Large liquid storage container suspension type active anti-wave system

Abstract

The invention discloses a large liquid storage container suspension type active wave prevention system, and relates to the technical field, and the system comprises the steps: obtaining vehicle operation data of a canning vehicle at T-N moments in a driving process, and calculating and obtaining a vehicle dynamic coefficient and steering data according to the vehicle operation data, the vehicle operation data comprises a vehicle real-time speed, a vehicle transverse acceleration, a vehicle longitudinal acceleration and a steering wheel rotation angle; the method comprises the following steps: collecting data of liquid in a liquid storage tank of a canned vehicle, calculating to obtain a liquid attribute coefficient, inputting the liquid attribute coefficient, a vehicle dynamic coefficient and steering data into a set attitude prediction model to obtain optimal attitude data of a swash plate, and pre-adjusting the swash plate based on the optimal attitude data of the swash plate; according to the invention, under the constraint of the dynamic attitude of the vehicle, the three mechanisms of pre-adjustment guide, pressure difference feedback correction and disturbance label identification are fused, and the suspension type active control and disturbance mode classification response of the fluctuation of the liquid in the large liquid storage container are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of liquid storage container wave control, in particular to a large liquid storage container suspension type active wave control system. BACKGROUND

[0002] In heavy vehicle transportation, as the core fluid carrying unit, the internal liquid of the large liquid storage container is prone to violent shaking and local wave accumulation when subjected to continuous vibration, vehicle turning impact or complex path disturbance, which seriously affects the safety of the container structure and the attitude stability of the carrying platform, and even causes secondary accidents.

[0003] Traditional liquid disturbance control methods rely on fixed wave control structures such as partitions, transverse rib plates or foam buffers. These structures are simple and have low design costs, and can suppress liquid free surface fluctuations to a certain extent. However, such passive wave control systems lack adaptability to dynamic disturbances. When the external disturbance source is strong or the attitude changes quickly, they often cannot respond in time, causing the liquid to accumulate or violently recoil in a certain direction.

[0004] In recent years, some research has attempted to introduce adjustable wave control plates. By adjusting the tension state of the inner plate structure through mechanical binding or hydraulic actuator units, it can respond to wave fluctuations in different operating scenarios and improve the control ability of liquid disturbance. Although this type of system has some active control ability, it generally has the following technical limitations: 1. Control lag problem is prominent. Current active adjustment strategies rely on passive correction after disturbance has occurred based on sensor response, and fail to achieve pre-judgment and guidance before disturbance; 2. Limited adjustment accuracy, lack of fine-grained recognition and classification of liquid dynamic disturbance patterns, and unable to achieve targeted compensation; 3. Insufficient closed loop execution, the wave control structure adjustment action and disturbance recognition are mostly decoupled, and the control system cannot achieve adaptive closed loop optimization. SUMMARY

[0005] To solve the above technical problems, the present application provides a large liquid storage container suspension type active wave control system, which comprises: A feedforward data acquisition module S11 acquires vehicle operating data of the tank truck at time T-N during driving, and calculates vehicle dynamic coefficients and steering data based on the vehicle operating data, wherein the vehicle operating data includes real-time vehicle speed, vehicle lateral acceleration, vehicle longitudinal acceleration and steering wheel rotation angle; The wave protection pre-adjustment module S12 collects the liquid data in the storage tank of the tank truck, calculates the liquid property coefficient, inputs the liquid property coefficient, vehicle dynamic coefficient and steering data into the established posture prediction model to obtain the best posture data of the wave protection board, and pre-adjusts the wave protection board based on the best posture data of the wave protection board; The real-time data acquisition module S13 collects the instantaneous pressure values of each pressure sensor at time T during the driving process of the vehicle through the multiple pressure sensors arranged at the wall ends of the storage tank of the tank vehicle, calculates the pressure difference value based on the instantaneous pressure values of the wall ends, and the pressure difference value includes the front and rear wall pressure difference value and the left and right wall pressure difference value; The wave protection re-adjustment module S14 performs secondary adjustment on the pre-adjusted wave protection board based on the pressure difference value, and generates a liquid wave state label, which includes a longitudinal impact label, a lateral shaking label and a mixed turbulent label.

[0006] Further, the logic for calculating the vehicle dynamic coefficient and the steering data according to the vehicle operation data is as follows: S111, based on the lateral acceleration and longitudinal acceleration of the tank truck, the instantaneous total acceleration modulus value is calculated; S112, the vehicle dynamic coefficient is generated according to the real-time speed of the vehicle and the instantaneous total acceleration modulus value; S113, based on the derivative of the steering wheel rotation angle with respect to time, the steering data of the tank truck is generated, and the steering data includes steering rate and steering label, and the steering label includes left steering label and right steering label.

[0007] Further, the generation logic of the steering label is as follows: Based on the preset rotation rate threshold, if the derivative of the steering wheel rotation angle with respect to time is positive and the rotation rate of the steering wheel is greater than the rotation rate threshold, the left steering label is generated; If the derivative of the steering wheel rotation angle with respect to time is negative and the rotation rate of the steering wheel is greater than the rotation rate threshold, the right steering label is generated.

[0008] Further, the step of collecting the liquid data in the storage tank of the tank truck and calculating the liquid property coefficient includes: S121, based on the tank truck factory setting or filling stage record, the liquid identification filled in the current storage tank is obtained, and the liquid data corresponding to the liquid identification is called from the liquid property database, and the liquid property data includes liquid density, liquid viscosity and surface tension value; S122, the liquid property coefficient is calculated through the normalization processing formula.

[0009] Further, the generation logic of the posture prediction model is as follows: Historical attitude prediction data is acquired and divided into a test set and a training set. The historical attitude prediction data includes liquid property coefficients, vehicle dynamic coefficients, steering data, and corresponding optimal attitude data of the wave deflector. The optimal attitude data of the wave deflector is executed by motors of four sets of wire harness mechanisms installed in the liquid storage tank. A regression network is constructed, with the liquid property coefficients, vehicle dynamic coefficients, and steering data in the training set as inputs and the corresponding optimal attitude data of the wave deflector as outputs. The regression network is then trained to obtain the initial attitude prediction network. The initial pose prediction network is validated using a test set. The initial pose prediction network whose output is less than or equal to a preset test error threshold is used as the pose prediction model.

[0010] Furthermore, the logic for generating the test error threshold is as follows: a1: Initialize the candidate test error threshold set. Construct an initial error threshold set containing M candidate test error thresholds, where M is a positive integer. Each candidate test error threshold is a candidate threshold evaluated by the regression network on the validation set. a2: Initialize the population: Treat each candidate test error threshold in the error threshold set as an individual to form the original population. The original population contains M individuals, and each individual represents a test error threshold. a3: Fitness evaluation: Under the test error threshold corresponding to each individual, the trained regression network is used to predict the test set, and it is calculated whether each test sample is judged as a qualified prediction under the error threshold; based on the judgment results of all test samples, the prediction accuracy or average residual is statistically analyzed, a fitness function is constructed, and the fitness value is calculated for each individual. The formula for calculating fitness value is:

[0011] In the formula, The test error threshold is fitness value, This represents the total number of samples in the test set. To return the network to the first The predicted value of the nth test sample, the nth The true value of each test sample This represents the candidate test error threshold for the current individual. This is an indicator function; its value is 1 if the condition inside the parentheses is true, and 0 otherwise. a4: Selection operation: Use the roulette wheel method to select individuals from the original population, and select two individuals with higher fitness values ​​to be the father and mother individuals respectively; a5: crossover operation: performing weighted mean crossover or interval recombination on the test error threshold of the selected parent individual and the mother individual to generate a new candidate test error threshold individual; a6: mutation operation: introducing a disturbance factor to the candidate test error threshold individual generated by crossover, performing a random disturbance operation to generate N new test error threshold individuals, N being an integer greater than zero; combining the N new test error threshold individuals into a new population and replacing the original population, and returning to step a3; a7: repeatedly performing the above steps a2-a6 until the fitness value of at least one individual in the new population is greater than or equal to the preset fitness threshold, or the number of iterations of the algorithm exceeds the preset maximum number of iterations; if any of the termination conditions is met, output the test error threshold individual with the highest fitness value as the optimal test error threshold of the regression network.

[0012] Further, the pressure sensor includes a front wall sensor, a rear wall sensor, a left wall sensor, and a right wall sensor.

[0013] Further, the step of calculating the pressure difference value is: S131, according to the instantaneous pressure value of each wall end, the average pressure value of each wall end is calculated and obtained; S132, based on the average pressure value of each wall end, the front-rear wall pressure difference value and the left-right wall pressure difference value are calculated and obtained.

[0014] Further, the step of performing secondary adjustment on the pre-adjusted wave board based on the pressure difference value is: S141, the motor power of the four groups of beam line mechanisms during the pre-adjustment of the wave board is obtained, and the pre-adjustment traction value of the four corners of the wave board is calculated and obtained; S142, based on the given traction displacement increment factor, the target traction value is calculated and obtained; S143, according to the target traction value and the pre-adjustment traction value of the four corners of the wave board, the motor control displacement amount of the four corners is generated; S144, the direction of each motor is judged and the traction correction power is calculated, and the control power command value of each motor is calculated according to the traction correction power, and the wave board is secondarily adjusted based on the control power command value.

[0015] Further, the generation logic of the liquid fluctuation state label is: The front-rear wall pressure difference value and the left-right wall pressure difference value are compared with the disturbance threshold value, if the front-rear wall pressure difference value or the left-right wall pressure difference value is greater than the disturbance threshold value, the liquid fluctuation state label is generated, if the front-rear wall pressure difference value and the left-right wall pressure difference value are both less than or equal to the disturbance threshold value, the liquid fluctuation state label is not generated; The absolute value of the left and right wall pressure difference at time T is multiplied by a label discrimination threshold proportion factor to generate left and right correction data, which is compared with the absolute value of the front and back wall pressure difference at time T, and if the left and right correction data is less than the absolute value of the front and back wall pressure difference at time T, a transverse shaking label is generated.

[0016] The absolute value of the front and back wall pressure difference at time T is multiplied by a label discrimination threshold proportion factor to generate front and back correction data, which is compared with the absolute value of the left and right wall pressure difference at time T, and if the front and back correction data is less than the absolute value of the left and right wall pressure difference at time T, a transverse shaking label is generated. Other cases generate a mixed turbulent label.

[0017] Compared with the prior art, the present application has the following beneficial effects: The present application firstly performs the pre-adjustment action of the wave breaker to guide the liquid storage structure to actively adapt to the dynamic attitude of the vehicle and realize the preliminary adjustment of the liquid stability by constructing an attitude prediction mechanism with the vehicle running state as a constraint condition, combining the physical properties of the liquid and the configuration of the storage tank; In addition, by introducing the real-time pressure difference from different wall surfaces in the liquid storage tank as a feedback signal, the motor power adjustment is driven by the pressure difference value change, and the micro-amplitude traction correction of the wave breaker is realized through the regulation of the beam mechanism, thereby making up for the nonlinear error and inertia delay of the pre-adjustment, ensuring that the liquid disturbance control process has dynamic response capability and closed-loop precision; To improve the recognition efficiency of the system to the disturbance mode, the present application sets a disturbance threshold and constructs a fluctuation state label generation function, accurately distinguishes three typical liquid disturbance forms of longitudinal impact, transverse shaking and mixed turbulent by dynamically comparing the pressure differences of the front and back walls and the left and right walls, and provides a clear classification identification; In summary, under the constraint of the dynamic attitude of the vehicle, the present application combines the triple mechanisms of pre-adjustment guidance, pressure difference feedback correction and disturbance label recognition, realizes the suspension type active control and disturbance mode classification response of the liquid fluctuation in the large liquid storage container. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments or the prior art, the drawings needed in the embodiments will be briefly introduced below, and obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0019] Figure 1 A flowchart of a large liquid storage container suspension type active wave breaker system according to Embodiment 1 of the present application is provided. Figure 2A wave protection demonstration perspective view of a large liquid storage container suspension type active wave protection system provided for the embodiment 2 of the present application; Figure 3 A wave protection demonstration side view of a large liquid storage container suspension type active wave protection system provided for the embodiment 2 of the present application. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme of the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0021] Embodiment 1 Please refer to Figure 1 The embodiment discloses a large liquid storage container suspension type active wave protection system, which comprises: The feedforward data acquisition module S11 acquires vehicle running data of the tank truck at T-N moment in the driving process, and calculates vehicle dynamic coefficients and steering data according to the vehicle running data, wherein the vehicle running data comprises real-time speed of the vehicle, lateral acceleration of the vehicle, longitudinal acceleration of the vehicle and steering wheel rotation angle. It should be noted that the real-time speed of the vehicle is acquired through the vehicle CAN bus; the vehicle acceleration is acquired through the inertial measurement unit IMU; and the steering wheel rotation angle is acquired through the steering angle sensor.

[0022] Specifically, the logic for calculating the vehicle dynamic coefficients and the steering data according to the vehicle running data is as follows: S111, calculating an instantaneous total acceleration module value based on the lateral acceleration and the longitudinal acceleration of the tank truck; which is expressed as:

[0023] In the formula, is the instantaneous total acceleration module value, is the lateral acceleration, is the longitudinal acceleration; It should be noted that the instantaneous total acceleration module value is used to reflect the acceleration and deceleration and lateral acceleration intensity of the vehicle as a whole. S112, generating vehicle dynamic coefficients according to the real-time speed of the vehicle and the instantaneous total acceleration module value; which is expressed as:

[0024] In the formula, is a vehicle dynamic coefficient, is a predetermined maximum instantaneous total acceleration modulus, is a vehicle real-time speed, is a preset maximum real-time speed, , are respectively an instantaneous total acceleration processing value and a real-time speed processing value, , and are all greater than 0; S113, based on the derivative of the steering wheel rotation angle with respect to time, generates the tank vehicle steering wheel rotation data, and the steering data includes steering rate and steering label, and the steering label includes left steering label and right steering label; The calculation formula is represented as:

[0025] wherein, is the steering rate of the steering wheel, is the derivative of the steering wheel rotation angle with respect to time; It should be noted that: It can be understood as the angular velocity of the steering wheel, when the high-speed driving is slightly corrected, the steering rate is small, When the steering wheel is turned sharply, the steering rate is large, It will increase significantly; Specifically, the generation logic of the steering label is: Based on the preset steering rate threshold, if the derivative of the steering wheel rotation angle with respect to time is positive, and the steering rate of the steering wheel is greater than the steering rate threshold, a left steering label is generated; If the derivative of the steering wheel rotation angle with respect to time is negative, and the steering rate of the steering wheel is greater than the steering rate threshold, a right steering label is generated.

[0026] The wave protection pre-adjustment module S12: collects the liquid data in the liquid storage tank of the tank vehicle, and calculates and obtains the liquid property coefficient, inputs the liquid property coefficient, the vehicle dynamic coefficient, and the steering data into the predetermined posture prediction model to obtain the best posture data of the wave protection plate, and pre-adjusts the wave protection plate based on the best posture data of the wave protection plate. It should be noted that: the best posture data of the wave protection plate is the best adjustment angle of the wave protection plate, which has positive and negative values, and the positive value is clockwise pre-adjustment, and the negative value is counterclockwise pre-adjustment; Specifically, the steps of collecting the liquid data in the liquid storage tank of the tank vehicle and calculating and obtaining the liquid property coefficient include: S121, based on the tank truck factory setting or filling stage record, the liquid identification filled in the current liquid storage tank is obtained, and the liquid data corresponding to the liquid identification is called from the liquid attribute database, and the liquid attribute data includes liquid density, liquid viscosity and surface tension value; It should be noted that the liquid attribute database is cloud-linked with various liquid attributes on the network S122, the liquid attribute coefficient is calculated by a normalization processing formula; Indicated as:

[0027] In the formula, is the liquid attribute coefficient, , , respectively, the weight factor of liquid density, liquid viscosity and liquid surface tension value, is the liquid density, is the liquid viscosity, is the liquid surface tension value, , and is the liquid density, liquid viscosity and liquid surface tension value It should be noted that: , and Based on the mean value of the maximum and minimum attribute values in the liquid attribute database; Specifically, the generation logic of the posture prediction model is: Obtain historical posture prediction data, divide the historical posture prediction data into test set and training set, the historical posture prediction data includes liquid attribute coefficient, vehicle dynamic coefficient, steering data and corresponding best posture data of the wave board, the best posture data of the wave board is executed by the motor of the four groups of beam line mechanism 22 arranged in the liquid storage tank 21; It should be noted that: Figure 2 The best posture data of the wave board 23 is specifically the inclination angle of the wave board, which is pre-adjusted by the stretching force output of the beam line mechanism 22 arranged on the side wall of the liquid storage tank 21, there are four groups of beam line mechanism 22, which are connected with the four corners of the wave board 22 through the traction rope 25, and the beam line mechanism 22 is driven by the motor; Referring to Figure 3 26 is a limiting device for adjusting the wave board 23 in cooperation with the traction rope 25, and 24 is a limiting ring arranged in the liquid storage tank 21 for binding the traction rope 25; It should be noted that in this invention, after a certain amount of solution is released from the filling truck, the current liquid level in the tank is obtained through an existing real-time liquid level acquisition device, which includes, but is not limited to, a guided wave radar level gauge, a hydrostatic level sensor, and a magnetostrictive level gauge. After obtaining the current liquid level in the tank, the height of the baffle plate 23 is adjusted based on the motor in the wire harness mechanism 22. This allows the height of the baffle plate to be further adjusted according to the current liquid level in the tank after a certain amount of solution is released, so that it fits the liquid surface after the solution is released, thereby achieving anti-wave control that changes with the liquid level. The motor in the wire harness mechanism 22 is a servo motor.

[0028] A regression network is constructed, with the liquid property coefficients, vehicle dynamic coefficients, and steering data in the training set as inputs and the corresponding optimal attitude data of the wave deflector as outputs. The regression network is then trained to obtain the initial attitude prediction network. The initial pose prediction network is validated using a test set. The initial pose prediction network whose output is less than or equal to the preset test error threshold is used as the pose prediction model. It should be noted that the regression network includes, but is not limited to, feedforward neural networks, multilayer perceptrons, one-dimensional convolutional neural networks, and recurrent neural networks. Specifically, the logic for generating the test error threshold is as follows: a1: Initialize the candidate test error threshold set. Construct an initial error threshold set containing M candidate test error thresholds, where M is a positive integer. Each candidate test error threshold is a candidate threshold evaluated by the regression network on the validation set. a2: Initialize the population: Treat each candidate test error threshold in the error threshold set as an individual to form the original population. The original population contains M individuals, and each individual represents a test error threshold. a3: Fitness evaluation: Under the test error threshold corresponding to each individual, the trained regression network is used to predict the test set, and it is calculated whether each test sample is judged as a qualified prediction under the error threshold; based on the judgment results of all test samples, the prediction accuracy or average residual is statistically analyzed, a fitness function is constructed, and the fitness value is calculated for each individual. The formula for calculating fitness value is:

[0029] In the formula, The test error threshold is fitness value, This represents the total number of samples in the test set. To return the network to the first The predicted value of the nth test sample, the nth a real value of the test sample, a candidate test error threshold value representing the current individual, a indicator function, if the condition in the parentheses is true, the value is 1, otherwise 0; It should be noted that the predicted value of the regression network for the first test sample is the predicted value of the regression network for the first test sample is the driving force output value of the motor in the beam mechanism 22 of the test sample; a4: selection operation: adopt roulette method to select individuals in the original population, select two individuals with higher fitness value as father and mother individuals; a5: crossover operation: weighted mean crossover or interval recombination is performed on the test error threshold values of the selected father and mother individuals to generate new candidate test error threshold individuals; a6: mutation operation: introduce a disturbance factor to the candidate test error threshold individuals generated by crossover, perform random disturbance operation to generate N new test error threshold individuals, N is an integer greater than zero; combine the N new test error threshold individuals into a new population and replace the original population, and return to step a3; a7: repeat the above steps a2-a6 until at least one individual in the new population has a fitness value greater than or equal to the preset fitness threshold, or the number of iterations exceeds the preset maximum number of iterations; if any of the termination conditions is met, output the test error threshold individual with the highest fitness value as the optimal test error threshold value of the regression network.

[0030] The real-time data acquisition module S13: through the multiple pressure sensors arranged at the ends of the walls inside the liquid storage tank of the tank vehicle, the instantaneous pressure values of each pressure sensor at time T during the driving process of the vehicle are acquired; according to the instantaneous pressure values of each wall end, the pressure difference values are calculated, and the pressure difference values include the front and rear wall pressure difference values and the left and right wall pressure difference values; Specifically, the pressure sensor includes a front wall sensor, a rear wall sensor, a left wall sensor and a right wall sensor; It should be noted that multiple pressure sensors are arranged on the front wall, rear wall, left wall and right wall in the liquid storage tank, and the numbering mode is as follows: The front wall pressure sensor set: ; The rear wall pressure sensor set: ; The left wall pressure sensor set: ; The right wall pressure sensor set: ; Specifically, the step of calculating the pressure difference value is: S131, Calculate and obtain the average pressure value of each wall end based on the instantaneous pressure value of each wall end; Represented as:

[0031] In the formula, This represents the average pressure value on the front wall. For the first The instantaneous pressure value collected by the front wall sensor at time T. This represents the average pressure value on the rear wall. For the first The instantaneous pressure value collected by the rear wall sensor. This represents the average pressure value on the left wall. For the first The instantaneous pressure value collected by the sensor on the left wall. This represents the average pressure value on the right wall. For the first The instantaneous pressure value collected by the sensor on the left wall. The number of sensors to be set for each wall end; It should be noted that: This refers only to the number of pressure sensors installed at a single wall end, not the total number of pressure sensors installed at all wall ends. S132, based on the average pressure value at each wall end, calculate and obtain the pressure difference between the front and rear walls and the pressure difference between the left and right walls; Represented as:

[0032] In the formula, Let T be the pressure difference between the front and rear walls. The pressure difference between the left and right walls at time T; It should be noted that the pressure difference between the front and rear walls... The pressure difference between the left and right walls is used to characterize the degree of inertial sloshing of the liquid in the direction of travel in a liquid storage tank. Used to characterize the intensity of lateral sway caused by the liquid storage tank when it turns; Wave-blocking readjustment module S14: Based on the pressure difference, the pre-adjusted wave-blocking plate is adjusted a second time, and liquid wave status labels are generated. The liquid wave status labels include longitudinal impact labels, lateral sway labels and mixed turbulence labels.

[0033] Specifically, the steps for secondary adjustment of the pre-adjusted baffle plate based on the pressure difference are as follows: S141, obtain the motor power of the four sets of wire harness mechanisms when pre-adjusting the wave deflector, and calculate the pre-adjustment traction value of the four corners of the wave deflector; Represented as:

[0034] in, For the first Pre-adjustment traction value of each motor during the pre-adjustment phase. The calibration factor for the shift value generated per unit power. For the first Motor power, ; It should be noted that: the moving value calibration coefficient Obtained through experimental calibration; S142, Calculate and obtain the target traction value based on the predetermined traction displacement increment factor; Represented as:

[0035]

[0036] in, To correct the target displacement longitudinally, To correct the target displacement laterally. This is a predetermined traction displacement increment factor; It should be noted that the predetermined traction displacement increment factor is determined based on historical data. S143, based on the target traction value and the pre-adjusted traction value of the four corners of the wave deflector, generate the motor control displacement of the four corners; Represented as:

[0037] In the formula, This represents the motor control displacement of the first motor at time T, and so on. S144: Determine the direction of each motor and calculate the traction correction power, and calculate the control power command value of each motor based on the traction correction power, and perform secondary adjustment of the baffle plate based on the control power command value. The formula for calculating traction correction power is:

[0038] in, For the first The traction correction power of each motor, if If >0, it indicates the "tighten the traction rope" direction, and the motor will start rotating forward. If it is less than 0, it indicates the direction of "releasing the traction rope". If the value is 0, the motor will remain in its current state without adjustment. The formula for calculating the control power command value is:

[0039] in, For time T Control power command value for each motor; Specifically, the logic for generating liquid fluctuation state tags is as follows: The pressure difference between the front and rear walls and the pressure difference between the left and right walls are compared with the disturbance threshold. If the pressure difference between the front and rear walls or the pressure difference between the left and right walls is greater than the disturbance threshold, a liquid fluctuation state label is generated. If the pressure difference between the front and rear walls and the pressure difference between the left and right walls are both less than or equal to the disturbance threshold, no liquid fluctuation state label is generated. The absolute value of the pressure difference between the left and right walls at time T is multiplied by the label discrimination threshold scaling factor to generate left and right correction data. This data is then compared with the absolute value of the pressure difference between the front and rear walls at time T. If the left and right correction data is less than the absolute value of the pressure difference between the front and rear walls at time T, then a new data set is generated.

[0040] Multiply the absolute value of the pressure difference between the front and rear walls at time T by the label discrimination threshold scaling factor to generate the corrected data. It is then compared with the absolute value of the pressure difference between the left and right walls at time T. If the corrected data before and after is less than the absolute value of the pressure difference between the left and right walls at time T, a lateral sway label is generated. In other cases, mixed turbulence tags are generated; The function for generating liquid fluctuation state labels is expressed as follows:

[0041] In the formula, This is the function for generating labels for liquid fluctuation states. To prevent the generation of label symbols, This is the disturbance threshold. The label discrimination threshold scaling factor.

[0042] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A large liquid storage container suspension active wave-damping system, characterized in that, The system includes: Feedforward data acquisition module S11: used to acquire vehicle operation data at time TN during the driving process of the tanker truck, and calculate and acquire vehicle dynamic coefficients and steering data based on the vehicle operation data. The vehicle operation data includes real-time vehicle speed, vehicle lateral acceleration, vehicle longitudinal acceleration and steering wheel rotation angle. S12: The anti-surge pre-adjustment module is used to collect liquid data in the storage tank of the tanker truck and calculate the liquid property coefficient. The liquid property coefficient, vehicle dynamic coefficient and steering data are input into the predetermined attitude prediction model to obtain the optimal attitude data of the anti-surge plate. Based on the optimal attitude data of the anti-surge plate, the anti-surge plate is pre-adjusted. Real-time data acquisition module S13: used to acquire the instantaneous pressure values ​​of each pressure sensor at time T during the vehicle's operation by using multiple pressure sensors installed on each wall end inside the storage tank of the tanker vehicle; and to calculate the pressure difference based on the instantaneous pressure values ​​of each wall end, wherein the pressure difference includes the pressure difference between the front and rear walls and the pressure difference between the left and right walls. Wave-blocking readjustment module S14: used to perform secondary adjustment on the pre-adjusted wave-blocking plate based on the pressure difference, and generate liquid wave status labels, which include longitudinal impact labels, lateral sway labels and mixed turbulence labels.

2. The large liquid storage container suspension active wave-damping system according to claim 1, characterized in that, The logic for calculating and obtaining vehicle dynamic coefficients and steering data based on vehicle operation data is as follows: S111, the instantaneous total acceleration modulus is obtained based on the lateral and longitudinal acceleration of the tank truck; S112, generate vehicle dynamic coefficients based on the vehicle's real-time speed and instantaneous total acceleration modulus; S113, Based on the derivative of the steering wheel rotation angle with respect to time, generate steering wheel rotation data for the tanker truck. The steering data includes steering rate and steering labels, and the steering labels include left steering labels and right steering labels.

3. The large liquid storage container suspension active wave-damping system according to claim 2, characterized in that, The logic for generating the turn tag is as follows: Based on a preset rotation rate threshold, if the derivative of the steering wheel rotation angle with respect to time is positive and the steering wheel rotation rate is greater than the rotation rate threshold, a left turn label is generated. If the derivative of the steering wheel rotation angle with respect to time is negative, and the steering wheel rotation rate is greater than the rotation rate threshold, then a right turn label is generated.

4. The large liquid storage container suspension active wave-damping system according to claim 3, characterized in that, The steps for collecting liquid data from the storage tank of a tanker truck and calculating the liquid property coefficients include: S121, based on the factory settings of the tanker truck or the filling stage record, obtain the liquid identifier currently filled in the storage tank, and call the liquid data corresponding to the liquid identifier from the liquid attribute database. The liquid attribute data includes liquid density, liquid viscosity and surface tension value. S122, the liquid property coefficients are calculated and obtained through the normalization formula.

5. A large liquid storage container suspension active wave-damping system according to claim 4, characterized in that, The generation logic of the attitude prediction model is as follows: Historical attitude prediction data is acquired and divided into a test set and a training set. The historical attitude prediction data includes liquid property coefficients, vehicle dynamic coefficients, steering data, and corresponding optimal attitude data of the wave deflector. The optimal attitude data of the wave deflector is executed by motors of four sets of wire harness mechanisms installed in the liquid storage tank. A regression network is constructed, with the liquid property coefficients, vehicle dynamic coefficients, and steering data in the training set as inputs and the corresponding optimal attitude data of the wave deflector as outputs. The regression network is then trained to obtain the initial attitude prediction network. The initial pose prediction network is validated using a test set. The initial pose prediction network whose output is less than or equal to a preset test error threshold is used as the pose prediction model.

6. The large liquid storage container suspension active wave-damping system according to claim 5, characterized in that, The logic for generating the test error threshold is as follows: a1: Initialize the candidate test error threshold set. Construct an initial error threshold set containing M candidate test error thresholds, where M is a positive integer. Each candidate test error threshold is a candidate threshold evaluated by the regression network on the validation set. a2: Initialize the population: Treat each candidate test error threshold in the error threshold set as an individual to form the original population. The original population contains M individuals, and each individual represents a test error threshold. a3: Fitness assessment: Under the test error threshold corresponding to each individual, the trained regression network is used to predict the test set, and it is calculated whether each test sample is judged as a qualified prediction under the error threshold. Based on the judgment results of all test samples, the prediction accuracy or average residual is statistically analyzed, a fitness function is constructed, and the fitness value is calculated for each individual. The formula for calculating fitness value is: In the formula, The test error threshold is fitness value, This represents the total number of samples in the test set. To return the network to the first The predicted value of the nth test sample, the nth The true value of each test sample This represents the candidate test error threshold for the current individual. This is an indicator function; its value is 1 if the condition inside the parentheses is true, and 0 otherwise. a4: Selection operation: Use the roulette wheel method to select individuals from the original population, and select two individuals with higher fitness values ​​to be the father and mother individuals respectively; a5: Crossover operation: Perform weighted mean crossover or interval recombination on the test error thresholds of the selected parent and mother individuals to generate new candidate test error threshold individuals; a6: Mutation operation: Introduce a perturbation factor into the candidate test error threshold individuals generated by crossover, perform a random perturbation operation, and generate N new test error threshold individuals, where N is an integer greater than zero; combine the N new test error threshold individuals into a new population and replace the original population, then return to step a3; a7: Repeat steps a2 to a6 above until the fitness value of at least one individual in the new population is greater than or equal to the preset fitness threshold, or the number of algorithm iterations exceeds the preset maximum number of iterations; If any termination condition is met, the individual with the highest fitness value is output as the optimal test error threshold for the regression network.

7. A large liquid storage container suspension active wave-damping system according to claim 6, characterized in that, The pressure sensor includes a front wall sensor, a rear wall sensor, a left wall sensor, and a right wall sensor.

8. A large liquid storage container suspension active wave-damping system according to claim 7, characterized in that, The steps for calculating the pressure difference are as follows: S131, Calculate and obtain the average pressure value of each wall end based on the instantaneous pressure value of each wall end; S132, based on the average pressure value at each wall end, calculate and obtain the pressure difference between the front and rear walls and the pressure difference between the left and right walls.

9. A large liquid storage container suspension active wave-damping system according to claim 8, characterized in that, The steps for secondary adjustment of the pre-adjusted baffle plate based on the pressure difference are as follows: S141, obtain the motor power of the four sets of wire harness mechanisms when pre-adjusting the wave deflector, and calculate the pre-adjustment traction value of the four corners of the wave deflector; S142, Calculate and obtain the target traction value based on the predetermined traction displacement increment factor; S143, based on the target traction value and the pre-adjusted traction value of the four corners of the wave deflector, generate the motor control displacement of the four corners; S144: Determine the direction of each motor and calculate the traction correction power, and calculate the control power command value of each motor based on the traction correction power, and perform secondary adjustment of the baffle plate based on the control power command value.

10. A large liquid storage container suspension active wave-damping system according to claim 9, characterized in that, The logic for generating liquid fluctuation state labels is as follows: The pressure difference between the front and rear walls and the pressure difference between the left and right walls are compared with the disturbance threshold. If the pressure difference between the front and rear walls or the pressure difference between the left and right walls is greater than the disturbance threshold, a liquid fluctuation state label is generated. If the pressure difference between the front and rear walls and the pressure difference between the left and right walls are both less than or equal to the disturbance threshold, no liquid fluctuation state label is generated. The absolute value of the pressure difference between the left and right walls at time T is multiplied by the label discrimination threshold scaling factor to generate left and right correction data. This data is then compared with the absolute value of the pressure difference between the front and rear walls at time T. If the left and right correction data is less than the absolute value of the pressure difference between the front and rear walls at time T, then a new data set is generated. Multiply the absolute value of the pressure difference between the front and rear walls at time T by the label discrimination threshold scaling factor to generate the corrected data. It is then compared with the absolute value of the pressure difference between the left and right walls at time T. If the corrected data before and after is less than the absolute value of the pressure difference between the left and right walls at time T, a lateral sway label is generated. In other cases, mixed turbulence labels are generated.

Images