A method for synchronously controlling multiple motors of a stranding machine

Abstract

The invention discloses a method for synchronously controlling multiple motors of a stranding machine, which relates to the field of electrical engineering technology, and comprises: equipping each motor with a high-precision encoder to monitor the rotation speed and position information in real time, and installing a sensor network to monitor the wire state and mechanical dynamic parameters; constructing a nonlinear dynamic model of the multiple motors of the stranding machine, applying the nonlinear model predictive control (NMPC) technology, predicting the future system state according to the nonlinear dynamic model and calculating the control instruction sequence; predicting the control strategy based on a deep neural network of historical data and combining it with a machine learning algorithm, and adaptively adjusting the load mutation and nonlinear effect; designing a dynamic weight distribution algorithm, and dynamically adjusting the weight parameters of each motor control strategy according to the real-time working condition analysis result; implementing the calculated control instruction to the motor controller, and using the feedback information of the encoder and the sensor network for closed-loop adjustment, optimizing the control strategy in real time according to the error, and maintaining highly synchronous stranding operation.

Description

Technical Field

[0001] The invention relates to the technical field of electrical engineering, and in particular to a method for synchronously controlling multiple motors of a stranding machine. Background Art

[0002] In the field of wire and cable manufacturing, the stranding machine is a core processing equipment, and its function and performance are directly related to the quality and production efficiency of cable products. With the rapid development of industrial automation technology, the stranding machine control system has evolved from a single motor drive to multi-motor synchronous control. In order to achieve more precise wire winding and higher production rates, traditional multi-motor synchronous control mainly relies on PID control strategy to maintain the consistency of line speed by adjusting the motor speed. However, this control method is often difficult to achieve ideal synchronous control effects when faced with complex nonlinear dynamic characteristics, load mutations and changes in the physical properties of wires, resulting in unstable cable quality and limited production efficiency.

[0003] Especially for high-end cable products, including communication cables and special cables, the twisting process requires extremely high precision. Minor synchronization errors will lead to cable structural defects and affect the final transmission performance. The shortcomings of existing technologies are mainly reflected in the following aspects: First, traditional control strategies lack comprehensive consideration of the nonlinear characteristics of the system, and are inaccurate in prediction and response to synchronization deviations caused by changes in wire tension, friction fluctuations and dynamic coupling effects between motors; second, the control strategy is not adaptable, and in the face of sudden load changes in the production process, including differences in wire material and wire diameter changes, it is impossible to make effective adjustments quickly, affecting the continuity and stability of production. The existing system is not efficient in utilizing real-time data, and lacks an effective mechanism to integrate historical data with real-time monitoring information to optimize the control strategy. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a multi-motor synchronous control method for a stranding machine to solve the problems that traditional control strategies lack comprehensive consideration of the nonlinear characteristics of the system, inaccurate predictions, and synchronization deviations caused by changes in wire tension, friction fluctuations, and dynamic coupling effects between motors; secondly, the control strategy is not adaptable, and in the face of sudden load changes in the production process, including differences in wire material and wire diameter, it is impossible to make effective adjustments quickly, affecting the continuity and stability of production. The existing system has low efficiency in utilizing real-time data, and lacks an effective mechanism to integrate historical data with real-time monitoring information to optimize the control strategy.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for synchronously controlling multiple motors of a stranding machine, which includes equipping each motor with a high-precision encoder to monitor the rotation speed and position information in real time, and installing a sensor network to monitor the wire state and mechanical dynamic parameters;

[0008] Construct a nonlinear dynamic model of the stranding machine's multiple motors, apply nonlinear model predictive control (NMPC) technology, predict the future system state based on the nonlinear dynamic model and calculate the control instruction sequence;

[0009] Deep neural networks based on historical data combined with machine learning algorithms, predictive control strategies, and adaptive adjustments to load mutations and nonlinear effects;

[0010] Design a dynamic weight allocation algorithm to dynamically adjust the weight parameters of each motor control strategy based on real-time operating condition analysis results;

[0011] The calculated control instructions are implemented to the motor controller, and the feedback information from the encoder and sensor network is used for closed-loop adjustment. The control strategy is optimized in real time according to the error to maintain highly synchronized stranding operations.

[0012] As a preferred solution of the multi-motor synchronous control method of the stranding machine of the present invention, wherein: a high-precision encoder is equipped on each motor to monitor the speed and position information in real time, and a sensor network is installed to monitor the wire state and mechanical dynamic parameters. The specific steps are:

[0013] Install piezoelectric force sensors and laser rangefinders at key locations on the stranding machine to monitor wire tension and displacement;

[0014] Monitor the working environment through temperature and humidity sensors;

[0015] The comprehensive index function is constructed by using the information of motor speed, position, wire tension and displacement, and the expression is:

[0016] ;

[0017] Among them, I s represents the comprehensive index function, n represents the number of motors, v m,i represents the real-time speed of the i-th motor, Represents the average speed of all motors, P m,i represents the position information of the i-th motor, represents the average value of the motor position, σ p Indicates the standard deviation of the motor position, F t,i represents the real-time tension of the i-th wire, represents the average value of wire tension, σ F Indicates the standard deviation of tension, D irepresents the real-time displacement of the i-th wire, represents the average value of wire displacement, σ D represents the standard deviation of displacement, λ represents the tension influence factor, and η represents the weight parameter of displacement deviation.

[0018] As a preferred solution of the method for synchronously controlling multiple motors of a stranding machine of the present invention, the nonlinear dynamic model of the multiple motors of the stranding machine is constructed, and the nonlinear model predictive control (NMPC) technology is used to predict the future system state and calculate the control instruction sequence according to the nonlinear dynamic model. The specific steps are as follows:

[0019] A nonlinear dynamic model is constructed based on motor torque, wire tension, friction, inertia and nonlinear dynamic functions. The expression is:

[0020] ;

[0021] in, represents the rate of change of the system state over time, f represents a nonlinear function, x(t) represents all necessary variables of the current state, u(t) represents the control vector, and d(t) is the external disturbance vector;

[0022] Based on the nonlinear dynamics model, a predictive control model is constructed to predict the state of the future time step Np. The expression is:

[0023] ;

[0024] Among them, k represents the current time step, t k represents the time point of the current time step, x k+e|k Indicates that at time step k, the state of the future e-th time step is predicted based on the current state, x k represents the actual state of the system at time step k, △t represents the sampling time interval;

[0025] The objective function is constructed based on the predictive control model, and the expression is:

[0026] ;

[0027] Among them, J represents the value of the objective function, r k+i represents the target state of the k+ith time step in the future, u represents the control vector, and u k+i-1|k represents the control input predicted at time step k when executing the i-1th control action, w1 and w2 are the weight coefficients of the state error and the control input respectively;

[0028] Using sequential quadratic programming SQP, under given constraints, the control sequence that minimizes the objective function J is solved. The control instruction expression is:

[0029] ;

[0030] Among them, N c To control the horizon, it indicates the number of future time steps considered in the optimization process, u k|k It means the control instruction formulated at time k, Indicates that at time k, the future Step pre-calculated control instructions;

[0031] According to the first step control instruction u obtained by optimization k|k , and send it to the motor controller for control.

[0032] As a preferred solution of the multi-motor synchronous control method of the stranding machine of the present invention, the deep neural network based on historical data is combined with a machine learning algorithm to predict the control strategy, and the specific steps are as follows:

[0033] Collect the operation data of the stranding machine under different working conditions, including motor speed, position, wire tension, displacement and environmental parameters;

[0034] Sort the collected data by timestamp and store them as a time series dataset;

[0035] Extract the motor speed change rate and tension fluctuation amplitude from the time series data set;

[0036] The motor speed change rate is expressed as:

[0037] ;

[0038] Wherein, △v represents the rate of change of motor speed, v(t) represents the motor speed at the current time t, v(t-△t) represents the motor speed at the previous time t-△t, and △t represents the time interval between two sampling time points;

[0039] The tension fluctuation amplitude is expressed as:

[0040] ;

[0041] Among them, δF represents the tension fluctuation amplitude, F(t) represents the tension value at time t, △t represents the time interval, and F(t-△t) represents the tension value in a small time period △t before time t;

[0042] The SN model is constructed by combining the long short-term memory network LSTM and the convolutional neural network CNN;

[0043] The extracted motor speed change rate, tension fluctuation amplitude, historical control instructions and external disturbances are input into the SN model, and the desired speed and target tension of the motor are output;

[0044] Set the previous moment's historical control command to C p , the external disturbance at the current moment is D;

[0045] After calculating the processed input vector, the SN model expression is:

[0046] ;

[0047] Where IP represents the preprocessed integrated input vector, w △v ,w δf ,w C ,w D They correspond to the motor speed change rate, tension fluctuation amplitude, historical control instructions, and weight coefficients of external disturbances, respectively, and sgn is a sign function;

[0048] The output layer of the SN model is expressed as:

[0049] ;

[0050] Among them, v d represents the desired motor speed output, T d Indicates the target tension output.

[0051] As a preferred solution of the method for synchronously controlling multiple motors of a stranding machine of the present invention, the adaptive adjustment of load mutation and nonlinear effect comprises the following specific steps:

[0052] Collect real-time data to accurately monitor motor speed, position, wire tension, displacement and environmental parameters;

[0053] By monitoring the rate of change of the motor current, it is possible to determine whether a load has occurred. The expression for the rate of change is:

[0054] ;

[0055] Wherein, △I / △t represents the instantaneous rate of change, △I represents the rate of change of current, △t represents the time interval, I(t) represents the current value at time t, and I(t-△t) represents the current value at time t-△t;

[0056] Calculate the average value of all instantaneous change rates △I / △t and standard deviation σ △I / △t ;

[0057] Once a sudden load change is detected, the parameter re-evaluation of the nonlinear dynamic model is triggered immediately, and the friction coefficient and material hardness coefficient in the model are adjusted according to the current working conditions;

[0058] Iteratively optimize the parameters f(x(t),u(t),d(t)) in the model based on the latest data.

[0059] As a preferred solution of the multi-motor synchronous control method of the stranding machine of the present invention, the dynamic weight allocation algorithm is designed to dynamically adjust the weight parameters of each motor control strategy according to the real-time working condition analysis results. The specific steps are as follows:

[0060] Extract key parameters from sensor networks in real time;

[0061] According to the extracted parameters, the real-time working conditions are divided into three categories: normal operation, mild load mutation, and severe load mutation;

[0062] Construct a feature vector for each category, the expression is:

[0063] X=[H, v m ,T, D I / D t ¯ , s D I / D t ,δF] ;

[0064] Among them, X represents the characteristic vector of the current working condition of the stranding machine, H represents the hardness of the wire, and v m Indicates the running speed of the motor, T indicates the ambient temperature;

[0065] Based on X and combined with the dynamic weight allocation function, the MLP model is constructed to automatically adjust the weight parameters w of each control strategy △v ,w δf ,w C ,w D , the MLP model expression is:

[0066] W ( X )= MLP ([ w D v , w δF , w C , w D ] 0 ; X ) ;

[0067] Where w(x) represents the dynamic weight assignment function, MLP represents the multi-layer perceptron, [w △v ,w δf ,w C ,w D ]0 represents the initial weight vector;

[0068] Use historical operating data and known optimal control strategies to input the MLP model to complete training;

[0069] Input the dynamically adjusted weights into the SN model to obtain a new input vector IP a , the adjusted SN model expression is:

[0070] I P a =W(X) ⋅ [ w D v ⋅ log (1+| D v|), w δF ⋅ δF , w C ⋅ C previous , w D ⋅ sgn(D) ⋅ ln (1+|D|)] .

[0071] As a preferred solution of the multi-motor synchronous control method of the stranding machine of the present invention, wherein: the implementation of the calculated control instructions to the motor controller and the closed-loop adjustment using the feedback information of the encoder and the sensor network are specifically performed as follows;

[0072] The control instructions obtained through nonlinear model predictive control (NMPC) technology and deep learning assisted optimization Sent to the controller of each motor;

[0073] Through high-precision encoders, force sensors and displacement sensors, the motor speed, position, wire tension, displacement and environmental parameters are collected in real time.

[0074] As a preferred solution of the multi-motor synchronous control method of the stranding machine of the present invention, wherein: the real-time optimization control strategy based on the error is used to maintain a highly synchronized stranding operation, and the specific steps are as follows:

[0075] Input the real-time data into the nonlinear dynamics model to obtain the current state of the system, compare the current state of the system with the target state to obtain the actual error e(t), and update the comprehensive index function I S The value of

[0076] Using the dynamic weight allocation algorithm, the weight parameters of the control strategy are automatically optimized through the MLP model according to the current operating condition feature vector X;

[0077] Combined with the updated weight parameters and SN model, the optimized control instructions are recalculated and transmitted to the motor controller to ensure the best match between the control instructions and the current system state;

[0078] After the control command is implemented, the control strategy is continuously adjusted according to the real-time information of the encoder and sensor network until the error is reduced, so that the system maintains a highly synchronized state.

[0079] In a second aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the multi-motor synchronous control method of a stranding machine as described in the first aspect of the present invention is implemented.

[0080] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the multi-motor synchronous control method for a stranding machine as described in the first aspect of the present invention is implemented.

[0081] The beneficial effects of the present invention are as follows: the present invention realizes precise real-time monitoring of the motor and wire status by integrating high-precision sensing and nonlinear dynamic modeling, and predicts the system status with the help of NMPC technology to generate the optimal control instruction sequence. At the same time, it innovatively integrates deep learning and machine learning algorithms, dynamically predicts control strategies based on historical data, effectively responds to sudden loads and nonlinear challenges, improves system adaptability and control accuracy, and can instantly adjust control weights according to changes in working conditions through a dynamic weight algorithm to ensure that the strategy is highly consistent with the actual situation. The closed-loop feedback system continuously optimizes the control strategy to ensure excellent synchronization of the stranding operation, overcoming the problems of inaccurate synchronization and poor adaptability of traditional methods in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0083] Figure 1 This is a flow chart of the multi-motor synchronous control method of the stranding machine in Example 1.

[0084] Figure 2 This is a flow chart of the judgment of the comprehensive index function in Example 1. DETAILED DESCRIPTION

[0085] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0086] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for synchronously controlling multiple motors of a stranding machine, comprising the following steps:

[0087] S1 is equipped with high-precision encoders on each motor to monitor speed and position information in real time, and a sensor network is installed to monitor wire status and mechanical dynamic parameters;

[0088] Each motor shaft is equipped with a magnetic high-precision encoder to ensure that the resolution is not less than 10,000 pulses per revolution, and the motor speed and accumulated pulse number are recorded in real time to calculate the absolute position;

[0089] The encoder signal is transmitted to the central processing unit CPU in real time through the digital interface CAN;

[0090] Install piezoelectric force sensors and laser rangefinders at key locations on the stranding machine to monitor wire tension and displacement;

[0091] Monitor the working environment through temperature and humidity sensors to ensure data accuracy is not affected by environmental factors;

[0092] All sensor data is collected to the CPU via wireless network, with a frequency of no less than 1kHz;

[0093] A comprehensive index function is constructed through the information of motor speed, position, wire tension and displacement to evaluate the immediate effect of synchronous control. The expression is:

[0094] ;

[0095] Among them, I s It represents the comprehensive index function, which reflects the efficiency and quality of multi-motor synchronous control. The smaller the value, the better the synchronization effect. n represents the number of motors. V m,i represents the real-time speed of the i-th motor, Represents the average speed of all motors, P m,i Indicates the position information of the i-th motor, expressed as the number of encoder pulses, reflecting the absolute position of the motor shaft. represents the average value of the motor position, σ p Indicates the standard deviation of the motor position, F t,i represents the real-time tension of the i-th wire, represents the average value of wire tension, σ F Indicates the standard deviation of tension, D i represents the real-time displacement of the i-th wire, represents the average value of wire displacement, σ D represents the standard deviation of displacement, λ represents the tension influence factor, which is used to balance the weight of the influence of speed and tension deviation on synchronization performance, and η represents the weight parameter of displacement deviation, which is used to emphasize the influence of displacement inconsistency on the overall synchronization performance;

[0096] The value range of Is is [ - 1,+ ∞ ] , the closer the value is to -1, the better the synchronization effect;

[0097] When Is<0, it indicates that the motor speed, position and wire state are basically synchronized. Carefully adjust the control parameters to optimize the synchronization accuracy.

[0098] When Is=0, it indicates perfect synchronization, maintaining the current control settings to ensure continuous optimal state;

[0099] When Is When it is 0, it means that there is an obvious synchronization deviation, and the control command is quickly adjusted according to the prediction model to correct the deviation;

[0100] The key parts of the stranding machine include but are not limited to the pay-off frame, stranding wheel, distribution drum and stranding die.

[0101] S2 builds a nonlinear dynamic model of the stranding machine's multiple motors and uses the nonlinear model predictive control (NMPC) technology to predict future system states and calculate control instruction sequences based on the nonlinear dynamic model.

[0102] A nonlinear dynamic model is constructed based on motor torque, wire tension, friction, inertia and nonlinear dynamic functions. The expression is:

[0103] ;

[0104] in, represents the rate of change of the system state over time, f represents a nonlinear function, x(t) represents all necessary variables of the current state, including but not limited to motor speed, motor position, wire tension, wire displacement and motor current, u(t) represents the control vector, and d(t) is the external disturbance vector;

[0105] Based on the nonlinear dynamics model, a predictive control model is constructed to predict the state of the future time step Np. The expression is:

[0106] ;

[0107] Among them, k represents the current time step, t k represents the time point of the current time step, x k+e|k Indicates that at time step k, the state of the future e-th time step is predicted based on the current state, x k represents the actual state of the system at time step k, i.e., the state vector, and △t represents the sampling time interval;

[0108] The objective function is constructed based on the predictive control model, and the expression is:

[0109] ;

[0110] Among them, J represents the value of the objective function, which measures the quality of the control performance. The smaller the value, the closer the control effect is to the ideal state. Indicates starting from i=1 to N p The end of the accumulation process, r k +i represents the target state of the k+ith time step in the future, u represents the control input vector, and u k +i|k represents the control input predicted at time step k when executing the i-1th control action, w1 and w2 are the weight coefficients of the state error and the control input, respectively;

[0111] Using sequential quadratic programming SQP, under given constraints, solve the objective function The minimum control sequence and control instruction expression are:

[0112] ;

[0113] Among them, N c To control the horizon, it indicates the number of future time steps considered in the optimization process, u k|k It means the control instruction formulated at time k, Indicates that at time k, the future Step pre-calculated control instructions;

[0114] According to the first step control instruction u obtained by optimization k|k , send it to the motor controller to implement control;

[0115] The nonlinear function f is obtained by:

[0116] Set the collected sample data to:

[0117] ;

[0118] Where N is the number of data points;

[0119] Assume that f can be expressed by a set of parameters θ, that is , the expression of the nonlinear function f is:

[0120] ;

[0121] Among them, θ 1、 θ 2、 θ3 represents the parameter to be estimated, n, m, l are the powers of each term, and the interaction term considers the interaction between state, control and disturbance;

[0122] External disturbance vector d(t), including but not limited to load change, ambient temperature change and friction change;

[0123] Constraints, including but not limited to motor maximum torque and wire tension limits;

[0124] The value range of the control command u depends on the specific motor characteristics and reflects the actual control intensity that can be applied. It must be ensured to be within the safe operating range of the motor.

[0125] S3 uses a deep neural network based on historical data and combines machine learning algorithms, predictive control strategies, and adaptive adjustments to load mutations and nonlinear effects;

[0126] Collect the operation data of the stranding machine under different working conditions, including motor speed, position, wire tension, displacement and environmental parameters;

[0127] Sort the collected data by timestamp and store them as a time series dataset;

[0128] Extract the motor speed change rate and tension fluctuation amplitude from the time series data set;

[0129] The motor speed change rate is expressed as:

[0130] ;

[0131] Wherein, △v represents the rate of change of motor speed, v(t) represents the motor speed at the current time t, v(t-△t) represents the motor speed at the previous time t-△t, and △t represents the time interval between two sampling time points;

[0132] The tension fluctuation amplitude is expressed as:

[0133] ;

[0134] Among them, δF represents the tension fluctuation amplitude, F(t) represents the tension value at time t, △t represents the time interval, and F(t-△t) represents the tension value in a small time period △t before time t;

[0135] The SN model is constructed by combining the long short-term memory network LSTM and the convolutional neural network CNN;

[0136] The extracted motor speed change rate, tension fluctuation amplitude, historical control instructions and external disturbances are input into the SN model, and the desired speed and target tension of the motor are output;

[0137] Set the historical control instruction of the previous moment to , the external disturbance at the current moment is D;

[0138] After calculating the processed input vector, the SN model expression is:

[0139] ;

[0140] Where IP represents the preprocessed integrated input vector, w △v ,w δF ,w C ,w D They correspond to the weight coefficients of the motor speed change rate, tension fluctuation amplitude, historical control instructions, and external disturbances, respectively. Log and ln represent natural logarithmic functions, which are used to increase the nonlinear response to the speed change rate. Sgn is a sign function. Sgn(D) means that if D>0, it is equal to 1, if D<0, it is equal to -1, and if D=0, it is equal to 0.

[0141] The output layer of the SN model is expressed as:

[0142] ;

[0143] Among them, v d represents the desired motor speed output, T d Indicates the target tension output;

[0144] Environmental parameters, including temperature, humidity, air pressure, and light intensity;

[0145] The SN model can not only accurately predict the future state of the system, but also flexibly adapt to environmental changes and optimize control strategies;

[0146] The external disturbance at the current moment is D, expressed as: D = d(t);

[0147] Collect real-time data to accurately monitor motor speed, position, wire tension, displacement and environmental parameters;

[0148] By monitoring the rate of change of the motor current, it is possible to determine whether a load has occurred. The expression for the rate of change is:

[0149] ;

[0150] Wherein, △I / △t represents the instantaneous rate of change, △I represents the rate of change of current, △t represents the time interval, I(t) represents the current value at time t, and I(t-△t) represents the current value at time t-△t;

[0151] Calculate the average value of all instantaneous change rates △I / △t and standard deviation σ △I / △t ;

[0152] Set the dynamic threshold T d , combined with the historical operation data of the stranding machine, especially the distribution of the motor current change rate, the dynamic threshold T d Can be set to the mean plus 2 times the standard deviation;

[0153] When △I / △t <T d When , it means that the motor is within the normal load variation range and the current control strategy remains unchanged;

[0154] When T d ≤△I / △t <T d When x2, it indicates that there is a slight load mutation on the motor, and the initial adaptive control adjustment is initiated, including slight adjustment of the motor output torque and speed, while strengthening monitoring;

[0155] When △I / △t>T d When x2, it means that the motor has a severe load mutation, and the emergency control strategy should be immediately implemented, including a significant adjustment of the control instructions to ensure system stability and send an alarm to the operator;

[0156] Once a sudden load change is detected, the parameter re-evaluation of the nonlinear dynamic model is triggered immediately, and the friction coefficient and material hardness coefficient in the model are adjusted according to the current working conditions;

[0157] Iteratively optimize the parameters f(x(t),u(t),d(t) in the model according to the latest data;

[0158] Based on the updated nonlinear dynamic model, the weight coefficients w1 and w2 in the objective function J of NMPC are adjusted in real time to adapt to the control requirements after the load mutation.

[0159] The current motor speed change rate, tension fluctuation amplitude, historical control instructions and adjusted external disturbances are input into the SN model to continuously learn and predict the best control strategy;

[0160] According to the real-time monitoring of load change rate, tension and position fluctuations, the w input to the SN model is dynamically adjusted. △v ,w δf ,w C ,w D , used to adapt to current working conditions;

[0161] The calculated control command is sent to the motor controller, and the control strategy is adjusted in a closed loop through real-time feedback from the encoder and sensor network. S , continuously optimize control instructions to ensure that the system maintains a highly synchronized stranding operation state;

[0162] System response, including actual changes in motor speed, actual fluctuations in wire tension, and actual movement of motor position.

[0163] S4 designs a dynamic weight allocation algorithm to dynamically adjust the weight parameters of each motor control strategy based on real-time working condition analysis results;

[0164] Extract key parameters from sensor networks in real time;

[0165] According to the extracted parameters, the real-time working conditions are divided into three categories: normal operation, mild load mutation, and severe load mutation;

[0166] Construct a feature vector for each category, the expression is:

[0167] X=[H, v m ,T, D I / D t ¯ , s D I / D t ,δF] ;

[0168] Among them, X represents the characteristic vector of the current working condition of the stranding machine, H represents the hardness of the wire, and v m Indicates the running speed of the motor, T indicates the ambient temperature;

[0169] Based on X and combined with the dynamic weight allocation function, the MLP model is constructed to automatically adjust the weight parameters w of each control strategy △v ,w δf ,w C ,w D , the MLP model expression is:

[0170] W ( X )= MLP ([ w D v , w δF , w C , w D ] 0 ; X ) ;

[0171] Where w(x) represents the dynamic weight assignment function, MLP represents the multi-layer perceptron, [w △v ,w δF ,w C ,w D ]0 represents the initial weight vector;

[0172] Use historical operating data and known optimal control strategies to input the MLP model to complete training;

[0173] Input the dynamically adjusted weights into the SN model to obtain a new input vector IP a , the adjusted SN model expression is:

[0174] I P a =W(X) ⋅ [ w D v ⋅ log (1+| D v|), w δF ⋅ δF , w C ⋅ C previous , w D ⋅ sgn(D) ⋅ ln (1+|D|)] ;

[0175] Key parameters, including but not limited to wire hardness, motor speed, ambient temperature, tension fluctuation amplitude, and the average and standard deviation of the motor current change rate, reflect the real-time working conditions of the current system and are the basis for dynamic weight adjustment;

[0176] The output expression of the dynamic weight allocation function w(x) is:

[0177] [ w D v ' , w δF ' , w C ' , w D ' ]=W([H, v m ,T, D I / D t ¯ , s D I / D t ,δF]) ;

[0178] in, They are respectively represented as the adjusted motor speed change rate, tension fluctuation amplitude, historical control instructions, and dynamic weight coefficients of external disturbances.

[0179] S5 implements the calculated control instructions to the motor controller and uses the feedback information from the encoder and sensor network to perform closed-loop adjustments, optimizing the control strategy in real time based on the error to maintain highly synchronized twisting operations;

[0180] The control instructions obtained through nonlinear model predictive control (NMPC) technology and deep learning assisted optimization Sent to the controller of each motor;

[0181] Through high-precision encoders, force sensors and displacement sensors, the motor speed, position, wire tension, displacement and environmental parameters are collected in real time;

[0182] Ensure that the acquisition frequency is no less than 1kHz to ensure the real-time and accuracy of the data;

[0183] Input the real-time data into the nonlinear dynamics model to obtain the current state of the system, compare the current state of the system with the target state to obtain the actual error e(t), and update the comprehensive index function I S The value of

[0184] Using the dynamic weight allocation algorithm, the weight parameters of the control strategy are automatically optimized through the MLP model according to the current operating condition feature vector X;

[0185] Combined with the updated weight parameters and SN model, the optimized control instructions are recalculated and transmitted to the motor controller to ensure the best match between the control instructions and the current system state;

[0186] After the control command is implemented, the control strategy is continuously adjusted according to the real-time information of the encoder and sensor network until the error is reduced, so that the system maintains a highly synchronized state;

[0187] The target state is the future state predicted by the nonlinear dynamics model;

[0188] e(t), including but not limited to speed error, position error, and tension error.

[0189] This embodiment also provides a computer device, which is suitable for the case of a multi-motor synchronous control method for a stranding machine, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the multi-motor synchronous control method for a stranding machine as proposed in the above embodiment.

[0190] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0191] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing synchronous control of multiple motors of a stranding machine as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0192] In summary, the present invention achieves precise real-time monitoring of the motor and wire status by integrating high-precision sensing and nonlinear dynamic modeling, and predicts the system status with the help of NMPC technology to generate the optimal control instruction sequence. At the same time, it innovatively integrates deep learning and machine learning algorithms, dynamically predicts control strategies based on historical data, effectively responds to sudden loads and nonlinear challenges, improves system adaptability and control accuracy, and can instantly adjust control weights according to changes in working conditions through a dynamic weight algorithm to ensure that the strategy is highly consistent with the actual situation. The closed-loop feedback system continuously optimizes the control strategy to ensure excellent synchronization of the stranding operation, overcoming the problems of inaccurate synchronization and poor adaptability of traditional methods in complex environments.

[0193] Example 2

[0194] Referring to Table 1, which is the second embodiment of the present invention, in order to further verify the technical solution of the present invention, experimental simulation data of a multi-motor synchronous control method for a stranding machine are provided.

[0195] This embodiment is directed to a multi-motor synchronous control method for a stranding machine, aiming to demonstrate its innovation and advantages. First, the stranding machine has undergone a comprehensive technical upgrade, and high-precision encoders and sensor networks have been installed at key locations. Specifically, each motor is equipped with an encoder with a resolution of up to 16 bits to track the rotation speed and position in real time. At the same time, piezoelectric force sensors are deployed to monitor wire tension, laser rangefinders track displacement, and temperature and humidity sensors are deployed to ensure accurate monitoring of environmental parameters.

[0196] In the embodiment, six motors were selected for synchronous control experiments, and different workloads and wire types were simulated to verify the universality and robustness of the method. The experimental environment parameters were set as follows: room temperature 25°C, relative humidity 50%RH to ensure the consistency of environmental factors.

[0197] The details are shown in the following table:

[0198] Table 1 Experimental record table

[0199]

[0200] By comparing the experimental data, it can be observed that although there are slight performance differences between different motors, they have achieved a high degree of synchronization as a whole. The comprehensive index function (I s ) are close, indicating that the control strategy effectively reduces the performance deviation between the motors. For example, the I sThe values ​​are distributed between 0.79 and 0.85, showing good consistency. In particular, when the external load suddenly increases or decreases (such as when Motor 2 encounters a temporary load mutation in a test), the system quickly adapts and optimizes the control instructions through real-time parameter adjustment of the nonlinear dynamic model and deep learning-assisted prediction, ensuring the tension (F t ) and displacement (D) fluctuations are kept within an acceptable range, and the error (e(t)) is reduced by about 30% compared with traditional control methods, significantly improving the accuracy and stability of synchronous operation.

[0201] The dynamic weight allocation algorithm (MLP model) of the present invention dynamically adjusts the weight parameters of the control strategy according to the real-time operating condition analysis results, further enhancing the adaptability to complex operating conditions. In the category of severe load mutation, by automatically optimizing the weight parameters, the matching degree between the control instructions and the actual needs is improved by at least 20%, effectively alleviating the control inaccuracy problem caused by nonlinear effects.

[0202] This embodiment fully demonstrates the innovation and superiority of the proposed multi-motor synchronous control method for stranding machines in practical applications. By integrating high-precision sensing, nonlinear dynamic model predictive control, deep learning and dynamic weight distribution strategy, highly synchronous control of stranding operations is achieved, which effectively solves the problems of low synchronization accuracy and poor adaptability of traditional methods under complex working conditions, and provides a solution for the field of stranding machine control.

[0203] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for synchronously controlling multiple motors of a stranding machine, characterized in that: include, Equip each motor with a high-precision encoder to monitor speed and position information in real time, and install a sensor network to monitor wire status and mechanical dynamic parameters; Construct a nonlinear dynamic model of the stranding machine's multiple motors, apply nonlinear model predictive control (NMPC) technology, predict the future system state based on the nonlinear dynamic model and calculate the control instruction sequence; Deep neural networks based on historical data combined with machine learning algorithms, predictive control strategies, and adaptive adjustments to load mutations and nonlinear effects; Design a dynamic weight allocation algorithm to dynamically adjust the weight parameters of each motor control strategy based on real-time operating condition analysis results; Implement the calculated control instructions to the motor controller, and use the feedback information from the encoder and sensor network to perform closed-loop adjustments, optimize the control strategy in real time based on the error, and maintain highly synchronized twisting operations; The deep neural network based on historical data combined with the machine learning algorithm predicts the control strategy, and the specific steps are as follows: Collect the operation data of the stranding machine under different working conditions, including motor speed, position, wire tension, displacement and environmental parameters; Sort the collected data by timestamp and store them as a time series dataset; Extract the motor speed change rate and tension fluctuation amplitude from the time series data set; The motor speed change rate is expressed as: ； Wherein, △v represents the rate of change of motor speed, v(t) represents the motor speed at the current time t, v(t-△t) represents the motor speed at the previous time t-△t, and △t represents the time interval between two sampling time points; The tension fluctuation amplitude is expressed as: ； Among them, δF represents the tension fluctuation amplitude, F(t) represents the tension value at time t, △t represents the time interval, and F(t-△t) represents the tension value in a small time period △t before time t; The SN model is constructed by combining the long short-term memory network LSTM and the convolutional neural network CNN; The extracted motor speed change rate, tension fluctuation amplitude, historical control instructions and external disturbances are input into the SN model, and the desired speed and target tension of the motor are output; Set the previous moment's historical control command to C p , the external disturbance at the current moment is D; After calculating the processed input vector, the SN model expression is: ； Where IP represents the preprocessed integrated input vector, w △v ,w δF ,w C ,w D They correspond to the motor speed change rate, tension fluctuation amplitude, historical control instructions, and weight coefficients of external disturbances, respectively, and sgn is a sign function; The output layer expression of the SN model is: ； Among them, v d represents the desired motor speed output, T d Indicates the target tension output; The specific steps of the adaptive adjustment of load mutation and nonlinear effect are as follows: Collect real-time data to accurately monitor motor speed, position, wire tension, displacement and environmental parameters; By monitoring the rate of change of the motor current, it is possible to determine whether a load has occurred. The expression for the rate of change is: ； in, represents the instantaneous rate of change, △I represents the rate of change of current, △t represents the time interval, I(t) represents the current value at time t, and I(t-△t) represents the current value at time t-△t; Calculate the average value of all instantaneous change rates △I / △t and standard deviation σ △I / △t ; Once a sudden load change is detected, the parameter re-evaluation of the nonlinear dynamic model is triggered immediately, and the friction coefficient and material hardness coefficient in the model are adjusted according to the current working conditions; Iteratively optimize the parameters f(x(t),u(t),d(t)) in the model based on the latest data, where f represents a nonlinear function, x(t) represents all necessary variables of the current state, u(t) represents the control vector, and d(t) is the external disturbance vector.

2. The method for synchronously controlling multiple motors of a stranding machine according to claim 1, characterized in that: The high-precision encoder is equipped on each motor to monitor the speed and position information in real time, and a sensor network is installed to monitor the wire state and mechanical dynamic parameters. The specific steps are as follows: Install piezoelectric force sensors and laser rangefinders at key locations on the stranding machine to monitor wire tension and displacement; Monitor the working environment through temperature and humidity sensors; The comprehensive index function is constructed by using the information of motor speed, position, wire tension and displacement, and the expression is: ； Among them, I s represents the comprehensive index function, n represents the number of motors, represents the real-time speed of the i-th motor, Represents the average speed of all motors, represents the position information of the i-th motor, represents the average value of the motor position, σ p Indicates the standard deviation of the motor position, F t,i represents the real-time tension of the i-th wire, represents the average value of wire tension, σ F Indicates the standard deviation of tension, D i represents the real-time displacement of the i-th wire, represents the average value of wire displacement, σ D represents the standard deviation of displacement, λ represents the tension influence factor, and η represents the weight parameter of displacement deviation.

3. The method for synchronously controlling multiple motors of a stranding machine according to claim 2, characterized in that: The nonlinear dynamic model of the stranding machine multi-motor is constructed, and the nonlinear model predictive control (NMPC) technology is used to predict the future system state and calculate the control instruction sequence according to the nonlinear dynamic model. The specific steps are as follows: A nonlinear dynamic model is constructed based on motor torque, wire tension, friction, inertia and nonlinear dynamic functions. The expression is: ； in, represents the rate of change of the system state over time, f represents a nonlinear function, x(t) represents all necessary variables of the current state, u(t) represents the control vector, and d(t) is the external disturbance vector; Based on the nonlinear dynamics model, a predictive control model is constructed to predict the state of the future time step Np. The expression is: ； Among them, k represents the current time step, represents the time point of the current time step, x k+e|k Indicates that at time step k, the state of the future e-th time step is predicted based on the current state, x k represents the actual state of the system at time step k, △t represents the sampling time interval; The objective function is constructed based on the predictive control model, and the expression is: ； Among them, J represents the value of the objective function, r k+i represents the target state of the k+ith time step in the future, u represents the control vector, and u k+i-1|k represents the control input predicted at time step k when executing the i-1th control action, w1 and w2 are the weight coefficients of the state error and the control input respectively; Using sequential quadratic programming SQP, under given constraints, the control sequence that minimizes the objective function J is solved. The control instruction expression is: ； Among them, N c To control the horizon, it indicates the number of future time steps considered in the optimization process, u k|k It means the control instruction formulated at time k, Indicates that at time k, the future Step pre-calculated control instructions; According to the first step control instruction u obtained by optimization k|k , and send it to the motor controller for control.

4. The method for synchronously controlling multiple motors of a stranding machine according to claim 3, characterized in that: The dynamic weight allocation algorithm is designed to dynamically adjust the weight parameters of each motor control strategy according to the real-time working condition analysis results. The specific steps are as follows: Extract key parameters from sensor networks in real time; According to the extracted parameters, the real-time working conditions are divided into three categories: normal operation, mild load mutation, and severe load mutation; Construct a feature vector for each category, the expression is: ； Among them, X represents the characteristic vector of the current working condition of the stranding machine, H represents the hardness of the wire, and v m Indicates the running speed of the motor, T indicates the ambient temperature; Based on X and combined with the dynamic weight allocation function, the MLP model is constructed to automatically adjust the weight parameters w of each control strategy △v ,w δF ,w C ,w D , the MLP model expression is: ； Where W(X) represents the dynamic weight assignment function, MLP represents the multi-layer perceptron, [w △v ,w δf ,w C ,w D ]0 represents the initial weight vector; Use historical operating data and known optimal control strategies to input the MLP model to complete training; Input the dynamically adjusted weights into the SN model to obtain a new input vector IP a , the adjusted SN model expression is: 。 5. The method for synchronously controlling multiple motors of a stranding machine according to claim 4, characterized in that: The step of implementing the calculated control command to the motor controller and performing closed-loop adjustment using feedback information from the encoder and the sensor network is as follows: The control instructions obtained through nonlinear model predictive control (NMPC) technology and deep learning assisted optimization Sent to the controller of each motor; Through high-precision encoders, force sensors and displacement sensors, the motor speed, position, wire tension, displacement and environmental parameters are collected in real time.

6. The method for synchronously controlling multiple motors of a stranding machine according to claim 5, characterized in that: The specific steps of optimizing the control strategy in real time according to the error to maintain highly synchronized stranding operation are as follows: Input the real-time data into the nonlinear dynamics model to obtain the current state of the system, compare the current state of the system with the target state to obtain the actual error e(t), and update the comprehensive index function I S The value of Using the dynamic weight allocation algorithm, the weight parameters of the control strategy are automatically optimized through the MLP model according to the current operating condition feature vector X; Combined with the updated weight parameters and SN model, the optimized control instructions are recalculated and transmitted to the motor controller to ensure the best match between the control instructions and the current system state; After the control command is implemented, the control strategy is continuously adjusted according to the real-time information of the encoder and sensor network until the error is reduced, so that the system maintains a highly synchronized state.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the stranding machine multi-motor synchronous control method described in any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the stranding machine multi-motor synchronous control method described in any one of claims 1 to 6 are implemented.

Images