Coiling roller safety state prediction control method based on multi-sensor cooperation

Abstract

The invention relates to the technical field of industrial control, in particular to a coiling roller safety state prediction control method based on multi-sensor cooperation. The method comprises the following steps: acquiring multi-modal data in real time through multiple sensors deployed on a coiling roller, and fusing the multi-modal data with digital twin model data to generate a state feature vector; solving a future prediction state value and uncertainty by using the state prediction model; and further, the model prediction controller synthesizes the actual state, the prediction state and the uncertainty, online optimization is performed to generate an optimal control sequence, the sequence is received and executed by the programmable logic controller, and closed-loop adjustment is performed on the control parameters of the coiling roller. Meanwhile, the system calculates the abnormal operation probability in real time and gives an over-limit alarm. Prediction, control and monitoring are deeply fused, active prevention of equipment faults is achieved, and the intelligent level, operation safety and production efficiency of a program control system are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial control technology, in particular to a coiling roll safety state prediction control method based on multi-sensor cooperation. BACKGROUND

[0002] In the field of industrial automation, the operation control of coiling rolls generally adopts program control systems. Such systems usually take programmable logic controllers (PLCs) as the hardware core, with a set of pre-set, static logic programs running inside; the programs strictly execute a series of determined instruction sequences according to process requirements, such as controlling coiling tension through PID loops, setting line speed synchronization of multi-stage drive units, and managing roll diameter calculation, to achieve precise control over the production process.

[0003] However, the existing program control technology has the following limitations in the application scenario of coiling rolls:

[0004] (1) In the traditional control architecture, the process control program executed by the PLC and the device state monitoring system running in parallel with it are independent in terms of data and logic, so the traditional program control system cannot perceive the dynamic changes in the state characteristics of the coiling roll, causing the control program to continue executing the original instruction sequence when the device health state deteriorates;

[0005] (2) The state monitoring results cannot directly and dynamically guide the program control system to adjust operating parameters to avoid risks, lacking closed-loop active control capabilities.

[0006] Therefore, a coiling roll safety state prediction control method based on multi-sensor cooperation is proposed. SUMMARY

[0007] The present application aims to provide a coiling roll safety state prediction control method based on multi-sensor cooperation.

[0008] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0009] A coiling roll safety state prediction control method based on multi-sensor cooperation, comprising:

[0010] Real-time acquisition of multi-modal data through multi-sensors deployed on the coiling roll and the conveying belt;

[0011] Preprocessing of the multi-modal data, and fusion of the preprocessed data with the pre-set coiling roll digital twin model data to generate a feature vector representing the current state of the coiling roll;

[0012] Based on the feature vector, in combination with a pre-established state prediction model and physical constraint conditions, analysis of the current state of the coiling roll to obtain its predicted state value and state uncertainty distribution;

[0013] According to the current actual state value, the predicted state value and the state uncertainty distribution, applying a model predictive controller to online optimize the control sequence of the coiler roll, and generating an optimal control sequence; receiving and executing the optimal control sequence by a programmable logic controller, and performing closed-loop adjustment on the control parameters;

[0014] The running abnormal probability of the coiler roll is calculated in real time by the monitoring and alarm module, and when the probability exceeds a preset safety threshold, an alarm is triggered to prompt the operator to intervene.

[0015] Further, the multi-modal data includes vibration data, temperature data, visual data and operation parameter data; wherein the operation parameter data includes motor torque, coiling tension and conveyor belt speed.

[0016] Further, the process of pre-processing the multi-modal data to generate pre-processed data includes:

[0017] The adaptive filtering technique is applied to the multi-modal data for denoising to obtain first multi-modal data;

[0018] The timestamp synchronization and interpolation method is used for data alignment on the first multi-modal data to obtain second multi-modal data;

[0019] The second multi-modal data is normalized to obtain pre-processed data.

[0020] Further, the process of fusing the pre-processed data and the preset coiler roll digital twin model data to generate a feature vector representing the current state of the coiler roll includes:

[0021] Based on the geometric parameters, material properties and operating conditions of the coiler roll, a digital twin model is constructed;

[0022] The time domain and frequency domain analysis method is used for feature extraction on the pre-processed data to obtain first features;

[0023] The real-time operation parameter data is input into the digital twin model, and the virtual sensor data under the same working condition is output, and the feature extraction is performed to obtain second features;

[0024] The first features and the second features are compared to obtain residual features; the first features, the second features and the residual features are spliced to obtain a feature vector representing the current state of the coiler roll.

[0025] Further, based on the feature vector, combined with the pre-established state prediction model and physical constraint conditions, the current state of the coiler roll is analyzed, and the predicted state value and the state uncertainty distribution are output.

[0026] inputting the feature vector into a pre-trained long short-term memory network model to obtain a predicted state value;

[0027] inputting the feature vector into a pre-trained physical information neural network model to output a probability vector of the coiler roll under different state categories;

[0028] calculating the highest state probability by using a Shannon entropy calculation formula to obtain a state entropy for representing state uncertainty distribution.

[0029] Further, the process of using a model predictive controller to optimize the control sequence of the coiler roll includes:

[0030] constructing a cost function of the model predictive controller, and the optimization objective is to minimize the deviation between the predicted state value and the current actual state value, reduce state uncertainty, and suppress the sharp change of the control amount;

[0031] solving the optimization problem by using a nonlinear programming algorithm to obtain the control sequence in the future period of time;

[0032] in each control period, the first instruction of the control sequence is issued to the programmable logic controller, and the rolling optimization is performed by using new measurement data to generate an optimal control sequence.

[0033] Further, the process of calculating the running abnormal probability of the coiler roll in real time by the monitoring and alarm module includes:

[0034] the edge controller continuously acquires and compares the predicted state value and the actual measurement value of the sensor;

[0035] combining the difference between the predicted state value and the actual measurement value with the state uncertainty distribution to form an abnormal score;

[0036] mapping the abnormal score to a running abnormal probability value between 0 and 1 by using a normalization function.

[0037] Compared with the prior art, the beneficial effects of the present application are:

[0038] 1. By combining the geometric parameters, material properties and operating conditions of the coiler roll to construct a digital twin model, and by extracting features through time-frequency domain analysis, combining the residual features of real-time and virtual data, and generating a comprehensive feature vector representing the current state of the coiler roll, the accuracy of state representation is improved; by simulating the actual working condition through the digital twin model, the state change of the coiler roll is dynamically captured, and the limitation that the traditional control system cannot perceive the decline of the equipment health state is overcome; the introduction of residual features can highlight the deviation between the actual and expected states, effectively improve the sensitivity of the feature vector to abnormal states, and provide a more reliable basis for subsequent prediction and control.

[0039] 2. By using a pre-trained long short-term memory network model to capture temporal features, high-precision predicted state values ​​are generated. Combined with the output state probability vector of the physical information neural network, the prediction results are ensured to conform to physical laws. The state uncertainty distribution is calculated by Shannon entropy, providing a quantitative confidence index for state prediction and improving the reliability of the prediction results. The state probability vector and uncertainty distribution provide a basis for anomaly detection, which helps to identify potential risks in a timely manner and make up for the shortcomings of the disconnect between traditional system state monitoring and control.

[0040] 3. By constructing a cost function that includes state deviation, uncertainty, and control smoothness, the system can comprehensively consider prediction accuracy and control stability to generate a smooth and efficient optimal control sequence, reducing drastic fluctuations in the control process. The system uses a nonlinear programming algorithm to efficiently solve the optimization problem and combines a rolling optimization strategy to continuously update the control sequence, ensuring that the control command can dynamically adapt to the state changes of the winding roll, thereby improving the adaptability and reliability of the winding roll safety state predictive control. Attached Figure Description

[0041] Fig. 1 This is a flowchart illustrating a winding roll safety state prediction and control method based on multi-sensor collaboration according to the present invention.

[0042] Fig. 2 This is a schematic diagram illustrating the process of generating feature vectors in this invention.

[0043] Fig. 3 This is a schematic diagram of the process for generating the optimal control sequence in this invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Please see Figs. 1 to 3 This invention provides a method for predictive control of the safety status of a winding roll based on multi-sensor collaboration, the technical solution of which is as follows:

[0046] Example 1

[0047] To improve the predictive control effect of the safety status of the take-up roll, a company used a multi-sensor collaborative predictive control method for the safety status of the take-up roll proposed in this invention. The flowchart of the method is as follows: Fig. 1 As shown, it specifically includes:

[0048] Real-time collection of multi-modal data through multiple sensors deployed on the coiling roller and the conveying belt;

[0049] Further, the multi-modal data includes vibration data, temperature data, visual data, and job parameter data; wherein the job parameter data includes motor torque, coiling tension, and conveying belt speed;

[0050] Further, the vibration data and the temperature data are collected by an acceleration sensor and a temperature sensor respectively installed on the bearing seat of the coiling roller; the visual data is obtained by shooting with an industrial camera erected directly above the coiling roller, while a coaxial ring LED light source is provided to eliminate the visual impact of surface reflection; the job parameter data is collected by a torque sensor installed on the motor shaft, a tension sensor on the production line, and a speed encoder of the conveying belt.

[0051] By collecting multi-modal data such as vibration, temperature, vision, and job parameters, the running state of the coiling roller is comprehensively captured, making up for the deficiency of traditional control systems that rely only on single or few data sources and cannot perceive dynamic state changes; multi-dimensional sensing data provides more abundant information sources for coiling roller safety state prediction control.

[0052] The multi-modal data is preprocessed, and the preprocessed data and the preset coiling roller digital twin model data are fused to generate a feature vector representing the current state of the coiling roller;

[0053] Further, the process of preprocessing the multi-modal data to generate preprocessed data includes:

[0054] The adaptive filtering technique is applied to the multi-modal data for denoising to obtain first multi-modal data;

[0055] The timestamp synchronization and interpolation method is used to align the first multi-modal data, to obtain second multi-modal data;

[0056] The second multi-modal data is normalized to obtain preprocessed data;

[0057] Further, the visual data in the multi-modal data is denoised using the Wiener filtering technique, and the adaptive filtering algorithm used for the remaining data is the least mean square algorithm;

[0058] Further, in the timestamp synchronization process, for sensors with different sampling rates, linear interpolation is used to align the low sampling rate data to the highest sampling rate reference, forming second multi-modal data that is completely synchronized in time;

[0059] Further, the aligned multi-modal data is scaled to the [0, 1] interval by the min-max normalization method.

[0060] The noise is removed by adaptive filtering technology, the multi-source data is aligned by timestamp synchronization and interpolation method, and the data scale is unified by normalization processing, so as to improve the quality and consistency of multi-modal data, and lay a foundation for subsequent feature extraction and state prediction.

[0061] Further, the generation process of the feature vector is as shown in Fig. 2 , which comprises:

[0062] Based on the geometric parameters, material properties and operating conditions of the coiler roll, a digital twin model is constructed;

[0063] The preprocessed data is subjected to feature extraction using time domain and frequency domain analysis methods to obtain first features;

[0064] The real-time job parameter data is input into the digital twin model, and the virtual sensor data under the same working condition is output, and the first features and the second features are compared to obtain residual features; the first features, the second features and the residual features are spliced to obtain a feature vector representing the current state of the coiler roll;

[0065] The first features and the second features are compared to obtain residual features; the first features, the second features and the residual features are spliced to obtain a feature vector representing the current state of the coiler roll;

[0066] Further, a three-dimensional model of the coiler roll is drawn using SolidWorks software, and the model is imported into a multi-physics simulation software, and material properties such as Young's modulus, Poisson's ratio, density, etc. are defined; then, parameterized scanning simulation is performed in the simulation software, and steady-state dynamics analysis and thermal analysis are performed for each motor speed and coiling tension parameter combination to simulate virtual sensor data under ideal conditions;

[0067] Further, the time domain and frequency domain analysis includes calculation of statistical quantities such as mean, variance, peak value, and frequency spectrum information obtained by fast Fourier transform;

[0068] Further, before the time domain and frequency domain analysis of the preprocessed data and the virtual sensor data, the preprocessed data is decomposed using Daubechies wavelet to extract low-frequency components and high-frequency components; then, the time domain and frequency domain analysis is performed on each layer of decomposition results to generate multi-scale features; by introducing multi-scale decomposition technology, the fine granularity of capturing dynamic changes of the coiler roll can be improved, and the representation ability of the feature vector can be improved.

[0069] The digital twin model is constructed by combining the geometric parameters of the coiling roller, material properties and operating conditions, and features are extracted through time-frequency domain analysis, combined with the residual features of real-time and virtual data to generate a comprehensive feature vector representing the current state of the coiling roller to improve the accuracy of state representation; the digital twin model can simulate actual working conditions, dynamically capture the state changes of the coiling roller, and overcome the limitations of traditional control systems that cannot perceive the decline of the health state of the equipment; at the same time, the introduction of residual features can highlight the deviation between the actual and expected states, effectively improve the sensitivity of the feature vector to abnormal states, and provide a reliable basis for state prediction and control.

[0070] Based on the feature vector, the current state of the coiling roller is analyzed in combination with a pre-established state prediction model and physical constraint conditions to obtain the predicted state value and state uncertainty distribution of the coiling roller;

[0071] Further, based on the feature vector, the current state of the coiling roller is analyzed in combination with a pre-established state prediction model and physical constraint conditions to output the predicted state value and state uncertainty distribution of the coiling roller, the process comprising:

[0072] The feature vector is input into a pre-trained long short-term memory network model to obtain the predicted state value;

[0073] The feature vector is input into a pre-trained physical information neural network model to output a probability vector of the coiling roller under different state categories;

[0074] The highest state probability is calculated using the Shannon entropy calculation formula to obtain the state entropy, which is used to represent the state uncertainty distribution;

[0075] Further, the long short-term memory network model comprises: 2 LSTM layers, 1 Dropout layer and 1 fully connected output layer; each LSTM layer has 128 neurons; the model takes the feature vector as input and outputs the future sensor prediction value; the training process uses the Adam optimizer with a learning rate of 0.001, the loss function is mean square error, and the training is performed for 100 Epochs with a batch size of 64;

[0076] Further, the physical information neural network model (PINN) is a fully connected neural network comprising 4 hidden layers, each layer having 256 neurons, the activation function using ReLU, and the output layer using the Softmax function for probability generation; the model takes the feature vector as input and outputs the probability of each type of coiling roller state; the loss function uses the weighted sum of the classification cross-entropy loss and the physical constraint loss;

[0077] Further, in actual production lines, the probability of occurrence of abnormal samples is very low, resulting in a small amount of available abnormal samples, thereby reducing the reliability of the physical information neural network model outputting the abnormal state probability; a generative adversarial network is introduced to generate a latent state distribution to assist the physical information neural network model in generating a more robust state probability vector; the feature vector is input into the generator to generate a simulated state probability vector, and the discriminator is used to distinguish the probability vector from the PINN and the generated probability vector; the probability vector generated by the GAN is weighted and averaged with the PINN output to generate a more robust state probability vector, improving the accuracy of the PINN in predicting abnormal probability.

[0078] By using a pre-trained long short-term memory network model to capture time sequence features, a high-precision predicted state value is generated, combined with a physical information neural network output state probability vector, and a state uncertainty distribution is calculated using Shannon entropy to provide a quantitative confidence index for state prediction; using the state probability vector and the uncertainty distribution can identify potential risks in a timely manner, making up for the shortcomings of traditional system state monitoring and control, thereby improving the reliability of the coiling roll safety state prediction control.

[0079] According to the current actual state value, the predicted state value and the state uncertainty distribution, a model predictive controller is applied to optimize the control sequence of the coiling roll online to generate an optimal control sequence; the optimal control sequence is received and executed by a programmable logic controller to perform closed-loop adjustment on the control parameters;

[0080] Further, the process of using a model predictive controller to optimize the control sequence of the coiling roll to generate an optimal control sequence is as shown in Fig. 3 , which includes:

[0081] A cost function of the model predictive controller is constructed, and the optimization goal is to minimize the deviation between the predicted state value and the current actual state value, reduce the state uncertainty, and suppress the drastic change of the control amount;

[0082] A nonlinear programming algorithm is used to solve the optimization problem to obtain the control sequence for a future period of time;

[0083] In each control period, the first instruction of the control sequence is issued to the programmable logic controller, and the new measurement data is used for rolling optimization to generate an optimal control sequence;

[0084] Further, the control period, the prediction horizon and the control horizon of the model predictive controller are set to 100 ms, 20 steps and 10 steps respectively; the state variable is defined as the actual state value, the predicted state value and the state entropy, and the control variable is set as a three-dimensional vector, including: motor output torque, coiling tension set value and conveyor belt speed; the constraint conditions include control amount constraint and control amount change rate constraint;

[0085] Further, the cost function comprises a state tracking term, an uncertainty suppression term and a control suppression term; wherein the state tracking term is used to penalize the gap between the predicted state value and the actual state value, the uncertainty suppression term directly penalizes the size of the predicted state uncertainty, and the control suppression term is used to penalize the magnitude of the control action, i.e. the size of the increment of the control variable;

[0086] Further, a nonlinear programming solver such as IPOPT is used to quickly solve the optimization problem on the edge controller to obtain a control sequence composed of control variables that can minimize the cost function;

[0087] Further, a DRL model is constructed, with state variables and control variables as the state space, and the cost function value as the reward function, and the action space of DRL is represented as the adjustment of the weights in the cost function. The DRL model is pre-trained offline using historical operation data, and at each control period, DRL adjusts the weights according to the real-time working conditions to optimize the cost function, so that the model predictive controller can adapt to the optimization requirements under different working conditions.

[0088] By constructing a cost function that includes state deviation, uncertainty and control quantity smoothness, the system can consider both prediction accuracy and control stability to generate a smooth and efficient optimal control sequence, reducing sharp fluctuations in the control process; using nonlinear programming algorithm to efficiently solve the optimization problem, combined with the rolling optimization strategy to continuously update the control sequence, to ensure that the control instruction can dynamically adapt to the state changes of the coiling roll, thereby improving the adaptability and reliability of the coiling roll safety state prediction control.

[0089] Through the monitoring and alarm module, the running abnormal probability of the coiling roll is calculated in real time, and when the probability exceeds the preset safety threshold, an alarm is triggered to prompt the operator to intervene.

[0090] Further, the process of calculating the running abnormal probability of the coiling roll in real time by the monitoring and alarm module comprises:

[0091] The predicted state value is continuously obtained and compared with the actual measurement value of the sensor by the edge controller;

[0092] The difference between the predicted state value and the actual measurement value is combined with the state uncertainty distribution to form an abnormal score;

[0093] The abnormal score is mapped to a running abnormal probability value between 0 and 1 by a normalization function;

[0094] Further, the deviation value between the predicted state value and the actual measurement value of the sensor is standardized, and the standardized result is weighted and summed with the state entropy to obtain the abnormal score;

[0095] Further, the Sigmoid function is used to map the anomaly score to a running anomaly probability value between 0 and 1, and when the value is greater than 0.85, an alarm is triggered to prompt the operator to intervene.

[0096] The edge controller continuously compares the predicted state value with the actual measured value, calculates the anomaly score in combination with the state uncertainty distribution, and generates a running anomaly probability in real time, thereby improving the timeliness and accuracy of anomaly detection and making up for the deficiency that the state monitoring result of the traditional system cannot directly guide the control.

[0097] The embodiment proposes a coiling roll safety state prediction control method based on multi-sensor cooperation. The method collects multi-modal data in real time through the multi-sensor deployed on the coiling roll, and fuses the data with the digital twin model data to generate a state feature vector. Then, the state prediction model is used to calculate the future predicted state value and uncertainty. Further, the model predictive controller integrates the actual state, the predicted state and the uncertainty to generate an optimal control sequence online, which is received and executed by the programmable logic controller to make closed-loop adjustment on the control parameters of the coiling roll. At the same time, the system calculates the running anomaly probability in real time and performs an overrun alarm. The present application deeply integrates prediction, control and monitoring, actively prevents equipment failure, and significantly improves the intelligent level, running safety and production efficiency of the program control system.

[0098] Embodiment two

[0099] The embodiment takes the optimization of the safety state prediction control of the coiling roll of a certain steel strip production enterprise as an example, and the specific process is as follows:

[0100] Multi-modal data is collected in real time through the multi-sensor deployed on the coiling roll and the conveying belt;

[0101] The enterprise uses the acceleration sensor installed on the bearing seat of the coiling roll, the attached thermocouple sensor and the industrial camera to collect vibration data, temperature data and visual data respectively; then, the torque sensor, the tension sensor and the speed encoder are used to collect the operating parameter data; the reference values are motor torque 2000 N·m, coiling tension 50 kN and conveying belt speed 1.5 m / s.

[0102] The multi-modal data is preprocessed, and the preprocessed data and the preset coiling roll digital twin model data are fused to generate a feature vector representing the current state of the coiling roll;

[0103] The temperature data and the visual data with lower sampling rates are aligned to the 1000 Hz sampling rate reference of the vibration data by the linear interpolation method to generate time-synchronized preprocessed data;

[0104] The enterprise uses SolidWorks to draw a three-dimensional model of the coiler roll, defines material properties such as Young's modulus of 200 GPa, Poisson's ratio of 0.3, and density of 7850 kg / m³, and imports COMSOL Multiphysics software for multi-physics field simulation to simulate virtual sensor data under the combination of motor speed and coiling tension;

[0105] According to the pre-processing data characteristics and simulation data characteristics under time-frequency domain analysis, a feature vector is generated;

[0106] Based on the feature vector, combined with the pre-established state prediction model and physical constraint conditions, the current state of the coiler roll is analyzed to obtain its predicted state value and state uncertainty distribution;

[0107] The enterprise inputs the feature vector into the pre-trained long short-term memory network model and physical information neural network model to obtain the sensor prediction value for the next 100 ms and the probability vector of the coiler roll under normal, slight abnormal and severe abnormal states, and uses the Shannon entropy formula to calculate the state entropy for representing the state uncertainty distribution;

[0108] The sensor prediction value, i.e., the predicted state value, is as follows: vibration amplitude of 0.02 m / s², temperature of 85°C, and tension deviation of 0.5 kN; the probability vector is [0.85, 0.12, 0.03], so the coiler roll is biased towards the normal state.

[0109] According to the current actual state value, predicted state value and state uncertainty distribution, a model predictive controller is applied to optimize the control sequence of the coiler roll online to generate the optimal control sequence; the programmable logic controller receives and executes the optimal control sequence to close-loop adjust the control parameters;

[0110] The settings of the model predictive controller are as follows: control period of 100 ms, prediction time domain of 20 steps, and control time domain of 10 steps; the state variable is defined as the actual state value, predicted state value and state entropy; the constraint ranges of motor torque, coiling tension and conveyor belt speed in the control variable are 2000-2500 N·m, 45-55 kN and 1.4-2.0 m / s, respectively;

[0111] The model predictive controller solves the cost function minimization problem by using the IPOPT nonlinear programming solver to optimize the coiler roll control sequence, and the generated optimal control sequence is as follows: the motor torque is adjusted to 2100 N·m, the coiling tension is adjusted to 48 kN, and the conveyor belt speed is adjusted to 1.6 m / s.

[0112] The monitoring and alarm module calculates the running abnormal probability of the coiler roll in real time, and triggers an alarm to prompt the operator to intervene when the probability exceeds the preset safety threshold.

[0113] The deviation values of vibration, temperature and tension data are 0.015 m / s², 0.4 ℃ and 0.3 kN respectively, and the calculated abnormal scores and abnormal probabilities are 0.62 and 0.65 respectively, which do not exceed the alarm threshold.

[0114] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for predictive control of the safety state of a take-up roll based on multi-sensor collaboration, characterized in that, include: Multimodal data is collected in real time by deploying multiple sensors on the take-up roller and conveyor belt; The multimodal data is preprocessed and fused with the preprocessed data and the preset digital twin model data of the roll to generate a feature vector representing the current state of the roll. Based on the feature vector, combined with the pre-established state prediction model and physical constraints, the current state of the winding roll is analyzed to obtain its predicted state value and state uncertainty distribution. Based on the current actual state value, predicted state value, and state uncertainty distribution, a model predictive controller is applied to optimize the control sequence of the winding roll online and generate the optimal control sequence; the optimal control sequence is received and executed by a programmable logic controller to perform closed-loop adjustment of the control parameters; The monitoring and alarm module calculates the probability of abnormal operation of the winding roller in real time, and triggers an alarm when the preset safety threshold is exceeded to prompt the operator to intervene.

2. The method for predictive control of the safety status of a winding roll based on multi-sensor collaboration according to claim 1, characterized in that, The multimodal data includes vibration data, temperature data, visual data, and operational parameter data; among which, the operational parameter data includes motor torque, winding tension, and conveyor belt speed.

3. The method for predictive control of the safety state of a winding roll based on multi-sensor collaboration according to claim 1, characterized in that, The process of preprocessing multimodal data to generate preprocessed data includes: Adaptive filtering techniques are applied to the multimodal data to remove noise, resulting in the first multimodal data. The first multimodal data is aligned using timestamp synchronization and interpolation methods to obtain the second multimodal data; The second multimodal data is normalized to obtain preprocessed data.

4. The method for predictive control of the safety state of a take-up roll based on multi-sensor collaboration according to claim 1, characterized in that, The process of fusing preprocessed data and preset digital twin model data of the roll to generate a feature vector characterizing the current state of the roll includes: A digital twin model is constructed based on the geometric parameters, material properties, and operating conditions of the winding roller; The first feature is obtained by using time-domain and frequency-domain analysis methods to extract features from the preprocessed data. Real-time operational parameter data is input into the digital twin model, and virtual sensor data under the same working conditions is output. Features are then extracted from the data to obtain the second feature. The first feature is compared with the second feature to obtain the residual feature; the first feature, the second feature and the residual feature are concatenated to obtain the feature vector representing the current state of the roll.

5. The method for predictive control of the safety state of a take-up roll based on multi-sensor collaboration according to claim 1, characterized in that, Based on feature vectors, combined with a pre-established state prediction model and physical constraints, the process of analyzing the current state of the winding roll and outputting its predicted state value and state uncertainty distribution includes: The feature vector is input into a pre-trained long short-term memory network model to obtain the predicted state value; The feature vector is input into a pre-trained physical information neural network model, which outputs a probability vector of the roll under different state categories. The highest state probability is calculated using the Shannon entropy formula to obtain the state entropy, which is used to characterize the distribution of state uncertainty.

6. The method for predictive control of the safety state of a winding roll based on multi-sensor collaboration according to claim 1, characterized in that, The process of using a model predictive controller to optimize the control sequence of the winding roll and generate the optimal control sequence includes: The cost function of the model predictive controller is constructed with the optimization objective of minimizing the deviation between the predicted state value and the current actual state value, reducing state uncertainty, and suppressing drastic changes in the control quantity. The optimization problem is solved using a nonlinear programming algorithm to obtain the control sequence for a future period of time. In each control cycle, the first instruction of the control sequence is sent to the programmable logic controller, and rolling optimization is performed using new measurement data to generate the optimal control sequence.

7. The method for predictive control of the safety state of a winding roll based on multi-sensor collaboration according to claim 1, characterized in that, The process of calculating the probability of abnormal operation of the take-up roll in real time through the monitoring and alarm module includes: The edge controller continuously acquires and compares the predicted state values ​​with the actual measurement values ​​from the sensors; The difference between the predicted state value and the actual measured value is combined with the state uncertainty distribution to form an anomaly score; The anomaly score is mapped to a runtime anomaly probability value between 0 and 1 using a normalization function.

Images