aspect of the present invention provides an intelligent shock absorption and attitude control system for battlefield medical equipment, comprising:

[0044] The first unit is used to acquire the posture sensor data of the battlefield medical equipment and calculate the posture angle data, collect the pressure data and displacement data on the shock absorber and determine the current stress state.

[0045] The second unit is used to construct a data matrix by arranging the attitude angle data and the current force state in a time sequence, perform wavelet transform on the data matrix to obtain frequency domain features, perform principal component analysis on the frequency domain features to extract feature vectors, map the feature vectors in the fuzzy space based on an adaptive fuzzy control algorithm to obtain membership values, select compensation rules according to the membership values ​​and optimize the weight coefficients of the compensation rules through a genetic algorithm, and perform a weighted sum of the obtained optimized weight coefficients and membership values ​​to obtain vibration reduction compensation values ​​and generate vibration reduction control commands.

[0046] The third unit is used to perform discrete Fourier transform on the vibration reduction compensation value to obtain spectral components, construct a state feedback matrix based on the spectral components and perform convolution operation with a preset feature compensation matrix to obtain a control gain matrix, perform least squares fitting on the vibration reduction compensation value based on the control gain matrix, calculate the current control parameters and attitude adjustment parameters of the vibration damper respectively, and control the vibration damper and Mecanum wheel of the battlefield medical equipment according to the current control parameters and attitude adjustment parameters.

[0047] The fourth unit is used to monitor the vibration reduction effect in real time. If the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude.

[0048] A third aspect of the present invention provides an electronic device, comprising:

[0049] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0050] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0051] This invention achieves efficient vibration reduction control for battlefield medical equipment under complex terrain conditions by using a vibration reduction compensation mechanism optimized by an adaptive fuzzy control algorithm combined with a genetic algorithm. This effectively reduces the impact of equipment vibration on medical operations and patients. By employing a method combining frequency domain analysis and state feedback control, and constructing a control gain matrix and performing least-squares fitting, the current control parameters and attitude adjustment parameters of the vibration damper are accurately calculated. This enables the battlefield medical equipment to automatically adjust its vibration reduction force and attitude according to terrain changes, improving the stability and adaptability of the medical equipment in the battlefield environment. The control parameters are automatically updated based on the current attitude, forming a closed-loop control mechanism that ensures that the battlefield medical equipment maintains its optimal working state in various complex environments, thereby improving the safety and effectiveness of medical care. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the intelligent shock absorption and attitude control method for battlefield medical equipment according to an embodiment of the present invention.

[0053] Figure 2 This is a flowchart illustrating the data processing of battlefield medical equipment in an embodiment of the intelligent shock absorption and attitude control method for battlefield medical equipment according to the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0056] Figure 1 This is a flowchart illustrating the intelligent shock absorption and attitude control method for battlefield medical equipment according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes:

[0057] Acquire the position sensor data of the battlefield medical equipment and calculate the attitude angle data; collect the pressure data and displacement data on the shock absorber and determine the current stress state.

[0058] The attitude angle data and the current force state are arranged in time sequence to construct a data matrix. Wavelet transform is performed on the data matrix to obtain frequency domain features. Principal component analysis is performed on the frequency domain features to extract feature vectors. The feature vectors are mapped in fuzzy space based on an adaptive fuzzy control algorithm to obtain membership values. Compensation rules are selected according to the membership values ​​and the weight coefficients of the compensation rules are optimized by a genetic algorithm. The optimized weight coefficients and the membership values ​​are weighted and summed to obtain the vibration reduction compensation value and generate vibration reduction control commands.

[0059] The vibration reduction compensation value is subjected to discrete Fourier transform to obtain spectral components. A state feedback matrix is ​​constructed based on the spectral components and convolved with a preset feature compensation matrix to obtain a control gain matrix. The vibration reduction compensation value is fitted with least squares based on the control gain matrix to calculate the current control parameters and attitude adjustment parameters of the vibration damper. The vibration damper and Mecanum wheel of the field medical equipment are controlled according to the current control parameters and attitude adjustment parameters.

[0060] The vibration reduction effect is monitored in real time. If the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude.

[0061] In one alternative implementation,

[0062] Acquiring posture sensor data from the battlefield medical equipment and calculating attitude angle data, collecting pressure and displacement data from the shock absorber and determining the current stress state includes:

[0063] The field medical equipment's posture sensor data and shock absorber sensing data are collected, wherein the posture sensor data includes acceleration data, angular velocity data, and geomagnetic data, and the shock absorber sensing data includes pressure data and displacement data.

[0064] The acceleration data, angular velocity data, and geomagnetic data are combined in time sequence, and the predicted state is calculated through a pre-set state equation. The predicted state is then corrected by combining the actual observation values ​​to obtain the attitude angle data of the battlefield medical equipment.

[0065] The pressure data and displacement data are paired by timestamp, and the current stress state of the shock absorber is determined by the ratio of the pressure change to the displacement change at adjacent times.

[0066] Data from the field medical equipment's attitude sensor and shock absorber sensors were collected. The attitude sensor data included triaxial acceleration, triaxial angular velocity, and triaxial geomagnetic data, provided by the inertial measurement unit mounted on the field medical equipment. The shock absorber sensor data included pressure and displacement data, provided by pressure and displacement sensors mounted on the shock absorber. The acquisition frequency was set to 100Hz to ensure data continuity and integrity.

[0067] The collected acceleration data includes linear acceleration values ​​in the x, y, and z directions, with units of m / s². 2 The numerical values ​​range from ±16g. Angular velocity data includes the rotational speed about the x, y, and z axes, measured in rad / s, and ranges from ±2000° / s. Geomagnetic data includes the magnetic field strength in the x, y, and z directions, measured in μT, and ranges from ±4900μT. For example, data collected at a certain time t might be: acceleration data (1.02, -0.58, 9.78) m / s², angular velocity data (0.03, 0.01, -0.02) rad / s, and geomagnetic data (23.5, -12.3, 42.1) μT.

[0068] Among the collected shock absorber sensing data, the pressure data represents the pressure value experienced by the shock absorber, measured in kPa, and typically ranging from 0 to 100 mm. For example, the data collected at time t might show a pressure of 350 kPa and a displacement of 25 mm.

[0069] Acceleration, angular velocity, and geomagnetic data are combined in time sequence, and the attitude angle data of the battlefield medical equipment is calculated using a Kalman filter algorithm. The state vector is defined as the equipment's Euler angles, including roll, pitch, and yaw. At time t, the state at the current time is predicted based on the state estimate from the previous time t-1 and the angular velocity data. Taking the roll angle as an example, the predicted value at the current time is equal to the estimate from the previous time plus the product of the x-axis component of the angular velocity and the sampling time interval. A similar calculation method is used for the pitch and yaw angles.

[0070] Observations are calculated using acceleration and geomagnetic data. Acceleration data can be used to calculate the direction of gravity, thereby obtaining the observed roll and pitch angles. For example, in a stationary state, the roll angle can be calculated using the arctangent functions of the y-axis and z-axis accelerations, and the pitch angle can be calculated using the arctangent function of the x-axis and the composite acceleration due to gravity. Geomagnetic data is used to calculate the observed yaw angle, using the geomagnetic components on the horizontal plane to calculate the direction relative to magnetic north.

[0071] The predicted value and the observed value are fused to obtain the final state estimation value. During the fusion process, different weights are assigned according to the uncertainties of the predicted value and the observed value. When the device is in a state of剧烈运动状态 (intense motion), the observed value may be disturbed, and at this time, the weight of the predicted value is increased; when the device is relatively stationary, the weight of the observed value is increased. Through this adaptive weight adjustment, stable attitude estimation can be obtained under different motion states.

[0072] Exemplarily, the attitude estimation result at a certain moment is: roll angle 5.2°, pitch angle -2.3°, yaw angle 78.1°. The pressure data and displacement data are paired according to the time stamp, and the force state of the shock absorber is analyzed. For two groups of data at time t and t+Δt, calculate the pressure change amount ΔP = P(t+Δt) - P(t) and the displacement change amount ΔX = X(t+Δt) - X(t), and then calculate the ratio K = ΔP / ΔX. This ratio represents the current stiffness characteristic of the shock absorber.

[0073] According to the mechanical characteristics of the shock absorber, set multiple stiffness thresholds K1, K2,..., K n , and divide the force state of the shock absorber into several intervals. For example, when K < K1, it is determined as a slightly stressed state; when K1 ≤ K < K2, it is determined as a moderately stressed state; when K ≥ K2, it is determined as a severely stressed state. For specific application scenarios, these thresholds can be determined through experiments. For example, K1 = 5 kPa / mm and K2 = 15 kPa / mm can be set.

[0074] In practical applications, in a certain measurement, the pressure data at two adjacent moments (10 ms apart) may be 320 kPa and 345 kPa respectively, and the displacement data may be 22 mm and 25 mm respectively. Calculate ΔP = 25 kPa, ΔX = 3 mm, and the ratio K = 8.33 kPa / mm, and it is determined as a moderately stressed state.

[0075] In this embodiment, by performing a time series combination of acceleration data, angular velocity data, and geomagnetic data, and introducing a preset state equation to calculate the predicted state, and then correcting the predicted state in combination with the actual observed value, accurate attitude angle data is obtained, realizing real-time perception and high-precision tracking of the attitude change of the battlefield medical equipment. The pressure data and displacement data of the shock absorber are paired according to the time stamp, and the force state of the shock absorber is determined by the ratio of the pressure change amount and the displacement change amount at adjacent moments, thus realizing the dynamic monitoring of the shock absorption performance of the equipment and improving the stability and reliability of the equipment in a complex battlefield environment.

[0076] Figure 2 This is the flowchart of the data processing of the battlefield medical equipment for the intelligent shock absorption and attitude control method of the embodiment of the present invention.

[0077] In an optional implementation manner,

[0078] A data matrix is ​​constructed by arranging the attitude angle data and the current force state in time sequence. Wavelet transform is performed on the data matrix to obtain frequency domain features. Principal component analysis is then performed on the frequency domain features to extract feature vectors. Based on an adaptive fuzzy control algorithm, these feature vectors are mapped in the fuzzy space to obtain membership values, including:

[0079] The attitude angle data and the current force state are arranged into a data matrix according to the sampling time sequence. After normalization, the attitude angle data and the current force state are sorted by timestamp, so that each row of the data matrix corresponds to the attitude angle data and the current force state at a sampling time.

[0080] The data matrix is ​​subjected to wavelet transform to obtain a wavelet coefficient matrix. The energy spectrum and signal-to-noise ratio of the wavelet coefficient matrix are calculated. The resolution of the scale parameter is adaptively adjusted based on the signal-to-noise ratio. The sliding window size of the translation parameter is dynamically selected according to the energy spectrum. The data matrix is ​​decomposed and reconstructed at multiple scales based on the scale parameter and the sliding window to extract the frequency domain features.

[0081] Calculate the covariance matrix of the frequency domain features, use the singular value decomposition method to perform eigenvalue decomposition on the covariance matrix to obtain an initial feature group, select the initial features in the initial feature group whose cumulative contribution rate reaches a preset contribution threshold to construct an orthogonal projection matrix, and multiply the frequency domain features with the orthogonal projection matrix to obtain the feature vector;

[0082] The feature vectors are clustered using a fuzzy C-means clustering algorithm to obtain cluster centers. An exponential function is used to construct a membership function. Based on the membership function, the feature vectors are mapped to a fuzzy space and initial membership values ​​are calculated. Based on the initial membership values, the cluster centers are iteratively optimized using a pre-set performance index function and gradient descent. The membership values ​​are then recalculated and output based on the optimized cluster centers.

[0083] The robot's joint attitude angle data and current force state are collected. The attitude angle data includes pitch, roll, and yaw angles, and the force state includes forces and torques in the X, Y, and Z axes. Taking a six-axis robot as an example, assume the collected data at a certain moment is as follows: pitch angle 23.5 degrees, roll angle 12.8 degrees, yaw angle 45.2 degrees, X-axis force 12.5 Newtons, Y-axis force 8.3 Newtons, Z-axis force 15.6 Newtons, X-axis torque 2.3 Nm, Y-axis torque 1.5 Nm, and Z-axis torque 3.2 Nm. Normalization was performed, for example, dividing the angle data by 180, the force data by 30, and the torque data by 5, resulting in normalized values ​​of 0.131, 0.071, 0.251, 0.417, 0.277, 0.520, 0.460, 0.300, and 0.640, respectively. Data was continuously collected at a frequency of once every 100 milliseconds for 2000 time points. The normalized data was then sorted by timestamp and organized into a data matrix. Each row of the matrix corresponds to the attitude angle data and force state at a sampling time. The matrix size is 2000×9.

[0084] For the data matrix, wavelet transform was used for frequency domain analysis. The de Bessie 5 wavelet was selected as the basis function, and the data matrix was decomposed into five levels. By calculating the energy spectrum of the wavelet coefficient matrix, it was found that in the third level decomposition, the low-frequency part accounted for 85% of the energy, while the high-frequency part accounted for 15%. The signal-to-noise ratio (SNR) was calculated to be 13.5 dB. Based on the SNR value, the resolution of the scaling parameter was dynamically adjusted to 0.05. Based on the energy spectrum distribution, a sliding window size of 64 sampling points was selected. Using the adjusted scaling parameter and sliding window size, the data matrix was decomposed and reconstructed at multiple scales to extract frequency domain features. The frequency domain feature matrix contains the energy distribution of the data in different frequency bands, including energy values ​​in five frequency bands: 0-5Hz, 5-10Hz, 10-20Hz, 20-40Hz, and 40-50Hz. The matrix size is 2000×45, and the energy distribution of each of the nine original data channels across the five frequency bands constitutes 45 features.

[0085] The covariance matrix of the frequency domain feature matrix is ​​calculated, resulting in a 45×45 matrix. Singular value decomposition (SVD) is then used to decompose the covariance matrix into eigenvalues, yielding 45 eigenvalues ​​and their corresponding eigenvectors. The eigenvalues, sorted from largest to smallest, are: 152.3, 87.6, 45.2, 23.8, 15.6, 9.4, and 5.2. The contribution rate and cumulative contribution rate of each eigenvalue are calculated. The cumulative contribution rate of the first three eigenvalues ​​reaches 80.2%, exceeding the preset 75% contribution threshold. The eigenvectors corresponding to the first three eigenvalues ​​are selected to construct an orthogonal projection matrix of size 45×3. Multiplying the frequency domain feature matrix by the orthogonal projection matrix yields a dimensionality-reduced eigenvector matrix of size 2000×3.

[0086] Fuzzy C-means clustering was applied to the obtained feature vector matrix for cluster analysis. The number of clusters was set to 4, the maximum number of iterations to 100, and the termination threshold to 0.001. The initial cluster centers were randomly selected as follows: center 1 (0.23, 0.45, 0.67), center 2 (0.56, 0.32, 0.18), center 3 (0.78, 0.12, 0.34), and center 4 (0.38, 0.62, 0.51). An exponential function was used to construct the membership function, with the exponent parameter set to 2.0. For each point in the feature vector, the Euclidean distance to each cluster center was calculated, and then these distances were converted into initial membership values ​​according to the membership function. For example, for the feature vector (0.35, 0.42, 0.55), the distances to the four cluster centers are calculated to be 0.23, 0.41, 0.53 and 0.15, respectively, which are converted into membership values ​​of 0.42, 0.18, 0.12 and 0.28.

[0087] The performance metric function was pre-set as the ratio of intra-cluster compactness to inter-cluster separation, with an initial value of 0.68. Gradient descent was used to iteratively optimize the cluster centers, with a learning rate of 0.05. After 62 iterations, the cluster centers converged to: center 1 (0.25, 0.48, 0.65), center 2 (0.58, 0.30, 0.16), center 3 (0.82, 0.15, 0.35), and center 4 (0.36, 0.65, 0.54). The performance metric function value decreased to 0.32, indicating a significant improvement in cluster quality. The membership values ​​of the feature vectors were recalculated based on the optimized cluster centers; for example, the new membership values ​​of the aforementioned feature vector (0.35, 0.42, 0.55) became 0.52, 0.15, 0.08, and 0.25.

[0088] In this embodiment, attitude angle data and force state are organized into a time-series data matrix, and after normalization and sorting, the comparability and consistency of data from different sources at the same time scale are ensured. Wavelet transform is used to decompose and reconstruct the data matrix at multiple scales. Under the condition of adaptive adjustment of scale parameters and dynamic adjustment of translation parameters, frequency domain features are effectively extracted, thereby enhancing useful information and suppressing noise in complex signals. Singular value decomposition is used to reduce the dimensionality of the covariance matrix of frequency domain features, retaining the main feature information and constructing an orthogonal projection matrix, which significantly improves the computational efficiency and robustness of subsequent analysis.

[0089] In one alternative implementation,

[0090] The compensation rule is selected based on the membership value, and the weight coefficients of the compensation rule are optimized using a genetic algorithm. The optimized weight coefficients are then weighted and summed with the membership value to obtain the damping compensation value and generate damping control commands, including:

[0091] A compensation rule base is constructed based on the membership value. The compensation rule base includes multiple compensation rules. Each compensation rule includes a membership value and a corresponding initial compensation value. Based on the membership value and the initial compensation value, the rule activation degree of each compensation rule is calculated using the algebraic product T-norm operator.

[0092] Initialize the rule weight coefficients corresponding to the compensation rule, encode the rule weight coefficients as chromosomes, construct a fitness function based on control error and control increment, use the roulette wheel algorithm to select the rule weight coefficients and perform arithmetic crossover operation to obtain crossover weight coefficients, perform non-uniform mutation operation on the crossover weight coefficients and iteratively optimize according to the fitness function to obtain optimized rule weight coefficients;

[0093] The optimized rule weight coefficient is multiplied by the rule activation degree to obtain the weighted activation degree. The weighted activation degree is multiplied by the initial compensation value and summed to obtain the compensation sum value. The weighted activation degrees are summed to obtain the activation sum value. The compensation sum value is divided by the activation sum value to obtain the compensation control amount.

[0094] The compensation control quantity is compared with the preset compensation upper and lower limits to obtain the amplitude limiting compensation value. The difference between the amplitude limiting compensation value and the compensation control quantity at the previous moment is calculated to obtain the control increment. The control increment is compared with the preset increment upper and lower limits to obtain the limited control increment. The limited control increment is added to the compensation control quantity at the previous moment to obtain the damping compensation value corresponding to the current moment and a damping control command is generated.

[0095] A compensation rule base is constructed based on the membership values ​​obtained from fuzzy clustering analysis. The compensation rule base contains multiple compensation rules, each corresponding to a specific equipment state and its vibration reduction compensation strategy. In the application scenario of battlefield medical equipment, four typical states are defined: stable state, slight vibration state, moderate vibration state, and severe vibration state, each corresponding to different combinations of membership values. For example, if the membership value of the equipment at a certain moment is [0.15, 0.65, 0.18, 0.02], it indicates that the equipment is currently mainly in a slight vibration state. For these four states, initial compensation values ​​are set to 0.1, 0.3, 0.6, and 0.9, respectively. These values ​​represent the compensation intensity that the shock absorber needs to apply, ranging from [0, 1], where 0 indicates no compensation and 1 indicates maximum compensation.

[0096] Based on the membership values ​​and initial compensation values, the rule activation degree of each compensation rule is calculated using the algebraic product T-norm operator. The algebraic product T-norm operator is a commonly used fuzzy "AND" operation in fuzzy logic, and the calculation method is direct multiplication. For each rule, the membership value of the corresponding state is multiplied by the initial compensation value of that rule to obtain the rule activation degree. For example, for the above membership values ​​[0.15, 0.65, 0.18, 0.02] and initial compensation values ​​[0.1, 0.3, 0.6, 0.9], the calculated rule activation degree is [0.015, 0.195, 0.108, 0.018]. The rule activation degree reflects the degree of influence of each rule in the current state.

[0097] Initialize the rule weight coefficients corresponding to the compensation rules. These weight coefficients are used to adjust the importance of each rule. Initial values ​​can be set to be equal, such as [0.25, 0.25, 0.25, 0.25], indicating that the four rules have the same importance. To achieve better control, these weight coefficients need to be optimized.

[0098] The rule weight coefficients are encoded into chromosomes, and a genetic algorithm is used for optimization. During the encoding process, each weight value is converted into a binary or real number code. For example, using real number encoding, the weight values ​​are directly used to construct the chromosome, such as [0.25, 0.25, 0.25, 0.25]. Multiple such chromosomes can be randomly generated in the initial population, with a population size of 50.

[0099] A fitness function is constructed based on control error and control increment. Control error is defined as the difference between the actual and target attitudes of the device, and control increment is defined as the change in control output between two consecutive control operations. The fitness function is designed as a weighted sum of the squares of the control error and the squares of the control increment, with the weighting coefficients adjustable according to actual needs. For example, increasing the weight of control error increases the emphasis on control precision, while increasing the weight of control increment increases the emphasis on control stability. A smaller fitness function value indicates better control performance.

[0100] A roulette wheel selection algorithm is used to perform arithmetic crossover operations based on the selection rule weight coefficients. The roulette wheel algorithm assigns selection probabilities based on the chromosome's fitness value; the smaller the fitness value, the higher the probability of selection. After selecting two parent chromosomes, arithmetic crossover is performed to generate two offspring chromosomes. The arithmetic crossover is performed by weighted averaging of genes at corresponding positions in the parent chromosomes. For example, for two parent chromosomes [0.2, 0.3, 0.3, 0.2] and [0.3, 0.2, 0.2, 0.3], with a crossover coefficient of 0.7, the generated offspring chromosomes are [0.23, 0.27, 0.27, 0.23] and [0.27, 0.23, 0.23, 0.27].

[0101] Non-uniform mutation is performed on the cross-weight coefficients. Non-uniform mutation is a mutation method in which the mutation amplitude gradually decreases with the number of iterations. The mutation probability is set to 0.1, meaning that each gene has a 10% probability of mutating. When mutation occurs, the gene value changes randomly within a certain range, and the magnitude of the change decreases with the number of iterations. For example, in early iterations, mutation may cause the gene value to change by ±0.1; while in later iterations, the mutation amplitude may decrease to ±0.01.

[0102] The mutated chromosomes are evaluated based on the fitness function, and individuals with lower fitness values ​​are retained for the next iteration. A maximum of 100 iterations is set, or convergence is considered achieved when the optimal fitness value changes by less than 0.001 over 10 consecutive generations, at which point iteration stops. After iterative optimization, optimized rule weight coefficients are obtained, for example, [0.15, 0.35, 0.40, 0.10], indicating that in the current application scenario, the rule weight is highest for moderate vibration and lowest for severe vibration.

[0103] The optimized rule weight coefficient is multiplied by the rule activation level to obtain the weighted activation level. For example, for rule activation levels [0.015, 0.195, 0.108, 0.018] and optimized rule weight coefficients [0.15, 0.35, 0.40, 0.10], the calculated weighted activation level is [0.00225, 0.06825, 0.0432, 0.0018]. The weighted activation level comprehensively considers both the activation degree and importance of the rule.

[0104] Multiply the weighted activation values ​​by the initial compensation values ​​and sum them to obtain the compensation sum; sum the weighted activation values ​​to obtain the activation sum. In the example above, the compensation sum is 0.00225×0.1+0.06825×0.3+0.0432×0.6+0.0018×0.9=0.0489; the activation sum is 0.00225+0.06825+0.0432+0.0018=0.1155. Divide the compensation sum by the activation sum to obtain the compensation control quantity. In the previous example, the compensation control quantity is 0.0489÷0.1155=0.4234, which represents the theoretical compensation intensity calculated based on the current equipment state.

[0105] The compensation control amount is compared with the preset upper and lower compensation limits to obtain the limiting compensation value. For example, if the upper limit of compensation is set to 0.8 and the lower limit to 0.1, then for the calculated compensation control amount of 0.4234, the value after limiting will still be 0.4234, because this value is within the upper and lower limits.

[0106] The control increment is obtained by calculating the difference between the limiting compensation value and the compensation control value at the previous moment. For example, if the compensation control value at the previous moment was 0.3825, then the control increment is 0.4234 - 0.3825 = 0.0409. The control increment reflects the rate of change of the compensation intensity.

[0107] The control increment is compared with the preset upper and lower limits to obtain the limited control increment. The upper limit of the increment is set to 0.05 and the lower limit to -0.05. For the calculated control increment of 0.0409, the limited value is still 0.0409 because this value is within the range of the upper and lower limits of the increment.

[0108] The limited control increment is added to the compensation control value from the previous moment to obtain the damping compensation value for the current moment. In the example above, the damping compensation value for the current moment is 0.3825 + 0.0409 = 0.4234. Based on this compensation value, a damping control command is generated and sent to the damping actuator.

[0109] The damping control command includes information such as compensation value and execution time. After receiving the command, the damping actuator adjusts the damper parameters, such as damping coefficient and stiffness, according to the compensation value. For example, a compensation value of 0.4234 may correspond to an adjustment that increases the damping coefficient by 40% and decreases the stiffness by 20%.

[0110] In this embodiment, a compensation rule base is established based on membership values ​​and initial compensation values. The activation degree of the rules is calculated using the algebraic product T-norm operator, which can flexibly reflect the control requirements under different working conditions. The fitness function is used to evaluate the quality of the rule weights. The weight coefficients are iteratively updated by combining operations such as roulette wheel selection, arithmetic crossover, and non-uniform mutation. This effectively avoids the problem that traditional compensation methods are prone to getting stuck in local optima or response lag in complex environments. The compensation control quantity is calculated by weighted summation of weighted activation degree and initial compensation value. It is further processed by amplitude limiting and incremental limitation to ensure that the output compensation value can maintain the sensitivity of dynamic response and avoid system instability caused by over-compensation. This realizes adaptive optimization and dynamic amplitude limiting adjustment of compensation control quantity, which can improve the vibration reduction performance of battlefield medical equipment and enhance the reliability of operation in complex battlefield environments.

[0111] In one alternative implementation,

[0112] The vibration reduction compensation value is subjected to discrete Fourier transform to obtain spectral components. A state feedback matrix is ​​constructed based on the spectral components and convolved with a preset feature compensation matrix to obtain a control gain matrix, including:

[0113] The vibration reduction compensation value is subjected to discrete Fourier transform to obtain the spectral components. The spectral components are decomposed into real and imaginary parts. The amplitude spectrum is calculated based on the square root of the sum of the squares of the real and imaginary parts. The phase spectrum is calculated based on the arctangent function of the quotient of the imaginary and real parts.

[0114] The amplitude spectrum and the phase spectrum are arranged in order of frequency components to construct a state feedback matrix. The state feedback matrix includes an amplitude spectrum sequence in the first row and a phase spectrum sequence in the second row. Multiple preset frequency weight coefficients are multiplied by the amplitude spectrum and phase spectrum of the corresponding frequency components in the state feedback matrix to obtain a weighted state feedback matrix.

[0115] A feature compensation matrix is ​​constructed, which includes multiple amplitude compensation coefficients in the first row and multiple phase compensation coefficients in the second row. The weighted state feedback matrix and the feature compensation matrix are subjected to a two-dimensional convolution operation. The control gain matrix is ​​calculated based on the sum of the products of the elements in the weighted state feedback matrix and the compensation coefficients at the corresponding positions in the feature compensation matrix.

[0116] The vibration reduction compensation values ​​are subjected to Discrete Fourier Transform (DFT) to obtain spectral components. DFT is a method for converting time-domain signals into frequency-domain representations. In practical implementation, Fast Fourier Transform (FFT) algorithms are used to improve computational efficiency. For vibration reduction control of battlefield medical equipment, the vibration reduction compensation values ​​of the most recent 64 sampling points are selected as the input sequence. For example, the vibration reduction compensation value sequence within a certain time period may be [0.42, 0.43, 0.45, 0.44, 0.46, ..., 0.38]. After DFT, 64 complex spectral components are obtained.

[0117] The spectral components are decomposed into real and imaginary parts. Each spectral component is a complex number containing both a real and an imaginary part. For example, the spectral component at the first frequency point might be 3.5 + 2.1i, where 3.5 is the real part and 2.1 is the imaginary part; the spectral component at the second frequency point might be 2.8 - 1.5i, where 2.8 is the real part and -1.5 is the imaginary part.

[0118] The amplitude spectrum is calculated using the square root of the sum of the squares of the real and imaginary parts. The amplitude spectrum represents the intensity of each frequency component. Taking the first frequency point as an example, the amplitude is sqrt(3.5^2 + 2.1^2) = sqrt(12.25 + 4.41) = sqrt(16.66) = 4.08, and the amplitude at the second frequency point is sqrt(2.8^2 + (-1.5)^2) = sqrt(7.84 + 2.25) = sqrt(10.09) = 3.18. The amplitude spectrum reflects the energy distribution of each frequency component in the original signal.

[0119] The phase spectrum is calculated using the arctangent function of the quotient of the imaginary and real parts. The phase spectrum represents the phase angle of each frequency component. Taking the first frequency point as an example, its phase is arctan(2.1 / 3.5) = arctan(0.6) = 0.54 radians ≈ 31 degrees. For the second frequency point, its phase is arctan((-1.5) / 2.8) = arctan(-0.54) = -0.49 radians ≈ -28 degrees. The phase spectrum reflects the relative time relationship of each frequency component. During the calculation, special cases such as zero real part or the signs of the real and imaginary parts need to be considered. Using the full arctan2 function ensures the accuracy of the phase calculation.

[0120] The amplitude and phase spectra are arranged in order of frequency components to construct the state feedback matrix. The state feedback matrix is ​​a 2×N matrix, where N is the number of frequency components. The first row is the amplitude spectrum sequence, and the second row is the phase spectrum sequence. For example, the first few columns of the state feedback matrix are: [[4.08, 3.18, 2.56, 1.95, ...], [0.54, -0.49, 1.25, -0.85, ...]]. Due to the symmetry of the Discrete Fourier Transform, usually only the first N / 2+1 frequency components are used, corresponding to the frequency range from 0 to half the sampling frequency.

[0121] The weighted state feedback matrix is ​​obtained by multiplying the preset frequency weighting coefficients by the amplitude and phase spectra of the corresponding frequency components in the state feedback matrix. The frequency weighting coefficients reflect the importance of different frequency components to the control effect. In battlefield medical equipment applications, the frequency range of 0.5Hz-10Hz is typically the focus, as frequency components within this range have a significant impact on equipment stability. Based on experience and experimental data, frequency weighting coefficients are set; for example, for four frequency points, they might be set to [1.0, 0.8, 0.6, 0.4], indicating that low-frequency components have higher weights and high-frequency components have lower weights. After multiplying by the weighting coefficients, the first few columns of the weighted state feedback matrix may become: [[4.08×1.0, 3.18×0.8, 2.56×0.6, 1.95×0.4, ...], [0.54×1.0, -0.49×0.8, 1.25×0.6, -0.85×0.4, ...]] = [[4.08, 2.54, 1.54, 0.78, ...], [0.54, -0.39, 0.75, -0.34, ...]].

[0122] Construct a characteristic compensation matrix. The characteristic compensation matrix is ​​also a 2×N matrix, with the first row containing multiple amplitude compensation coefficients and the second row containing multiple phase compensation coefficients. These compensation coefficients are pre-set based on the device's dynamic characteristics and control objectives. For example, for the first four frequency points, the characteristic compensation matrix might be: [[0.9, 1.2, 1.5, 0.8, ...], [0.1, -0.2, 0.3, -0.1, ...]]. An amplitude compensation coefficient greater than 1 indicates enhancement of that frequency component, while a coefficient less than 1 indicates suppression of that frequency component; the phase compensation coefficients are used to adjust the phase relationship, affecting the timing characteristics of the control.

[0123] The weighted state feedback matrix and the feature compensation matrix are subjected to a two-dimensional convolution operation. Two-dimensional convolution is a common operation in image and signal processing, capable of extracting local features of signals and enhancing pattern recognition capabilities. In this embodiment, a sliding window method is used for convolution calculation, with a window size of 3×2 and a stride of 1. For each 3×2 submatrix in the weighted state feedback matrix, its elements are multiplied by the corresponding compensation coefficients in the feature compensation matrix, and then summed to obtain a convolution result. For example, the first three columns of the weighted state feedback matrix are convolved with the first three columns of the feature compensation matrix: 4.08×0.9+2.54×1.2+1.54×1.5+0.54×0.1+(-0.39)×(-0.2)+0.75×0.3=3.67+3.05+2.31+0.05+0.08+0.23=9.39. This value is used as an element of the control gain matrix. By using a sliding window, the convolution results at different positions are calculated to form a complete control gain matrix.

[0124] The control gain matrix is ​​calculated by summing the products of the elements in the weighted state feedback matrix and the corresponding compensation coefficients in the feature compensation matrix. The dimension of the control gain matrix depends on the parameters of the convolution operation and the size of the input matrix. For a 2×N input matrix and a 3×2 convolution window, the resulting control gain matrix is ​​2×(N-2). For example, for N=32, the control gain matrix is ​​2×30. Each element of the control gain matrix corresponds to a specific combination of frequency domain features used to adjust the control strategy.

[0125] The control gain matrix is ​​used to optimize the parameters of the vibration damping controller. For example, the elements in the first row of the control gain matrix can be used to adjust the proportional gain, and the elements in the second row can be used to adjust the integral gain. This control gain optimization method based on frequency domain characteristics can adopt different control strategies for interference at different frequencies, improving the accuracy and robustness of the control.

[0126] In this embodiment, the compensation value is mapped from the time domain to the frequency domain using the Discrete Fourier Transform, and the amplitude spectrum and phase spectrum are extracted. This comprehensively reflects the energy distribution and phase characteristics of the vibration reduction signal at different frequency components, improving the ability to describe the dynamic characteristics of the signal. By constructing the amplitude spectrum and phase spectrum into a state feedback matrix and introducing preset frequency weighting coefficients for weighting, key information can be adaptively highlighted and invalid components suppressed for different frequency characteristics, thereby enhancing the pertinence and sensitivity of feedback control. Through the combination of amplitude compensation coefficient and phase compensation coefficient, a two-dimensional compensation mechanism is effectively provided. By performing two-dimensional convolution between the weighted state feedback matrix and the feature compensation matrix, not only is the deep fusion of spectral features and compensation parameters achieved, but the distribution of control gain can also be dynamically optimized through the superposition effect of the convolution results, effectively improving the stability and reliability of battlefield medical equipment in complex vibration environments.

[0127] In one alternative implementation,

[0128] Based on the control gain matrix, the vibration reduction compensation value is fitted using least squares to calculate the current control parameters and attitude adjustment parameters of the shock absorber. Controlling the shock absorber and Mecanum wheel of the field medical equipment according to the current control parameters and attitude adjustment parameters includes:

[0129] Obtain the control gain matrix, multiply the control gain matrix by a preset first control coefficient and add a bias coefficient to obtain the shock absorber current control quantity, calculate the sum of squares of the difference between the shock absorber current control quantity and the preset expected current value and construct a first objective function, and under the constraints that the current change rate does not exceed the preset change rate upper limit and the current amplitude does not exceed the preset amplitude limit, minimize the function value corresponding to the first objective function to solve for the first control coefficient and bias coefficient as current control parameters;

[0130] The attitude control input is obtained by multiplying the control gain matrix by a pre-set first attitude coefficient and superimposing the attitude error by a second attitude coefficient. A Lyapunov function containing the attitude error and the pre-acquired angular velocity is constructed. The derivative of the Lyapunov function is weighted with the sum of the squares of the attitude control input to construct a second objective function. The function value corresponding to the second objective function is minimized to obtain the first attitude coefficient and the second attitude coefficient as attitude adjustment parameters.

[0131] The product of the current control parameter and the attitude error is added to the shock absorber current control quantity to obtain the actual control current of the shock absorber. The product of the attitude adjustment parameter and the difference between the actual control current of the shock absorber and the shock absorber current control quantity is added to the attitude control input to obtain the wheel system control command. The shock absorber and Mecanum wheel are controlled in a coordinated manner according to the actual control current of the shock absorber and the wheel system control command.

[0132] Obtain the control gain matrix, calculated using the aforementioned frequency domain analysis method. This matrix contains frequency domain characteristic information about the status of battlefield medical equipment. The control gain matrix is ​​typically a 2×30 matrix, where the first row corresponds to the amplitude characteristics and the second row corresponds to the phase characteristics. For example, some elements of the control gain matrix might be [[9.39, 8.25, 7.16, ...], [3.45, 2.87, 2.32, ...]]. These elements reflect the importance and influence of different frequency characteristics.

[0133] The damper current control value is obtained by multiplying the control gain matrix by a pre-set first control coefficient and then adding the bias coefficient. The first control coefficient is in matrix form and has the same dimension as the control gain matrix. It is used to adjust the influence of various frequency domain characteristics on the current control. The initial value can be set to an all-1 matrix, that is, all elements are 1. The bias coefficient is a scalar used to provide the base current value, and the initial value can be set to 0.5 amperes. For example, if the first few elements of the first control coefficient matrix are [[0.02, 0.03, 0.02, ...], [0.01, 0.01, 0.01, ...]], then the calculation process of the shock absorber current control quantity is: 9.39×0.02+8.25×0.03+7.16×0.02+...+3.45×0.01+2.87×0.01+2.32×0.01+...+0.5=0.1878+0.2475+0.1432+...+0.0345+0.0287+0.0232+...+0.5≈1.8 Amperes.

[0134] Calculate the sum of squares of the differences between the damper current control quantity and the preset desired current value, and construct the first objective function. The desired current value is determined based on the balance between damping effect and energy consumption, and is usually set to 1.5 amperes. The sum of squares reflects the degree of deviation between the current control quantity and the desired value, and is calculated as the square of (current control quantity - desired current value), with the sum of squares being (1.8 - 1.5). 2 =0.09. The first objective function is defined as the weighted sum of the squared differences and a regularization term. The regularization term is used to prevent the control parameters from being too large and can be expressed as 0.01 times the sum of the squares of the elements of the first control coefficient.

[0135] Under the constraints that the rate of change of current does not exceed a preset upper limit and the current amplitude does not exceed a preset amplitude limit, the function value corresponding to the first objective function is minimized, and the first control coefficient and bias coefficient are solved as current control parameters. The upper limit of the rate of change is set to 0.5 amperes / second, indicating that the current value changes by a maximum of 0.5 amperes per second; the amplitude limit is set to 3 amperes, indicating that the current does not exceed 3 amperes. The Lagrange multiplier method is used to handle the constrained optimization problem, and the constraints are introduced into the objective function. The gradient descent method is used to iteratively solve the problem, updating the first control coefficient and bias coefficient until convergence. For example, after optimization, the first few elements of the first control coefficient matrix may become [[0.018, 0.027, 0.019, ...], [0.009, 0.011, 0.012, ...]], and the bias coefficient becomes 0.48 amperes.

[0136] The attitude control input is obtained by multiplying the control gain matrix by a pre-set first attitude coefficient and then adding the attitude error multiplied by a second attitude coefficient. The attitude error is defined as the difference between the current attitude and the target attitude, including three components: roll angle error, pitch angle error, and yaw angle error. For example, if the current attitude is [5.2°, -2.3°, 78.1°] and the target attitude is [0°, 0°, 90°], then the attitude error is [5.2°, -2.3°, -11.9°]. The first attitude coefficient is in matrix form, with the same dimensions as the control gain matrix, and its initial value can be set to an all-0.1 matrix. The second attitude coefficient is a 3×3 diagonal matrix, with diagonal elements corresponding to the control gains for roll, pitch, and yaw angles, respectively, and its initial value can be set to [0.5, 0.5, 0.2]. The calculation process of the attitude control input combines the frequency domain characteristics and the influence of the attitude error, forming a three-dimensional vector, each corresponding to the control quantity of the three rotation axes.

[0137] A Lyapunov function is constructed, incorporating attitude errors and pre-acquired angular velocities. The Lyapunov function is a function used to prove system stability; in this embodiment, it is defined as a weighted sum of the squares of the attitude errors and the squares of the angular velocities. Angular velocity data is acquired through an inertial measurement unit, representing the rotational rate about three axes. For example, the measured angular velocity at a certain moment is [0.03, 0.01, -0.02] radians / second. The Lyapunov function value reflects the total energy of the system, and this value should gradually decrease as the system stabilizes.

[0138] A second objective function is constructed by weighting the derivative of the Lyapunov function with the sum of squares of the attitude control inputs. The derivative of the Lyapunov function represents the rate of change of the system's energy; a negative value is desired, indicating a gradual decrease in system energy. The derivative calculation involves attitude error, angular velocity, and their derivatives, and is approximated using numerical difference methods. The second objective function is defined as a weighted combination of the derivative of the Lyapunov function and the sum of squares of the attitude control inputs, aiming to minimize (make negative) the derivative of the Lyapunov function while ensuring the control inputs are not excessive.

[0139] The function value corresponding to the second objective function is minimized, and the first and second attitude coefficients are obtained as attitude adjustment parameters. Simulated annealing is used for global optimization, with an initial temperature of 100°C, a final temperature of 0.1°C, and a cooling coefficient of 0.95. At each temperature, the current parameter values ​​are randomly perturbed, and new objective function values ​​are calculated. The Metropolis criterion is used to determine whether to accept the new parameters. As the temperature gradually decreases, the perturbation amplitude decreases, eventually converging to a better solution. For example, the first few elements of the optimized first attitude coefficient matrix may be [[0.12, 0.09, 0.08, ...], [0.05, 0.06, 0.04, ...]], and the second attitude coefficient becomes [0.6, 0.55, 0.18], which can provide effective attitude control while ensuring system stability.

[0140] The actual control current of the shock absorber is obtained by superimposing the product of the current control parameter and the attitude error onto the shock absorber current control value. The current control parameter includes a first control coefficient and a bias coefficient, and its product with the attitude error represents the dynamic adjustment of the current based on the attitude state. For example, if the roll angle error is 5.2°, the corresponding current control parameter is 0.02, then the adjustment amount is 5.2 × 0.02 = 0.104 amperes. After superimposing the adjustment amount, the actual control current of the shock absorber becomes 1.8 + 0.104 = 1.904 amperes.

[0141] The attitude adjustment parameters are multiplied by the product of the actual control current of the shock absorber and the difference between the actual control current and the control current of the shock absorber, and then added to the attitude control input to obtain the wheel train control command. The attitude adjustment parameters include a first attitude coefficient and a second attitude coefficient; their product with the current difference represents the feedback adjustment of the attitude control based on the damping state. For example, if the current difference is 0.104 amperes, and the corresponding attitude adjustment parameter is 0.3, then the adjustment amount is 0.104 × 0.3 = 0.0312. After adding the adjustment amount, the corresponding component of the attitude control input increases by 0.0312, forming the final wheel train control command.

[0142] The shock absorber and Mecanum wheel are controlled collaboratively based on the actual control current of the shock absorber and the wheel system control commands. The shock absorber control current is sent to the magnetorheological shock absorber via a current driver to adjust the damping characteristics of the shock absorber. The Mecanum wheel is an omnidirectional wheel, capable of omnidirectional movement and stationary rotation. The wheel system control commands are sent to the four drive motors of the Mecanum wheel via motor drivers to control the speed and direction of each wheel. For example, when the roll angle needs to be adjusted, it is achieved by differentially controlling the speed of the left and right wheels; when the pitch angle needs to be adjusted, it is achieved by differentially controlling the speed of the front and rear wheels; when the yaw angle needs to be adjusted, omnidirectional rotation is achieved by coordinating the speed of all four wheels.

[0143] In this embodiment, the control gain matrix is ​​combined with preset coefficients, and the shock absorber current control quantity is obtained after superimposing the bias coefficient. Then, by constructing an objective function based on the desired current value, and optimizing the first control coefficient and bias coefficient under the condition of limited current change rate and amplitude, the current control parameters can achieve high-precision adjustment of the shock absorber current while ensuring safety and stability, thereby improving the damping performance and the smoothness of the current response. A Lyapunov function is constructed using attitude error and angular velocity information, and an objective function is established based on the derivative of this function and the weighted sum of squares of the attitude control input. By minimizing the objective function to solve the attitude adjustment parameters, the stability and convergence of the attitude control process are effectively guaranteed. It can not only correct the attitude error in real time, but also improve the robustness and dynamic response capability of the attitude adjustment process. By superimposing the current control parameters with the attitude error to obtain the actual control current, and combining it with the attitude adjustment parameters to correct the attitude control input, a coordinated control command for the shock absorber and Mecanum wheel is generated simultaneously. This can take into account both the vehicle's damping effect and attitude stability in complex terrain and strong vibration environments.

[0144] In one alternative implementation,

[0145] Real-time monitoring of the vibration reduction effect; if the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude, including:

[0146] Vibration acceleration signals are collected by an accelerometer installed on the shock absorber, and vibration displacement signals are collected by a displacement sensor. The root mean square value of the vibration acceleration signal and the maximum amplitude of the vibration displacement signal are used as evaluation indicators of the shock absorption effect. When the evaluation indicators exceed the preset shock absorption threshold, parameter optimization is triggered.

[0147] The current attitude is obtained, and the sum of squares of the difference between the current control value of the shock absorber and the expected current value is fitted by least squares to obtain the updated current control parameters. The Lyapunov function constructed by the attitude error and angular velocity is optimized to obtain the updated attitude adjustment parameters.

[0148] The updated current control parameters and the updated attitude adjustment parameters are applied to the coordinated control of the shock absorber and the Mecanum wheel, respectively, until the evaluation index is less than the preset shock absorption threshold.

[0149] Vibration acceleration signals are collected by an accelerometer mounted on the shock absorber. The accelerometer is installed between the fixed and movable ends of the shock absorber, with a sampling frequency of 1000Hz, a range of ±16g, and a resolution of 0.01g. During the operation of the field medical equipment, the accelerometer collects vibration acceleration signals in real time, recording the vibration experienced by the shock absorber. For example, the vibration acceleration data collected at a certain moment might be [-0.25g, 0.35g, -0.18g, 0.22g, ...], reflecting the changes in acceleration experienced by the shock absorber at different times.

[0150] Vibration displacement signals are acquired using a displacement sensor. The displacement sensor, a linear variable differential transformer (LVDT) type, is mounted on the piston rod of the shock absorber to measure changes in piston displacement. The displacement sensor has a range of ±50 mm, a resolution of 0.1 mm, and a sampling frequency of 1000 Hz, consistent with the accelerometer. For example, the vibration displacement data acquired at a certain moment might be [2.5 mm, 3.2 mm, 1.8 mm, -0.5 mm, ...], reflecting the compression and extension states of the shock absorber at different times.

[0151] The root mean square (RMS) value of the vibration acceleration signal and the maximum amplitude of the vibration displacement signal are used as evaluation indicators for vibration reduction effect. The RMS value of the vibration acceleration signal is calculated by first squaring each acceleration sampling point, then summing the results, dividing by the number of sampling points, and finally taking the square root. For example, for the aforementioned acceleration data [-0.25g, 0.35g, -0.18g, 0.22g, ...], the RMS value is approximately 0.26g. The maximum amplitude of the vibration displacement signal is calculated by taking the maximum absolute value of the displacement data. For the aforementioned displacement data [2.5mm, 3.2mm, 1.8mm, -0.5mm, ...], the maximum amplitude is 3.2mm.

[0152] Parameter optimization is triggered when the evaluation indicators exceed the preset vibration reduction thresholds. The preset vibration reduction thresholds are determined based on the specific application scenario and vibration reduction requirements of the field medical equipment, and are typically set to a root mean square acceleration value not exceeding 0.3g and a maximum displacement amplitude not exceeding 5mm. The parameter optimization process is triggered when the detected root mean square acceleration value exceeds 0.3g or the maximum displacement amplitude exceeds 5mm. For example, if the detected root mean square acceleration value is 0.35g, exceeding the threshold of 0.3g, the parameter optimization program is automatically started.

[0153] The current attitude is obtained, and the sum of squared differences between the shock absorber current control value and the desired current value is fitted using least squares to obtain the updated current control parameters. The current attitude is obtained through an attitude sensor, including roll angle, pitch angle, and yaw angle. For example, the attitude data at a certain moment might be [2.5°, -1.8°, 85.3°]. Based on the current attitude data, the current control parameters (first control coefficient and bias coefficient) are used as independent variables, and the sum of squared current differences is used as the dependent variable. The most recent 100 sets of data are collected, including the current control parameters and the corresponding sum of squared current differences under different attitudes, and the parameters are fitted using the least squares method.

[0154] For example, a linear relationship exists between the sum of squares of the current difference and the elements of the first control coefficient and the bias coefficient. The parameters are updated iteratively using gradient descent after calculating their gradients until convergence. For instance, in the original current control parameters, the first few elements of the first control coefficient matrix are [[0.018, 0.027, 0.019, ...], [0.009, 0.011, 0.012, ...]], and the bias coefficient is 0.48 amperes. After least squares fitting, the updated parameters might become: the first few elements of the first control coefficient matrix are [[0.022, 0.025, 0.020, ...], [0.010, 0.013, 0.011, ...]], and the bias coefficient is 0.52 amperes. These updated parameters allow the current control quantity to better adapt to the current attitude state, reducing the difference from the desired current value.

[0155] The Lyapunov function, constructed from attitude error and angular velocity, is optimized to obtain updated attitude adjustment parameters. The Lyapunov function is defined as a weighted combination of the sum of squared attitude errors and the sum of squared angular velocities. Its derivative should be negative, indicating that the system energy gradually decreases and the system tends to stabilize. Based on this principle, an optimization problem is constructed: to find the optimal attitude adjustment parameters (first attitude coefficient and second attitude coefficient) that minimize the derivative of the Lyapunov function while controlling the input to be moderate.

[0156] The particle swarm optimization algorithm is used to solve this optimization problem. 50 particles are initialized, each representing a set of candidate attitude adjustment parameters. The number of iterations is set to 100, the inertia weight to 0.8, the individual learning factor to 1.5, and the swarm learning factor to 2.0. In each iteration, the velocity and position of each particle are updated, the objective function value corresponding to each particle is evaluated, and the individual optimal position and the global optimal position are updated. For example, in the original attitude adjustment parameters, the first few elements of the first attitude coefficient matrix are [[0.12, 0.09, 0.08, ...], [0.05, 0.06, 0.04, ...]], and the second attitude coefficient is [0.6, 0.55, 0.18]. After particle swarm optimization, the updated parameters may become: the first few elements of the first attitude coefficient matrix are [[0.14, 0.10, 0.09, ...], [0.06, 0.05, 0.05, ...]], and the second attitude coefficient is [0.65, 0.58, 0.20]. These updated parameters can improve the stability and response speed of attitude control.

[0157] The updated current control parameters and updated attitude adjustment parameters are applied to the coordinated control of the shock absorber and Mecanum wheel, respectively, until the evaluation index is less than the preset damping threshold. The updated current control parameters are then substituted into the current control formula to calculate the new actual control current of the shock absorber; the updated attitude adjustment parameters are substituted into the attitude control formula to calculate the new wheel system control command. For example, using the updated parameters, the actual control current of the shock absorber may become 2.1 amperes, and the three components of the wheel system control command may become [0.35, -0.28, 0.15], corresponding to the control inputs of the roll axis, pitch axis, and yaw axis, respectively.

[0158] The control system sends the actual control current of the shock absorber to the current driver of the magnetorheological shock absorber to adjust the damping characteristics of the shock absorber; it also sends the wheel system control commands to the motor driver of the Mecanum wheel to control the speed and direction of each wheel. It continuously monitors vibration acceleration and displacement signals, calculates evaluation indicators, and checks whether the vibration reduction requirements are met. If the evaluation indicators are lower than the preset vibration reduction threshold, the current parameter settings are maintained; if the evaluation indicators still exceed the threshold, the parameter optimization process continues.

[0159] In this embodiment, acceleration and displacement sensors installed on the shock absorber are used to collect vibration acceleration and displacement signals, respectively. The root mean square value of the acceleration signal and the maximum amplitude of the displacement signal are used as evaluation indicators of the shock absorption effect, which can intuitively reflect the shock absorption performance of the equipment in terms of vibration intensity and displacement amplitude. The least squares fitting method is used to model and correct the difference between the current control quantity and the expected value to obtain updated current control parameters, thereby ensuring the accuracy and convergence speed of the shock absorption current adjustment. The updated current control parameters and attitude adjustment parameters are applied to the coordinated control of the shock absorber and the Mecanum wheel, so that the system can continuously adjust to the optimal state of shock absorption effect until the evaluation index drops below the threshold.

[0160] A second aspect of the present invention provides an intelligent shock absorption and attitude control system for battlefield medical equipment, comprising:

[0161] The first unit is used to acquire the posture sensor data of the battlefield medical equipment and calculate the posture angle data, collect the pressure data and displacement data on the shock absorber and determine the current stress state.

[0162] The second unit is used to construct a data matrix by arranging the attitude angle data and the current force state in a time sequence, perform wavelet transform on the data matrix to obtain frequency domain features, perform principal component analysis on the frequency domain features to extract feature vectors, map the feature vectors in the fuzzy space based on an adaptive fuzzy control algorithm to obtain membership values, select compensation rules according to the membership values ​​and optimize the weight coefficients of the compensation rules through a genetic algorithm, and perform a weighted sum of the obtained optimized weight coefficients and membership values ​​to obtain vibration reduction compensation values ​​and generate vibration reduction control commands.

[0163] The third unit is used to perform discrete Fourier transform on the vibration reduction compensation value to obtain spectral components, construct a state feedback matrix based on the spectral components and perform convolution operation with a preset feature compensation matrix to obtain a control gain matrix, perform least squares fitting on the vibration reduction compensation value based on the control gain matrix, calculate the current control parameters and attitude adjustment parameters of the vibration damper respectively, and control the vibration damper and Mecanum wheel of the battlefield medical equipment according to the current control parameters and attitude adjustment parameters.

[0164] The fourth unit is used to monitor the vibration reduction effect in real time. If the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude.

[0165] A third aspect of the present invention provides an electronic device, comprising:

[0166] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0167] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0168] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent shock absorption and attitude control of battlefield medical equipment, characterized in that, include: Acquire the position sensor data of the battlefield medical equipment and calculate the attitude angle data; collect the pressure data and displacement data on the shock absorber and determine the current stress state. The attitude angle data and the current force state are arranged in time sequence to construct a data matrix. Wavelet transform is performed on the data matrix to obtain frequency domain features. Principal component analysis is performed on the frequency domain features to extract feature vectors. The feature vectors are mapped in the fuzzy space based on an adaptive fuzzy control algorithm to obtain membership values. Compensation rules are selected according to the membership values, and the rule weight coefficients of the compensation rules are optimized by a genetic algorithm. The optimized weight coefficients and the membership values ​​are weighted and summed to obtain the vibration reduction compensation value and generate vibration reduction control commands. The vibration reduction compensation value is subjected to discrete Fourier transform to obtain spectral components. A state feedback matrix is ​​constructed based on the spectral components and convolved with a preset feature compensation matrix to obtain a control gain matrix. The vibration reduction compensation value is fitted with least squares based on the control gain matrix to calculate the current control parameters and attitude adjustment parameters of the vibration damper. The vibration damper and Mecanum wheel of the field medical equipment are controlled according to the current control parameters and attitude adjustment parameters. The vibration reduction effect is monitored in real time. If the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude. Based on the control gain matrix, the vibration reduction compensation value is fitted using least squares to calculate the current control parameters and attitude adjustment parameters of the shock absorber. Controlling the shock absorber and Mecanum wheel of the field medical equipment according to the current control parameters and attitude adjustment parameters includes: Obtain the control gain matrix, multiply the control gain matrix by a preset first control coefficient and add a bias coefficient to obtain the shock absorber current control quantity, calculate the sum of squares of the difference between the shock absorber current control quantity and the preset expected current value and construct a first objective function, and under the constraints that the current change rate does not exceed the preset change rate upper limit and the current amplitude does not exceed the preset amplitude limit, minimize the function value corresponding to the first objective function to solve for the first control coefficient and bias coefficient as current control parameters; The attitude control input is obtained by multiplying the control gain matrix by a pre-set first attitude coefficient and superimposing the attitude error by a second attitude coefficient. A Lyapunov function containing the attitude error and the pre-acquired angular velocity is constructed. The derivative of the Lyapunov function is weighted with the sum of the squares of the attitude control input to construct a second objective function. The function value corresponding to the second objective function is minimized to obtain the first attitude coefficient and the second attitude coefficient as attitude adjustment parameters. The product of the current control parameter and the attitude error is added to the shock absorber current control quantity to obtain the actual control current of the shock absorber. The product of the attitude adjustment parameter and the difference between the actual control current of the shock absorber and the shock absorber current control quantity is added to the attitude control input to obtain the wheel system control command. The shock absorber and Mecanum wheel are controlled in a coordinated manner according to the actual control current of the shock absorber and the wheel system control command.

2. The method according to claim 1, characterized in that, Acquiring posture sensor data from the battlefield medical equipment and calculating attitude angle data, collecting pressure and displacement data from the shock absorber and determining the current stress state includes: The field medical equipment's posture sensor data and shock absorber sensing data are collected, wherein the posture sensor data includes acceleration data, angular velocity data, and geomagnetic data, and the shock absorber sensing data includes pressure data and displacement data. The acceleration data, angular velocity data, and geomagnetic data are combined in time sequence, and the predicted state is calculated through a pre-set state equation. The predicted state is then corrected by combining the actual observation values ​​to obtain the attitude angle data of the battlefield medical equipment. The pressure data and displacement data are paired by timestamp, and the current stress state of the shock absorber is determined by the ratio of the pressure change to the displacement change at adjacent times.

3. The method according to claim 1, characterized in that, A data matrix is ​​constructed by arranging the attitude angle data and the current force state in time sequence. Wavelet transform is performed on the data matrix to obtain frequency domain features. Principal component analysis is then performed on the frequency domain features to extract feature vectors. Based on an adaptive fuzzy control algorithm, these feature vectors are mapped in the fuzzy space to obtain membership values, including: The attitude angle data and the current force state are arranged into a data matrix according to the sampling time sequence. After normalization, the attitude angle data and the current force state are sorted by timestamp, so that each row of the data matrix corresponds to the attitude angle data and the current force state at a sampling time. The data matrix is ​​subjected to wavelet transform to obtain a wavelet coefficient matrix. The energy spectrum and signal-to-noise ratio of the wavelet coefficient matrix are calculated. The resolution of the scale parameter is adaptively adjusted based on the signal-to-noise ratio. The sliding window size of the translation parameter is dynamically selected according to the energy spectrum. The data matrix is ​​decomposed and reconstructed at multiple scales based on the scale parameter and the sliding window to extract the frequency domain features. Calculate the covariance matrix of the frequency domain features, use the singular value decomposition method to perform eigenvalue decomposition on the covariance matrix to obtain an initial feature group, select the initial features in the initial feature group whose cumulative contribution rate reaches a preset contribution threshold to construct an orthogonal projection matrix, and multiply the frequency domain features with the orthogonal projection matrix to obtain the feature vector; The feature vectors are clustered using a fuzzy C-means clustering algorithm to obtain cluster centers. An exponential function is used to construct a membership function. Based on the membership function, the feature vectors are mapped to a fuzzy space and initial membership values ​​are calculated. Based on the initial membership values, the cluster centers are iteratively optimized using a pre-set performance index function and gradient descent. The membership values ​​are then recalculated and output based on the optimized cluster centers.

4. The method according to claim 1, characterized in that, The compensation rule is selected based on the membership value, and the rule weight coefficient of the compensation rule is optimized using a genetic algorithm. The optimized weight coefficient is then weighted and summed with the membership value to obtain the damping compensation value and generate damping control commands, including: A compensation rule base is constructed based on the membership value. The compensation rule base includes multiple compensation rules. Each compensation rule includes a membership value and a corresponding initial compensation value. Based on the membership value and the initial compensation value, the rule activation degree of each compensation rule is calculated using the algebraic product T-norm operator. Initialize the rule weight coefficients corresponding to the compensation rule, encode the rule weight coefficients as chromosomes, construct a fitness function based on control error and control increment, use the roulette wheel algorithm to select the rule weight coefficients and perform arithmetic crossover operation to obtain crossover weight coefficients, perform non-uniform mutation operation on the crossover weight coefficients and iteratively optimize according to the fitness function to obtain optimized rule weight coefficients; The optimized rule weight coefficient is multiplied by the rule activation degree to obtain the weighted activation degree. The weighted activation degree is multiplied by the initial compensation value and summed to obtain the compensation sum value. The weighted activation degrees are summed to obtain the activation sum value. The compensation sum value is divided by the activation sum value to obtain the compensation control amount. The compensation control quantity is compared with the preset compensation upper and lower limits to obtain the amplitude limiting compensation value. The difference between the amplitude limiting compensation value and the compensation control quantity at the previous moment is calculated to obtain the control increment. The control increment is compared with the preset increment upper and lower limits to obtain the limited control increment. The limited control increment is added to the compensation control quantity at the previous moment to obtain the damping compensation value corresponding to the current moment and a damping control command is generated.

5. The method according to claim 1, characterized in that, The vibration reduction compensation value is subjected to discrete Fourier transform to obtain spectral components. A state feedback matrix is ​​constructed based on the spectral components and convolved with a preset feature compensation matrix to obtain a control gain matrix, including: The vibration reduction compensation value is subjected to discrete Fourier transform to obtain the spectral components. The spectral components are decomposed into real and imaginary parts. The amplitude spectrum is calculated based on the square root of the sum of the squares of the real and imaginary parts. The phase spectrum is calculated based on the arctangent function of the quotient of the imaginary and real parts. The amplitude spectrum and the phase spectrum are arranged in order of frequency components to construct a state feedback matrix. The state feedback matrix includes an amplitude spectrum sequence in the first row and a phase spectrum sequence in the second row. Multiple preset frequency weight coefficients are multiplied by the amplitude spectrum and phase spectrum of the corresponding frequency components in the state feedback matrix to obtain a weighted state feedback matrix. A feature compensation matrix is ​​constructed, which includes multiple amplitude compensation coefficients in the first row and multiple phase compensation coefficients in the second row. The weighted state feedback matrix and the feature compensation matrix are subjected to a two-dimensional convolution operation. The control gain matrix is ​​calculated based on the sum of the products of the elements in the weighted state feedback matrix and the compensation coefficients at the corresponding positions in the feature compensation matrix.

6. The method according to claim 1, characterized in that, Real-time monitoring of the vibration reduction effect; if the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude, including: Vibration acceleration signals are collected by an accelerometer installed on the shock absorber, and vibration displacement signals are collected by a displacement sensor. The root mean square value of the vibration acceleration signal and the maximum amplitude of the vibration displacement signal are used as evaluation indicators of the shock absorption effect. When the evaluation indicators exceed the preset shock absorption threshold, parameter optimization is triggered. The current attitude is obtained, and the sum of squares of the difference between the current control value of the shock absorber and the expected current value is fitted by least squares to obtain the updated current control parameters. The Lyapunov function constructed by the attitude error and angular velocity is optimized to obtain the updated attitude adjustment parameters. The updated current control parameters and the updated attitude adjustment parameters are applied to the coordinated control of the shock absorber and the Mecanum wheel, respectively, until the evaluation index is less than the preset shock absorption threshold.

7. An intelligent shock absorption and attitude control system for battlefield medical equipment, used to implement the method of any one of claims 1-6, characterized in that, include: The first unit is used to acquire the posture sensor data of the battlefield medical equipment and calculate the posture angle data, collect the pressure data and displacement data on the shock absorber and determine the current stress state. The second unit is used to construct a data matrix by arranging the attitude angle data and the current force state in a time sequence, perform wavelet transform on the data matrix to obtain frequency domain features, perform principal component analysis on the frequency domain features to extract feature vectors, map the feature vectors in the fuzzy space based on an adaptive fuzzy control algorithm to obtain membership values, select compensation rules according to the membership values, optimize the rule weight coefficients of the compensation rules through a genetic algorithm, and perform a weighted sum of the obtained optimized weight coefficients and membership values ​​to obtain the vibration reduction compensation value and generate vibration reduction control commands. The third unit is used to perform discrete Fourier transform on the vibration reduction compensation value to obtain spectral components, construct a state feedback matrix based on the spectral components and perform convolution operation with a preset feature compensation matrix to obtain a control gain matrix, perform least squares fitting on the vibration reduction compensation value based on the control gain matrix, calculate the current control parameters and attitude adjustment parameters of the vibration damper respectively, and control the vibration damper and Mecanum wheel of the battlefield medical equipment according to the current control parameters and attitude adjustment parameters. The fourth unit is used to monitor the vibration reduction effect in real time. If the vibration reduction effect does not reach the preset vibration reduction threshold, the current control parameters and attitude adjustment parameters are updated based on the current attitude.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images