An artificial intelligence driven crane adaptive anti-interference control method and system

Abstract

The present invention provides an artificial intelligence driven crane adaptive anti-interference control method and system. In the technical field of crane adaptive anti-interference control, the present invention first collects multiple groups of historical data of crane operation to form a historical data set, and divides the historical data set into a training set and a verification set in proportion, uses a neural network to build a crane interference prediction model, uses the training set and the verification set to train and optimize the model, generates a swing interference prediction model, collects various data of the crane operation in real time, uses the swing interference prediction model to predict the swing interference coefficient, generates a wind interference coefficient through the weight, the cross-sectional area of ​​the weight, the wind speed and the air density; generates a vibration interference coefficient through the vibration frequency and amplitude of the boom; adjusts the system gain through the above interference coefficient, and then controls the output of the controller through the gain to achieve the purpose of adaptive anti-interference.

Description

Technical Field

[0001] The present invention relates to the technical field of crane adaptive anti-interference control, and in particular to an artificial intelligence driven crane adaptive anti-interference control method and system. Background Art

[0002] Cranes are widely used in industry and construction, and their main function is to lift and move heavy objects. However, cranes may be subject to various disturbances during operation, such as wind, vibration, and swing, which may affect the stability and performance of the crane. Traditional crane control methods are usually based on pre-defined control strategies, which may not adapt to changes and disturbances in the external environment, resulting in a decrease in control performance. In order to improve the anti-interference ability and adaptability of cranes, artificial intelligence technology has been introduced into crane control. By using artificial intelligence algorithms such as neural networks, fuzzy logic, and genetic algorithms, cranes can achieve adaptive adjustment through PID controllers, thereby improving their stability and performance.

[0003] In the prior art, publication number CN111941432B discloses a high-performance robotic arm artificial intelligence output feedback control method, including establishing a robotic arm dynamics model and a system state equation; designing a state observer to estimate the unknown state of the entire system; combining the state observer, designing a disturbance observer to estimate and compensate for the external disturbance, modeling error and network approximation error of the system; based on a composite observer, designing an adaptive neural network output feedback controller for the robotic arm; designing a neural network weight adaptation law to achieve the adaptation of the composite observer, the adaptation of the controller and the rapid approximation of the uncertainties in the system modeling parameters, thereby solving the technical problems of high-performance and high-precision control. This prior art takes a robotic arm system as a research object, and realizes that the robotic arm joint position output can accurately track the expected position under the conditions where the robotic arm system has unknown external disturbances, unknown modeling parameters, model uncertainties and only position signals.

[0004] The shortcomings of this prior art are: 1. This prior art relies on accurate system model design. When there is great uncertainty or dynamic changes in the system, accurate modeling is very difficult, which may lead to inaccurate or even failed models. If the system model is inaccurate, the performance of the observer will also be affected. If the modeling fails, the entire system will be meaningless. 2. This prior art is usually designed for specific systems and specific problems, and has low versatility and flexibility. In different application scenarios or system changes, this prior art may need to be redesigned and adjusted, and has low adaptability. 3. This prior art is relatively limited in adapting to dynamic changes in the system and real-time updates, and it is difficult to achieve online learning and adaptation. In an environment where the system changes frequently, this prior art may not be able to adapt to changes in a timely manner, affecting the control effect.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The object of the present invention is to provide an artificial intelligence driven crane adaptive anti-interference control method and system to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] An artificial intelligence driven crane adaptive anti-interference control method, the specific steps include:

[0009] Step 1: Collect multiple sets of historical data of crane operation to form a historical data set. Each set of data includes weight mass, weight cross-sectional area, wind speed, air density, weight swing angle, swing angular velocity, rope length, boom vibration frequency and amplitude; generate a swing interference coefficient based on the weight swing angle and swing angular velocity in each set of data;

[0010] Step 2: According to the historical data set, 80% of the data set is divided into a training set and 20% is divided into a validation set. A crane interference prediction model is constructed based on a neural network, and the swing interference coefficient is used as a label. The crane interference prediction model is trained using the training set. After the training, the crane interference prediction model is evaluated using the validation set. The hyperparameters of the crane interference prediction model are adjusted according to the evaluation results to generate a swing interference prediction model.

[0011] Step 3: Collecting crane working data in real time, the working data including weight mass, weight cross-sectional area, wind speed, air density, suspension rope length, boom vibration frequency and amplitude; inputting the crane working data into the swing interference prediction model to generate a swing interference coefficient;

[0012] Step 4: Generate a wind interference coefficient through the weight mass, weight cross-sectional area, wind speed and air density in the real-time collected crane working data; generate a vibration interference coefficient through the boom vibration frequency and amplitude in the real-time collected crane working data;

[0013] Step 5: Obtain the proportional gain, differential gain and integral gain in the PID control system of the real-time crane, adjust the proportional gain of the system by the wind disturbance coefficient, adjust the integral gain of the system by the swing disturbance coefficient, and adjust the differential gain of the system by the vibration disturbance coefficient;

[0014] Step 6: Control the output of the crane controller by adjusting the proportional gain, derivative gain and integral gain.

[0015] Furthermore, the specific logic for generating the swing interference coefficient is: assigning weights to the swing angle of the weight and the square of the swing angular velocity to generate the swing interference coefficient, and the specific formula is:

[0016]

[0017] in, is the swing interference coefficient, θ is the swing angle of the weight, ω is the angular velocity of the weight, α is the angle weight coefficient, β is the angular velocity weight coefficient, α+β=1, 0<β<0.1α<α<1.

[0018] Furthermore, the specific logic for selecting mean square error as the loss function when optimizing the crane interference prediction model is: for each parameter, calculate the error between the predicted value and the actual value, square the error of each sample, sum all the square errors and take the average; the specific formula is:

[0019]

[0020] MSE is the mean square error, n is the number of samples, y i is the actual swing interference coefficient of the i-th sample, is the predicted swing interference coefficient of the i-th sample.

[0021] Furthermore, the specific logic for generating the wind interference coefficient is: the wind force is obtained by approximating the wind speed, air density, wind coefficient and cross-sectional area of ​​the weight; the specific formula for obtaining the wind force is:

[0022]

[0023] Among them, F is wind force, V is wind speed, ρ is air density, C d is the wind force coefficient, and M is the cross-sectional area of ​​the weight.

[0024] Then, the wind interference coefficient is generated according to the wind force and the mass of the heavy object. The specific formula for generating the wind interference coefficient is:

[0025]

[0026] in, is the wind disturbance coefficient, m is the mass of the heavy object, and g is the acceleration of gravity.

[0027] Furthermore, the specific logic for generating the vibration interference coefficient is to generate the vibration interference coefficient through the boom vibration frequency and amplitude; the specific formula is:

[0028]

[0029] in, is the vibration interference coefficient, A is the vibration amplitude of the boom, and f is the vibration frequency of the boom.

[0030] Furthermore, the wind disturbance is adjusted by adjusting the proportional gain in the controller, and the formula for adjusting the proportional gain is:

[0031]

[0032] Among them, K Pnew is the adjusted proportional gain, K Pbase is the original proportional gain, α F is the wind adaptation coefficient, is the wind disturbance coefficient,

[0033] The swing disturbance is adjusted by adjusting the integral gain, based on the formula:

[0034]

[0035] Among them, K Inew is the adjusted integral gain, K Ibase is the original integral gain, is the swing interference coefficient; α B is the swing adaptation coefficient;

[0036] The vibration disturbance is adjusted by adjusting the differential gain, based on the formula:

[0037]

[0038] Among them, K dnew is the adjusted differential gain, K dbase is the original differential gain, α F is the vibration adaptive coefficient, is the vibration interference coefficient.

[0039] Furthermore, the specific formula for obtaining the output of the crane controller is:

[0040]

[0041] Where u(t) is the controller output, K Pnew is the adjusted proportional gain, K Inew is the adjusted integral gain, K dnew is the adjusted differential gain, e(t) is the error signal, and t represents the time variable.

[0042] The present invention further provides an artificial intelligence driven crane adaptive anti-interference control system, which is used to execute the above artificial intelligence driven crane adaptive anti-interference control method, comprising:

[0043] The acquisition module is used to collect multiple groups of historical data of the crane during operation to form a historical data set, each group of data includes the weight, the cross-sectional area of ​​the weight, the wind speed, the air density, the weight swing angle, the swing angular velocity, the length of the suspension rope, the boom vibration frequency and amplitude; the swing interference coefficient is generated according to the weight swing angle and the swing angular velocity in each group of data;

[0044] A modeling optimization module is used to divide 80% of the data set into a training set and 20% into a validation set according to the historical data set, build a crane interference prediction model based on a neural network, and use the swing interference coefficient as a label; use the training set to train the crane interference prediction model, and after the training, use the validation set to evaluate the crane interference prediction model, adjust the hyperparameters of the crane interference prediction model according to the evaluation results, and generate a swing interference prediction model;

[0045] A prediction module is used to collect crane working data in real time, wherein the working data includes weight mass, cross-sectional area of ​​weight, wind speed, air density, length of suspension rope, boom vibration frequency and amplitude; the crane working data is input into a swing interference prediction model to generate a swing interference coefficient;

[0046] An analysis module is used to generate a wind interference coefficient through the weight mass, the cross-sectional area of ​​the weight, the wind speed and the air density in the crane working data collected in real time; and to generate a vibration interference coefficient through the boom vibration frequency and amplitude in the crane working data collected in real time;

[0047] A gain adjustment module is used to obtain the proportional gain, differential gain and integral gain in the PID control system of the real-time crane, adjust the proportional gain of the system through the wind interference coefficient, adjust the integral gain of the system through the swing interference coefficient, and adjust the differential gain of the system through the vibration interference coefficient;

[0048] Output control module, used to control the output of the crane controller through adjusted proportional gain, derivative gain and integral gain.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] When dealing with more complex, changeable and difficult-to-calculate interference factors, the present invention constructs a model and uses large-scale data for training and optimization, which can not only make the results more accurate, but also discover and capture the complex patterns and characteristics of the system, and can better utilize data for adaptive control.

[0051] The present invention can continuously adapt to new data and changes through online learning and updating mechanisms, meeting the control needs in dynamic environments, in environments with dynamic changes and high uncertainty.

[0052] The present invention divides interference into wind interference, swing interference and vibration interference, which are adjusted through proportional gain, differential gain and integral gain respectively, so as to control the system more specifically and meticulously. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of the control method in the present invention;

[0054] Figure 2 It is a structural schematic diagram of the control system in the present invention. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0056] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0057] Example:

[0058] See also Figure 1 , the present invention provides a technical solution:

[0059] An artificial intelligence driven crane adaptive anti-interference control method, the specific steps include:

[0060] Step 1: Collect multiple sets of historical data of crane operation to form a historical data set. Each set of data includes weight mass, weight cross-sectional area, wind speed, air density, weight swing angle, swing angular velocity, rope length, boom vibration frequency and amplitude; generate a swing interference coefficient based on the weight swing angle and swing angular velocity in each set of data;

[0061] In this embodiment, the mass of the heavy object is directly measured, and the cross-sectional area of ​​the heavy object is measured by projection. The specific operation is as follows: the heavy object is placed in a flat and open place, a light source is used to project the heavy object, the area of ​​the projection is calculated, and then the ratio of the cross-sectional area of ​​the heavy object to the projection is calculated according to the distance between the projection and the real object. The cross-sectional area of ​​the heavy object is calculated by the relationship between the projection area and the above ratio. The swing angle of the heavy object and the length of the suspension rope are directly measured by the position sensor. The vibration frequency f, amplitude A and swing angular velocity ω of the crane boom are directly measured by the acceleration sensor. The wind speed V is measured by the wind speed sensor, and the temperature T and pressure P are obtained by the temperature sensor and the atmospheric pressure sensor. The air density ρ is obtained according to the formula:

[0062]

[0063] Among them, ρ is the air density, R is the gas constant, R is usually 8.314J / (mol·K), P is the gas pressure, and T is the absolute temperature of the gas.

[0064] The specific logic for generating the swing interference coefficient is: assigning weights to the swing angle of the weight and the square of the swing angular velocity according to certain weights to generate the swing interference coefficient. The specific formula is:

[0065]

[0066] in, is the swing interference coefficient, θ is the swing angle of the weight, ω is the angular velocity of the weight, α is the angle weight coefficient, β is the angular velocity weight coefficient, α+β=1, 0<β<0.1α<α<1; Swing interference coefficient It reflects the impact of the swing disturbance on the system. The larger the value, the greater the interference caused by the swing disturbance to the system. The swing angle θ of the weight reflects the angle of the weight swing. The larger the value, the greater the weight swing. The angular velocity ω of the weight swing is the angular velocity of the weight swing. The larger the value, the greater the angular velocity of the weight swing. α is the angle weight coefficient reflecting the contribution weight of the angle to the swing disturbance. The larger the value, the greater the contribution of the swing angle to the swing disturbance. β is the angular velocity weight coefficient, reflecting the contribution weight of the angular velocity to the swing disturbance. The larger the value, the greater the contribution of the swing angular velocity to the swing disturbance. The influence of the swing angle on the swing disturbance is much higher than the influence of the swing angular velocity on the swing disturbance. However, in the process of studying the swing disturbance, the influence of the swing angular velocity on the swing disturbance is not small enough to be ignored. Therefore, when setting the weight, the weight of the square of the swing angular velocity is chosen to be one order of magnitude lower than the weight of the square of the swing angle. This formula reflects the interference caused by the swing of the weight to the system by combining the swing angle of the weight and the square of the swing angular velocity according to a certain weight.

[0067] Step 2: According to the historical data set, 80% of the data set is divided into a training set and 20% is divided into a validation set. A crane interference prediction model is constructed based on a neural network, and the swing interference coefficient is used as a label. The crane interference prediction model is trained using the training set. After the training, the crane interference prediction model is evaluated using the validation set. The hyperparameters of the crane interference prediction model are adjusted according to the evaluation results to generate a swing interference prediction model.

[0068] The specific evaluation and adjustment logic of the validation set and test set for the crane interference prediction model is as follows: select the mean absolute error loss function, calculate the loss value on the validation set, set the preset loss value, compare the calculated loss value with the preset loss value, and if the calculated loss value is greater than or equal to the preset loss value, adjust it by adding a regularization number; if the calculated loss value is less than the preset loss value, the model does not need to be adjusted to generate a swing interference prediction model.

[0069] The specific logic for choosing mean square error as the loss function when optimizing the crane interference prediction model is: for each parameter, calculate the error between the predicted value and the actual value, square the error of each sample, sum all the squared errors and take the average; the specific formula is:

[0070]

[0071] MSE is the mean square error, n is the number of samples, y i is the actual swing interference coefficient of the i-th sample, is the predicted swing interference coefficient of the i-th sample.

[0072] In this embodiment, the loss values ​​on the training set and the validation set are observed: if the loss value of the model on the training set is very low, but the loss value on the validation set is very high, it means that the model may be overfitting; if the loss values ​​of the model on both the training set and the validation set are very high, it means that the model may not fit the training data well and there is an underfitting problem; the complexity of the model can be penalized by adding a regularization term to the loss function to solve the overfitting and underfitting problems. This embodiment uses L1 regularization to deal with the overfitting and underfitting problems.

[0073] L1 regularization: L1 regularization penalizes or rewards the complexity of the model by adding the L1 norm of the model weights in the loss function. Specifically, for the linear regression model, the loss function of L1 regularization can be expressed as:

[0074]

[0075] Among them, θ is the parameter of the model, MAE is the mean absolute error loss function, and γ is the regularization parameter used to control the strength of regularization. is the L1 norm (also known as the Manhattan norm), which represents the sum of the absolute values ​​of the model parameters. L1 regularization tends to produce sparse weights, so even a small number of features have a large impact on the model, which helps feature selection and model interpretability.

[0076] Step 3: Collecting crane working data in real time, the working data including weight mass, weight cross-sectional area, wind speed, air density, suspension rope length, boom vibration frequency and amplitude; inputting the crane working data into the swing interference prediction model to generate a swing interference coefficient;

[0077] Step 4: Generate a wind interference coefficient through the weight mass, weight cross-sectional area, wind speed and air density in the real-time collected crane working data; generate a vibration interference coefficient through the boom vibration frequency and amplitude in the real-time collected crane working data;

[0078] The wind speed V is approximated to obtain the wind force F, and the specific logic is based on approximating the wind force by the wind speed, air density, wind force coefficient and cross-sectional area of ​​the weight. The specific formula is:

[0079]

[0080] Among them, F is wind force, V is wind speed, ρ is air density, C d is the wind force coefficient, and M is the cross-sectional area of ​​the weight.

[0081] Then, the wind interference coefficient is generated according to the wind force and the mass of the heavy object. The specific formula for generating the wind interference coefficient is:

[0082]

[0083] in, is the wind interference coefficient, m is the mass of the heavy object, g is the acceleration of gravity, and F is the wind force; It reflects the impact of wind interference on the system. The larger the value, the greater the interference of wind interference on the system. Wind force F reflects the size of wind acting on the heavy object when the crane is working. The larger the value, the greater the wind force acting on the heavy object. The mass m of the heavy object is the mass of the load when the crane is working. The larger the value, the heavier the load. This formula combines wind force and the mass of heavy objects to reflect the degree of influence of wind interference on the system.

[0084] The specific logic for generating the vibration interference coefficient is to generate the vibration interference coefficient through the boom vibration frequency and amplitude; the specific formula is:

[0085]

[0086] in, is the vibration interference coefficient, A is the vibration amplitude of the boom, and f is the vibration frequency of the boom. It reflects the impact of vibration interference on the system. The larger the value, the greater the interference of the vibration interference on the system. The boom vibration amplitude A reflects the vibration amplitude of the boom when the crane vibrates during operation. The larger the value, the greater the vibration amplitude. The boom vibration frequency f reflects the vibration frequency of the boom when the crane vibrates during operation. The larger the value, the higher the vibration frequency. This formula generates a vibration interference coefficient that can comprehensively reflect the vibration interference by analyzing the vibration amplitude and frequency of the boom when the crane vibrates during operation, reflecting the degree of influence of vibration interference on the system.

[0087] Step 5: Obtain the proportional gain, differential gain and integral gain in the PID control system of the real-time crane, adjust the proportional gain of the system by the wind disturbance coefficient, adjust the integral gain of the system by the swing disturbance coefficient, and adjust the differential gain of the system by the vibration disturbance coefficient;

[0088] In this embodiment, the wind interference is adjusted by adjusting the proportional gain in the controller. The formula for adjusting the proportional gain is:

[0089]

[0090] Among them, K Pnew is the adjusted proportional gain, K Pbase is the original proportional gain, α F is the wind adaptation coefficient, is the wind interference coefficient. This formula dynamically adjusts the proportional gain according to the wind interference coefficient and the wind adaptive coefficient, so that the crane system can better cope with wind interference and improve the stability and control performance of the system. The adjusted proportional gain K Pnew It reflects the adjustment of the proportional control parameters when the system faces wind disturbance. The larger the value, the more sensitive the system is to wind disturbance and the more obvious the adjustment effect. Pbase Reflects the proportional gain of the original system. The larger its value, the greater the proportional gain of the original system; wind force adaptation coefficient α F It reflects the influence of wind disturbance on proportional gain adjustment. The larger the value, the more significant the influence of wind disturbance on proportional gain adjustment, and the more sensitive and flexible the system is in responding to wind disturbance. It is a sigmoid function, which is used to make the system response a gradual process. Since wind interference is generally the greatest interference to the operation of the crane, the interference degree may be very large. If the system immediately responds to the interference to the same degree, the sudden strong control is likely to cause safety problems. In addition, since wind interference may decrease instantly, if this happens, the sudden strong control command will cause greater interference to the system, so the sigmoid function is used. Make the system response a gradual process to prevent the above situation from happening; wind interference coefficient Reflects the impact of wind interference on the system. The larger the value, the greater the interference of wind interference on the system.

[0091] The swing disturbance is adjusted by adjusting the integral gain, based on the formula:

[0092]

[0093] Among them, K Inew is the adjusted integral gain, K Ibase is the original integral gain, is the swing interference coefficient, α B is the swing adaptive coefficient. This formula can realize the adaptive adjustment of the system to the swing disturbance. The adjusted integral gain K Inew It reflects the adjustment of the integral control parameters when the system faces swing disturbance. The adjusted integral gain K Inew The larger the value, the more active and flexible the system is in adjusting the integral control parameters. Ibase Reflects the differential gain of the original system. The larger its value, the greater the differential gain in the original system. The swing adaptive coefficient α B It reflects the degree of adjustment of the integral control parameters when the system faces swing disturbance. The swing adaptive coefficient α B The larger the value, the more actively the system adjusts the integral control parameters when it predicts the existence of swing disturbance, so as to more effectively offset the influence of swing disturbance on the system; Swing disturbance coefficient It reflects the impact of swing interference on the system. The larger the value, the greater the interference caused by the swing interference to the system.

[0094] The vibration disturbance is adjusted by adjusting the differential gain, based on the formula:

[0095]

[0096] Among them, K dnew is the adjusted differential gain, K dbase is the original differential gain, α F is the vibration adaptive coefficient, is the vibration interference coefficient. This formula is constructed to achieve the adaptive adjustment of the system to vibration interference; the adjusted differential gain K dnew It reflects the degree of adjustment of the differential gain parameter when the system faces vibration disturbance. The larger the value, the stronger the system response to the vibration disturbance. That is, when the system predicts the existence of vibration disturbance, it will adjust the differential gain parameter more actively to more effectively offset the influence of vibration disturbance on the system. The original differential gain K dbase Reflects the differential gain of the original system. The larger its value, the greater the differential gain in the original system; vibration adaptive coefficient α F It reflects the degree of adjustment of the differential gain parameter by the system when facing vibration interference. The larger the value, the more actively the system will adjust the differential gain parameter when it predicts the existence of vibration interference, so as to more effectively offset the influence of vibration interference on the system. Reflects the impact of vibration interference on the system. The larger the value, the greater the interference of vibration interference on the system.

[0097] Step 6: Control the output of the crane controller by adjusting the proportional gain, derivative gain and integral gain.

[0098] In this embodiment, the specific formula for obtaining the output of the crane controller is:

[0099]

[0100] Where u(t) is the controller output, K Pnew is the adjusted proportional gain, K Inew is the adjusted integral gain, K dnew is the adjusted differential gain, e(t) is the error signal, and t represents the time variable. This is the output formula of the adjusted PID controller. The significance of constructing this formula is to adjust the proportion K adaptively. Pnew , integral K Inew and differential K dnew Gain enables the control system to better cope with various disturbances (such as wind, swing and vibration disturbances), thereby improving the stability and control performance of the system; u(t) is the output of the controller, which is used to adjust the input of the system to reduce errors and achieve system goals. The larger the u(t), the greater the adjustment of the controller output, and the more drastic the system's correction action on the error. e(t) is the error signal, which represents the difference between the expected value and the actual value. The larger its value, the greater the deviation between the actual output and the expected value, and the more the system needs to make adjustments; K Pnew e(t) proportional control part, K Pnew e(t) reflects the adjustment strength of the current error of the system. The adjusted proportional gain enables the system to flexibly respond to the current error; The integral control part reflects the adjustment strength of the system's past accumulated errors. The adjusted integral gain can eliminate steady-state errors and improve the long-term accuracy of the system; The differential control part reflects the adjustment strength of the system error change rate. The adjusted differential gain can predict the error trend and slow down the fluctuation and oscillation of the system response.

[0101] See also Figure 2 The present invention also provides an artificial intelligence driven crane adaptive anti-interference control system, which is used to implement the above artificial intelligence driven crane adaptive anti-interference control method, including:

[0102] The acquisition module is used to collect multiple groups of historical data of the crane during operation to form a historical data set, each group of data includes the weight, the cross-sectional area of ​​the weight, the wind speed, the air density, the weight swing angle, the swing angular velocity, the length of the suspension rope, the boom vibration frequency and amplitude; the swing interference coefficient is generated according to the weight swing angle and the swing angular velocity in each group of data;

[0103] A modeling optimization module is used to divide 80% of the data set into a training set and 20% into a validation set according to the historical data set, build a crane interference prediction model based on a neural network, and use the swing interference coefficient as a label; use the training set to train the crane interference prediction model, and after the training, use the validation set to evaluate the crane interference prediction model, adjust the hyperparameters of the crane interference prediction model according to the evaluation results, and generate a swing interference prediction model;

[0104] A prediction module is used to collect crane working data in real time, wherein the working data includes weight mass, cross-sectional area of ​​weight, wind speed, air density, length of suspension rope, boom vibration frequency and amplitude; the crane working data is input into a swing interference prediction model to generate a swing interference coefficient;

[0105] An analysis module is used to generate a wind interference coefficient through the weight mass, the cross-sectional area of ​​the weight, the wind speed and the air density in the crane working data collected in real time; and to generate a vibration interference coefficient through the boom vibration frequency and amplitude in the crane working data collected in real time;

[0106] A gain adjustment module is used to obtain the proportional gain, differential gain and integral gain in the PID control system of the real-time crane, adjust the proportional gain of the system through the wind interference coefficient, adjust the integral gain of the system through the swing interference coefficient, and adjust the differential gain of the system through the vibration interference coefficient;

[0107] Output control module, used to control the output of the crane controller through adjusted proportional gain, derivative gain and integral gain.

[0108] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0109] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. An artificial intelligence driven crane adaptive anti-interference control method, characterized in that: The following steps are involved: Step 1: Collect multiple sets of historical data of crane operation to form a historical data set. Each set of data includes weight mass, weight cross-sectional area, wind speed, air density, weight swing angle, swing angular velocity, rope length, boom vibration frequency and amplitude; generate a swing interference coefficient based on the weight swing angle and swing angular velocity in each set of data; Step 2: Divide 80% of the data set into a training set and 20% into a validation set based on the historical data set, build a crane interference prediction model based on a neural network, and use the swing interference coefficient as a label; The crane interference prediction model is trained using the training set. After the training, the crane interference prediction model is evaluated using the validation set. The hyperparameters of the crane interference prediction model are adjusted according to the evaluation results to generate a swing interference prediction model. Step 3: Collecting crane working data in real time, the working data including weight mass, weight cross-sectional area, wind speed, air density, suspension rope length, boom vibration frequency and amplitude; inputting the crane working data into the swing interference prediction model to generate a swing interference coefficient; Step 4: Generate a wind interference coefficient through the weight mass, weight cross-sectional area, wind speed and air density in the real-time collected crane working data; generate a vibration interference coefficient through the boom vibration frequency and amplitude in the real-time collected crane working data; Step 5: Obtain the proportional gain, differential gain and integral gain in the PID control system of the real-time crane, adjust the proportional gain of the system by the wind disturbance coefficient, adjust the integral gain of the system by the swing disturbance coefficient, and adjust the differential gain of the system by the vibration disturbance coefficient; Step 6: Control the output of the crane controller by adjusting the proportional gain, derivative gain and integral gain.

2. The method for adaptive anti-interference control of a crane driven by artificial intelligence according to claim 1 is characterized in that: The specific logic for generating the swing interference coefficient is: The weights are assigned to the swing angle and the square of the swing angular velocity of the weight to generate the swing interference coefficient. The specific formula is as follows: in, is the swing interference coefficient, θ is the swing angle of the weight, ω is the angular velocity of the weight, α is the angle weight coefficient, β is the angular velocity weight coefficient, α+β=1, 0<β<0.1α<α<1.

3. The method for adaptive anti-interference control of a crane driven by artificial intelligence according to claim 2 is characterized in that: The specific logic for choosing mean square error as the loss function when optimizing the crane interference prediction model is: for each parameter, calculate the error between the predicted value and the actual value, square the error of each sample, sum all the squared errors and take the average; the specific formula is: MSE is the mean square error, n is the number of samples, y i is the actual swing interference coefficient of the i-th sample, is the predicted swing interference coefficient of the i-th sample.

4. The method for adaptive anti-interference control of a crane driven by artificial intelligence according to claim 3 is characterized in that: The specific logic for generating the wind interference coefficient is: The wind force is obtained by approximating the wind speed by using wind speed, air density, wind force coefficient and cross-sectional area of ​​the weight; the specific formula for obtaining the wind force is: Among them, F is wind force, V is wind speed, ρ is air density, C d is the wind force coefficient, M is the cross-sectional area of ​​the weight; Then, the wind interference coefficient is generated according to the wind force and the mass of the heavy object. The specific formula for generating the wind interference coefficient is: in, is the wind disturbance coefficient, m is the mass of the heavy object, and g is the acceleration of gravity.

5. The method for adaptive anti-interference control of a crane driven by artificial intelligence according to claim 4 is characterized in that: The specific logic for generating the vibration interference coefficient is to generate the vibration interference coefficient through the boom vibration frequency and amplitude; the specific formula is: in, is the vibration interference coefficient, A is the vibration amplitude of the boom, and f is the vibration frequency of the boom.

6. The method for adaptive anti-interference control of a crane driven by artificial intelligence according to claim 5 is characterized in that: Wind disturbance is regulated by adjusting the proportional gain in the controller. The proportional gain is adjusted based on the formula: Among them, K Pnew is the adjusted proportional gain, K Pbase is the original proportional gain, α F is the wind adaptation coefficient, is the wind interference coefficient; The swing disturbance is adjusted by adjusting the integral gain, based on the formula: Among them, K Inew is the adjusted integral gain, K Ibase is the original integral gain, is the swing interference coefficient; α B is the swing adaptation coefficient; The vibration disturbance is adjusted by adjusting the differential gain, based on the formula: Among them, K dnew is the adjusted differential gain, K dbase is the original differential gain, α F is the vibration adaptive coefficient, is the vibration interference coefficient.

7. The method for adaptive anti-interference control of a crane driven by artificial intelligence according to claim 6 is characterized in that: The specific formula based on which the output of the crane controller is obtained is: Where u(t) is the controller output, K Pnew is the adjusted proportional gain, K Inew is the adjusted integral gain, K dnew is the adjusted differential gain, e(t) is the error signal, and t represents the time variable.

8. An artificial intelligence driven crane adaptive anti-interference control system, the system is used to implement the crane adaptive anti-interference control method according to any one of claims 1 to 7, comprising: The acquisition module is used to collect multiple groups of historical data of the crane during operation to form a historical data set, each group of data includes the weight, the cross-sectional area of ​​the weight, the wind speed, the air density, the weight swing angle, the swing angular velocity, the length of the suspension rope, the boom vibration frequency and amplitude; the swing interference coefficient is generated according to the weight swing angle and the swing angular velocity in each group of data; Modeling optimization module, which is used to divide 80% of the data set into a training set and 20% into a validation set based on the historical data set, build a crane interference prediction model based on a neural network, and use the swing interference coefficient as a label; The crane interference prediction model is trained using the training set. After the training, the crane interference prediction model is evaluated using the validation set. The hyperparameters of the crane interference prediction model are adjusted according to the evaluation results to generate a swing interference prediction model. A prediction module is used to collect crane working data in real time, wherein the working data includes weight mass, cross-sectional area of ​​weight, wind speed, air density, length of suspension rope, boom vibration frequency and amplitude; the crane working data is input into a swing interference prediction model to generate a swing interference coefficient; An analysis module is used to generate a wind interference coefficient through the weight mass, the cross-sectional area of ​​the weight, the wind speed and the air density in the crane working data collected in real time; and to generate a vibration interference coefficient through the boom vibration frequency and amplitude in the crane working data collected in real time; A gain adjustment module is used to obtain the proportional gain, differential gain and integral gain in the PID control system of the real-time crane, adjust the proportional gain of the system through the wind interference coefficient, adjust the integral gain of the system through the swing interference coefficient, and adjust the differential gain of the system through the vibration interference coefficient; Output control module, used to control the output of the crane controller through adjusted proportional gain, derivative gain and integral gain.

Images