Electrical steel strip rolling process intelligent scheduling and energy consumption prediction system

Abstract

The invention relates to the technical field of steel production, and discloses an electrical steel strip rolling process intelligent scheduling and energy consumption prediction system, which comprises a data acquisition module, a dynamic scheduling module, an energy consumption prediction module, an anomaly detection module and an optimization feedback module. The data acquisition module acquires data by using a distributed sensor network; the dynamic scheduling module optimizes scheduling based on an improved genetic algorithm; the energy consumption prediction module adopts a deep space-time convolutional neural network to predict energy consumption; the anomaly detection module detects anomaly based on an isolated forest algorithm; and the optimization feedback module adjusts a scheduling strategy through a reinforcement learning framework. The system realizes intelligent scheduling of the rolling process and accurate prediction of energy consumption, can effectively improve production efficiency, reduce energy consumption, timely discover abnormity and optimize a scheduling scheme, and has remarkable economic benefits and production management advantages.

Description

Technical Field

[0001] The present invention relates to the technical field of steel production, and specifically to an intelligent scheduling and energy consumption prediction system for the rolling process of electrical steel strips. Background Art

[0002] In the field of modern steel production, the rolling process of electrical steel strips is a key link determining product quality and enterprise benefits. With the advancement of industrial intelligence, many problems have emerged in the traditional rolling process in terms of scheduling and energy consumption management, making it difficult to meet the growing demands of the industry.

[0003] The scheduling of the traditional rolling process mainly relies on manual experience or simple rule algorithms, lacking the effective ability to cope with complex production environments and multi-objective optimization. In the face of sudden situations such as equipment failures and order changes, it is unable to quickly adjust the scheduling plan, resulting in low production efficiency and low equipment utilization rate. For example, when arranging rolling tasks, the load capacity of equipment and the sequence between processes are often not fully considered, causing excessive wear of equipment or too long waiting time between processes, wasting a large amount of production resources.

[0004] In terms of energy consumption management, due to the lack of accurate prediction means, enterprises are difficult to plan energy supply in advance and optimize the production process. Traditional energy consumption prediction methods often rely on simple statistical models or empirical formulas, unable to accurately capture the influence of various complex factors on energy consumption during the rolling process. This makes enterprises have great blindness in energy procurement and use. Either the production progress is affected due to insufficient energy reserves, or waste is caused due to excessive energy, increasing production costs.

[0005] In data collection and processing, there are also obvious defects in previous technical means. The layout of sensors is not reasonable enough to comprehensively and accurately obtain key data during the rolling process. Moreover, the collected data lacks effective integration and analysis, and a large amount of valuable information is left unused, unable to provide strong support for production decisions.

[0006] In addition, with the intensification of market competition, customers have higher and higher requirements for the quality and delivery period of electrical steel strips. Enterprises need to improve production efficiency and reduce energy consumption on the premise of ensuring product quality to enhance their market competitiveness. However, the existing rolling process scheduling and energy consumption management technologies cannot meet these strict requirements, and there is an urgent need for an innovative intelligent system to solve these problems and achieve efficient, energy-saving and intelligent production during the rolling process. Summary of the Invention

[0007] The purpose of the present invention is to provide an intelligent scheduling and energy consumption prediction system for the rolling process of electrical steel strips to solve the problems raised in the above background art.

[0008] To achieve the above object, the present invention provides the following technical solutions: an intelligent scheduling and energy consumption prediction system for the rolling process of electrical steel strips, the system comprising:

[0009] A data acquisition module, configured to collect temperature, pressure, speed and energy consumption data in the rolling process in real time through a distributed sensor network. The sensor network includes a temperature sensor array, a pressure sensor array and a speed encoder, and each sensor is embedded in key nodes of the rolling mill equipment in a spatial grid distribution manner;

[0010] A dynamic scheduling module, configured to perform multi-objective optimization scheduling on the rolling process based on an improved genetic algorithm. The improved genetic algorithm includes process priority coding, dynamic resource constraint processing and cross-mutation probability adaptive adjustment mechanism;

[0011] An energy consumption prediction module, configured to perform multi-scale feature extraction on the collected time series data by using a deep spatio-temporal convolutional neural network, and predict the energy consumption value in the future process stage;

[0012] An anomaly detection module, configured to detect data anomalies in the rolling process in real time based on the isolation forest algorithm, and dynamically update the anomaly determination threshold in combination with the sliding window technology;

[0013] An optimization feedback module, configured to dynamically adjust the scheduling strategy through a reinforcement learning framework, and the framework includes state space modeling, reward function design and policy gradient optimization mechanism.

[0014] Preferably, in the data acquisition module, the arrangement method of the temperature sensor array is: micro thermocouples are embedded at equal intervals along the axial and circumferential directions on the surface of the roll, and the temperature data is uploaded to the edge computing node in real time through a wireless transmission protocol.

[0015] Preferably, in the dynamic scheduling module, the improved genetic algorithm specifically includes the following steps:

[0016] Generate an initial population based on process priority, where the chromosome coding uses a mixed coding of binary and real numbers;

[0017] Introduce a dynamic penalty function to handle equipment load overrun and process conflict constraints;

[0018] Use the simulated annealing strategy to adaptively adjust the crossover and mutation probabilities, and the calculation formula is:

[0019]

[0020] where, P c is the dynamically adjusted crossover probability, P c0 is the initial crossover probability, P m is the dynamically adjusted mutation probability, P m0is the initial mutation probability, α and β are attenuation coefficients, and t is the number of iterations.

[0021] Preferably, in the energy consumption prediction module, the deep spatio-temporal convolutional neural network includes the following structure:

[0022] The input layer receives time series data and extracts local time series features through a one-dimensional convolutional layer;

[0023] The spatio-temporal attention mechanism layer dynamically assigns weights to different sensor data;

[0024] The dilated convolutional layer expands the receptive field to capture long-term dependencies;

[0025] The output layer predicts the energy consumption value through a fully connected network regression.

[0026] Preferably, in the anomaly detection module, the improvement of the isolation forest algorithm includes:

[0027] Divide the data subset based on a sliding window, and independently construct isolation trees for each subset;

[0028] Introduce the Mahalanobis distance to replace the Euclidean distance to eliminate the dimensional difference of sensor data;

[0029] Dynamically update the anomaly threshold, and the calculation formula is:

[0030] T th = μ score + k·σ score

[0031] Where, T th is the dynamically updated anomaly determination threshold, μ score is the mean of historical anomaly scores, σ score is the standard deviation, and k is the dynamic adjustment coefficient.

[0032] Preferably, in the optimization feedback module, the reward function of the reinforcement learning framework is designed as:

[0033]

[0034] Where, R is the reinforcement learning reward value, E actual is the actual energy consumption, E pred is the predicted energy consumption, T idle is the device idle time, Δ violate is the constraint violation amount, T total is the total scheduling time, and w1, w2, and w3 are weight coefficients.

[0035] Preferably, the dynamic scheduling module further includes a conflict resolution sub-module for generating an alternative scheduling plan using a fuzzy logic rule base when process conflicts occur. The rule base includes rolling force thresholds, temperature tolerance intervals, and equipment priority levels.

[0036] Preferably, the energy consumption prediction module further includes a data augmentation sub-module for processing the input time series data as follows:

[0037] Generating extended data with noise through time warping techniques;

[0038] Synthesizing virtual data under extreme working conditions using a generative adversarial network;

[0039] Applying wavelet transform to remove high-frequency noise and retain low-frequency trend components.

[0040] Preferably, the anomaly detection module further includes a false alarm filtering sub-module for smoothing the anomaly scores output by the isolation forest through a Kalman filter and removing physically infeasible anomaly points in combination with the process knowledge base.

[0041] Preferably, the system further includes a visualization interaction module for real-time displaying the rolling process status through a three-dimensional virtual twin model. The model integrates sensor data, scheduling plans, and energy consumption prediction results and supports multi-dimensional data drilling and analysis.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] In terms of data acquisition, by embedding key nodes of rolling mill equipment in a spatial grid distribution manner through a distributed sensor network, temperature, pressure, speed, and energy consumption data can be comprehensively and accurately collected. For example, a temperature sensor array evenly embeds micro-thermocouples along the axial and circumferential directions on the surface of the roll and wirelessly transmits the data, ensuring the real-time and accuracy of temperature data and providing a reliable basis for subsequent precise analysis. This enables enterprises to grasp various parameters of the rolling process in real time, promptly discover potential problems, and lay a foundation for optimizing the production process.

[0044] The dynamic scheduling module is based on an improved genetic algorithm, with a process priority encoding, a dynamic resource constraint handling, and a cross-mutation probability adaptive adjustment mechanism, effectively solving the deficiencies of traditional scheduling methods. It generates an initial population based on process priorities, uses a hybrid encoding of binary and real numbers to improve the algorithm's search efficiency; introduces a dynamic penalty function to handle equipment load overrun and process conflict constraints, ensuring the orderly progress of production; and simulates an annealing strategy to adaptively adjust the crossover and mutation probabilities, enabling the algorithm to better converge to the optimal solution. In addition, the conflict resolution sub-module uses a fuzzy logic rule base to generate alternative scheduling plans during process conflicts, greatly improving the flexibility and adaptability of production scheduling, reducing equipment idle time, increasing equipment utilization rate, thereby enhancing the overall production efficiency and reducing production costs.

[0045] The energy consumption prediction module uses a deep spatio-temporal convolutional neural network. Through the collaborative work of a one-dimensional convolutional layer, a spatio-temporal attention mechanism layer, a dilated convolutional layer, and an output layer, it can perform multi-scale feature extraction on the collected time-series data and accurately predict the energy consumption value in the future process stage. The data augmentation sub-module further improves the prediction accuracy and the generalization ability of the model. This enables enterprises to plan energy supply in advance, reasonably arrange production plans, avoid energy waste, reduce energy costs, and enhance the scientific nature and initiative of enterprises in energy management.

[0046] The anomaly detection module is based on an improved isolation forest algorithm, combined with a sliding window technique to dynamically update the anomaly determination threshold, and can detect data anomalies in the rolling process in a timely and accurate manner. The false alarm filtering sub-module effectively eliminates false alarms through the cooperation of a Kalman filter and a process knowledge base, improving the reliability of anomaly detection. This helps enterprises to promptly discover equipment failures and process anomalies, take corresponding measures for repair, avoid the occurrence of production accidents, and ensure the continuity of production and product quality.

[0047] The optimization feedback module dynamically adjusts the scheduling strategy through a reinforcement learning framework and optimizes the scheduling plan in real time according to the actual production situation. The reward function comprehensively considers factors such as the difference between the actual energy consumption and the predicted energy consumption, equipment idle time, and constraint violation amount, guiding the scheduling strategy to develop in a more optimal direction and further improving the performance and efficiency of the system. The visualization interaction module uses a three-dimensional virtual twin model to display the rolling process status in real time, integrates various data, and supports multi-dimensional data drilling analysis, providing an intuitive and comprehensive production monitoring interface for operators. This facilitates operators to promptly understand the production situation, make accurate decisions, and improves the convenience and scientific nature of production management. Description of the Drawings

[0048] Figure 1 It is the working principle diagram of the intelligent scheduling and energy consumption prediction system for the electrical steel strip rolling process described in the present invention;

[0049] Figure 2 It is a step diagram of improving the isolation forest algorithm in the anomaly detection module;

[0050] Figure 3 It is a workflow diagram of the reinforcement learning framework in the optimization feedback module;

[0051] Figure 4 It is a workflow diagram of the conflict resolution sub-module in the dynamic scheduling module. Specific implementation manners

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0053] Please refer to Figures 1-4 , the present invention provides an intelligent scheduling and energy consumption prediction system for the rolling process of electrical steel strips, and its overall implementation solution is as follows:

[0054] The data acquisition module collects temperature, pressure, speed, and energy consumption data in the rolling process in real time through a distributed sensor network. This sensor network is composed of a temperature sensor array, a pressure sensor array, and a speed encoder, and each sensor is embedded in the key nodes of the rolling mill equipment in a spatial grid distribution manner. Such a distribution can comprehensively and accurately obtain various key data during the operation of the rolling mill, providing a reliable basis for subsequent analysis and decision-making. For example, the temperature sensor can monitor the temperature change of the roll during the rolling process, the pressure sensor can sense the magnitude of the rolling force in real time, and the speed encoder can accurately measure the rolling speed of the steel strip.

[0055] The dynamic scheduling module performs multi-objective optimization scheduling on the rolling process based on an improved genetic algorithm. This improved genetic algorithm has a process priority encoding, a dynamic resource constraint processing, and a cross-mutation probability adaptive adjustment mechanism. Through these mechanisms, this module can comprehensively consider various factors, such as the load capacity of the equipment, the sequence between processes, etc., so as to formulate a more reasonable scheduling plan and improve the overall efficiency of the rolling process.

[0056] The energy consumption prediction module uses a deep spatio-temporal convolutional neural network to perform multi-scale feature extraction on the collected time-series data and predict the energy consumption value in the future process stage. The deep spatio-temporal convolutional neural network can effectively mine the potential features in the data, accurately capture the change laws of the data in time and space, and then achieve accurate prediction of the energy consumption, providing strong support for the enterprise to reasonably arrange production and control costs.

[0057] The anomaly detection module uses the Isolation Forest algorithm to detect data anomalies in the rolling process in real time and combines the sliding window technique to dynamically update the anomaly determination threshold. This module can promptly identify anomalies in the rolling process, such as equipment failures and abnormal process parameters, enabling staff to take timely measures for handling and ensuring the smooth progress of production.

[0058] The optimization feedback module dynamically adjusts the scheduling strategy through a reinforcement learning framework. This framework includes state space modeling, reward function design, and policy gradient optimization mechanisms. Through continuous learning and optimization, the scheduling strategy can be dynamically adjusted according to the actual production situation, further improving the performance and efficiency of the system.

[0059] The following further illustrates the implementation of the present invention in combination with Embodiments 1 to 5.

[0060] Embodiment 1:

[0061] In this embodiment, the specific implementation details of the data acquisition module and the dynamic scheduling module are further elaborated.

[0062] For the data acquisition module, the layout of the temperature sensor array is crucial for the accuracy and reliability of the acquired data. Micro thermocouples are embedded at equal intervals along the axial and circumferential directions on the roll surface. This layout can comprehensively and evenly monitor the temperature changes at different positions on the roll surface. The micro thermocouples can quickly sense temperature changes and convert the temperature signals into electrical signals. Through the wireless transmission protocol, these temperature data can be uploaded to the edge computing node in real time, avoiding the problems of complex wiring and easy damage caused by wired transmission, and also facilitating the rapid processing and transmission of data.

[0063] In the dynamic scheduling module, the improved genetic algorithm is the core technology. When generating the initial population based on the operation priority, the chromosome encoding adopts a mixed encoding of binary and real numbers. This encoding method combines the advantages of binary encoding and real number encoding. Binary encoding is beneficial for crossover and mutation operations in the genetic algorithm, while real number encoding can more intuitively represent actual parameters such as the operation priority, improving the search efficiency and accuracy of the algorithm.

[0064] In dealing with equipment load overlimit and operation conflict constraints, a dynamic penalty function is introduced. When the equipment load exceeds the limit or conflicts occur between operations, the dynamic penalty function will penalize the corresponding chromosomes according to the actual situation, reducing their competitiveness in the population and prompting the algorithm to search for the optimal solution in the direction of meeting the constraint conditions.

[0065] The adaptive adjustment of the crossover and mutation probabilities is another key feature of this algorithm. The simulated annealing strategy is used to adaptively adjust the crossover and mutation probabilities, and the calculation formula is:

[0066]

[0067] Among them, P c is the crossover probability after dynamic adjustment, which determines the probability of crossover operation between two chromosomes in the genetic algorithm. The crossover operation can exchange part of the genes between different chromosomes, thus generating new individuals and increasing the diversity of the population. P c0 is the initial crossover probability, which is the crossover probability value set at the beginning of the algorithm. α is the attenuation coefficient, which is used to control the attenuation speed of the crossover probability with the iteration number t. As the number of iterations increases, the crossover probability gradually decreases. This is because in the initial stage of the algorithm, a larger crossover probability is needed to explore a wider solution space; while in the later stage of the algorithm, in order to avoid destroying the relatively good solutions that have been found, the crossover probability needs to be gradually reduced.

[0068] P m is the mutation probability after dynamic adjustment. The mutation operation can randomly change some genes in the chromosome to prevent the algorithm from falling into a local optimal solution. P m0 is the initial mutation probability, which is the mutation probability value set at the beginning of the algorithm. β is the attenuation coefficient, which is used to control the change speed of the mutation probability with the iteration number t. Similar to the crossover probability, the mutation probability is larger in the initial stage of the algorithm and gradually decreases in the later stage.

[0069] In addition, the dynamic scheduling module also includes a conflict resolution sub-module. When process conflicts occur, this sub-module uses a fuzzy logic rule base to generate an alternative scheduling plan. The rolling force threshold, temperature tolerance interval, and equipment priority level in the rule base are important bases for formulating alternative plans. For example, when the rolling force exceeds the set threshold and the temperature is outside the tolerance interval, according to the equipment priority level, the process with a lower priority is adjusted first to avoid equipment damage and ensure product quality. In this way, the process conflict problem can be quickly and effectively solved in a complex production environment, ensuring the continuity and stability of production.

[0070] Embodiment 2:

[0071] The deep spatio-temporal convolutional neural network of the energy consumption prediction module has a unique structure. The input layer receives the collected time-series data, which contains information such as temperature, pressure, and speed changing with time during the rolling process. The one-dimensional convolutional layer is responsible for extracting local time-series features. It slides the convolutional kernel on the time-series data and performs a convolutional operation on the local data to capture the change trends and features of the data in a short period of time. For example, it can discover the rapid rise or fall trend of temperature in a small period of time, as well as the fluctuation of pressure.

[0072] The spatio-temporal attention mechanism layer is a key part of this network. It can dynamically allocate weights to different sensor data. During the actual rolling process, the importance of different sensor data for energy consumption prediction may vary. For example, in some cases, temperature data may have a greater impact on energy consumption; while in other cases, rolling speed data may be more critical. The spatio-temporal attention mechanism layer automatically assigns appropriate weights to different sensor data by learning the features and patterns of the data, highlighting the role of important data and improving the accuracy of prediction.

[0073] The dilated convolution layer is used to expand the receptive field to capture long-term dependencies. The receptive field of the traditional convolution layer is limited and it is difficult to capture long-term features in the data. The dilated convolution layer introduces holes in the convolution kernel, enabling the convolution operation to obtain data information in a larger range, and thus being able to discover dependencies in the data on a longer time scale. For example, it can discover the periodic energy consumption change patterns that occur at regular intervals during the rolling process.

[0074] The output layer predicts the energy consumption value through a fully connected network regression. The fully connected network comprehensively processes the features extracted by the previous layers and finally outputs the predicted energy consumption value.

[0075] The energy consumption prediction module also includes a data augmentation sub-module. Extended data with noise is generated through time warping technology. The time warping technology can simulate the possible time series changes in actual production, such as small time deviations in data acquisition, etc., to generate extended data with a certain amount of noise, increasing the diversity of the data and improving the generalization ability of the model. An adversarial generative network is used to synthesize virtual data under extreme working conditions. In actual production, extreme working conditions may rarely occur, but energy consumption prediction under these conditions is equally important. The adversarial generative network can generate virtual data under extreme temperature, pressure and other conditions, enabling the model to learn the data characteristics under these special conditions, thereby improving the prediction ability of the model under extreme working conditions. Wavelet transform is applied to remove high-frequency noise and retain low-frequency trend components. Wavelet transform can effectively separate high-frequency noise and low-frequency trends in the data, removing noise interference and enabling the model to more accurately learn the essential characteristics of the data, improving the accuracy of prediction.

[0076] Example 3:

[0077] The anomaly detection module is based on the Isolation Forest algorithm and has been improved in many aspects in practical applications. The data is divided into subsets based on a sliding window, and an isolation tree is constructed independently for each subset. The sliding window technique can dynamically select data for analysis. As time goes by, the window slides continuously, and new data is included in the analysis scope. An isolation tree is constructed independently for each subset, which can capture the local features and distribution of the data more meticulously. For example, during a certain period, if there are abnormal fluctuations in the rolling speed, the isolation tree can be constructed by dividing the data subset through the sliding window to detect this abnormal situation in a timely manner.

[0078] The Mahalanobis distance is introduced to replace the Euclidean distance to eliminate the dimensional difference of sensor data. In actual production, the data collected by different sensors has different dimensions. For example, the unit of temperature is degrees Celsius, and the unit of pressure is Pascal, etc. The Euclidean distance will produce deviations when dealing with data with different dimensions, while the Mahalanobis distance takes into account the covariance structure of the data, can eliminate the influence of dimensional differences, and can measure the similarity and difference between data points more accurately, thus improving the accuracy of anomaly detection.

[0079] The anomaly threshold is updated dynamically, and the calculation formula is:

[0080] T th = μ score + k·σ score

[0081] Among them, T th is the dynamically updated anomaly determination threshold, which is a key indicator for judging whether the data is abnormal. μ score is the mean of historical anomaly scores, which reflects the average level of anomaly scores in historical data. σ score is the standard deviation, which is used to measure the dispersion degree of historical anomaly scores. k is a dynamically adjusted coefficient, which is adjusted according to the actual production situation and experience. For example, when the production environment is relatively stable, k can be appropriately reduced to improve the sensitivity of anomaly detection; while when the production environment fluctuates greatly, k can be appropriately increased to avoid excessive false alarms.

[0082] The anomaly detection module also includes a false alarm filtering sub-module. The anomaly scores output by the Isolation Forest are smoothed through a Kalman filter. The Kalman filter is an optimal linear filter, which can use the historical information and current observations of the data to make an optimal estimate of the anomaly scores, remove noise and fluctuations, and make the anomaly scores more stable and reliable. Combining with the process knowledge base, physically infeasible anomaly points are eliminated. The process knowledge base contains various physical limitations and empirical knowledge of the rolling process. For example, according to process knowledge, a certain combination of temperature and pressure is impossible to occur in actual production. If the Isolation Forest algorithm detects such a data point as an anomaly, combining with the process knowledge base can determine that this is a false alarm and thus eliminate it to improve the reliability of anomaly detection.

[0083] Example 4:

[0084] In the reinforcement learning framework of the optimization feedback module, the design of the reward function plays a key role in optimizing the scheduling strategy. The reward function formula is:

[0085]

[0086] where R is the reinforcement learning reward value, which is an important indicator to measure the quality of the scheduling strategy. E actual is the actual energy consumption, that is, the energy actually consumed during the rolling process, which reflects the energy usage under the current scheduling strategy. E pred is the predicted energy consumption, which is the energy consumption value predicted by the energy consumption prediction module. By comparing the actual energy consumption and the predicted energy consumption, the accuracy of the prediction can be evaluated, and at the same time, it can also reflect the control effect of the scheduling strategy on energy consumption. w1 is the weight coefficient, which is used to adjust the importance of the difference between the actual energy consumption and the predicted energy consumption in the reward function. If you want to pay more attention to the accuracy of energy consumption prediction, w1 can be appropriately increased; if you are more concerned about other factors, w1 can be correspondingly decreased.

[0087] T idle is the equipment idle time, T total is the total scheduling time, represents the proportion of the equipment idle time in the total scheduling time. Too long equipment idle time will reduce production efficiency, so this term is used to encourage the scheduling strategy to reduce the equipment idle time and improve the equipment utilization rate. w2 is the weight coefficient, which is used to control the weight of the equipment idle time proportion in the reward function.

[0088] Δ violate is the constraint violation amount, ∑Δ violate represents the sum of all constraint violation amounts. In actual production, there are various constraint conditions, such as equipment load limits, process sequence, etc. When the scheduling strategy violates these constraints, a constraint violation amount will be generated. w3 is the weight coefficient, which is used to measure the influence degree of the constraint violation situation on the reward function. If you want to strictly abide by the constraint conditions, w3 can be set larger; if the requirement for the strictness of the constraints is relatively low, w3 can be appropriately reduced.

[0089] Through this reward function, the reinforcement learning framework can dynamically adjust the scheduling strategy according to the actual production situation. When the actual energy consumption is close to the predicted energy consumption, the equipment idle time is reduced, and the constraint violation amount is decreased, the reward value will increase, indicating that the current scheduling strategy is optimized; on the contrary, the reward value will decrease, prompting the algorithm to find a better scheduling strategy, so as to achieve efficient scheduling and optimization of the rolling process.

[0090] Example 5:

[0091] The visualization interaction module displays the rolling process status in real time through a 3D virtual twin model. The 3D virtual twin model integrates sensor data, scheduling schemes, and energy consumption prediction results, providing an intuitive and comprehensive production monitoring interface for operators.

[0092] The sensor data is transmitted to the 3D virtual twin model in real time, and information such as the temperature and pressure of the rollers and the rolling speed of the steel strip is presented in a visual manner. For example, by using different colors and values to display the temperature at different positions of the rollers, operators can clearly see whether there are abnormal conditions such as local overheating of the rollers at a glance. The pressure data can be displayed through charts or dynamically changing values, facilitating operators to understand the magnitude and change trend of the rolling force at any time.

[0093] The scheduling scheme is presented in an intuitive way in the 3D virtual twin model to show the execution sequence of the processes and the allocation of equipment. Operators can clearly see which equipment each process is carried out on and the sequence relationship between the processes. If the scheduling scheme changes, the model will be updated in real time, enabling operators to promptly grasp the latest production arrangements.

[0094] The energy consumption prediction results are also integrated into the 3D virtual twin model. Through charts or prediction curves, operators can understand the energy consumption situation in future process stages in advance, thereby reasonably arranging energy supply and production plans. For example, if it is predicted that the energy consumption will increase significantly during a certain period, operators can adjust the equipment parameters or optimize the scheduling scheme in advance to reduce energy consumption.

[0095] This model also supports multi-dimensional data drilling analysis. Operators can click on various elements in the model to obtain more detailed data information. For example, by clicking on a certain roller, the temperature change curve, pressure data, and related equipment operating status at different time points of the roller can be viewed. Through multi-dimensional data drilling analysis, operators can deeply understand various details in the production process, promptly discover potential problems, and make more accurate decisions, improving the management level and efficiency of production.

[0096] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or sequence between these entities or operations. Moreover, the terms "including", "comprising", or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or elements inherent to such process, method, article, or device.

[0097] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent scheduling and energy consumption prediction system for the rolling process of electrical steel strip, characterized in that, It includes the following modules: The data acquisition module is used to collect temperature, pressure, speed and energy consumption data in the rolling process in real time through a distributed sensor network. The sensor network includes a temperature sensor array, a pressure sensor array and a speed encoder, and each sensor is embedded in key nodes of the rolling mill equipment in a spatial grid distribution manner; The dynamic scheduling module is used to perform multi-objective optimization scheduling on the rolling process based on an improved genetic algorithm. The improved genetic algorithm includes process priority coding, dynamic resource constraint processing and cross-mutation probability adaptive adjustment mechanism; The energy consumption prediction module is used to extract multi-scale features from the collected time-series data by using a deep spatio-temporal convolutional neural network and predict the energy consumption value in the future process stage; The anomaly detection module is used to detect data anomalies in the rolling process in real time based on the isolation forest algorithm and dynamically update the anomaly determination threshold in combination with the sliding window technology; The optimization feedback module is used to dynamically adjust the scheduling strategy through a reinforcement learning framework. The framework includes state space modeling, reward function design and policy gradient optimization mechanism.

2. The system according to claim 1, wherein In the data acquisition module, the arrangement method of the temperature sensor array is as follows: miniature thermocouples are embedded equidistantly along the axial and circumferential directions on the surface of the roll, and the temperature data is uploaded to the edge computing node in real time through a wireless transmission protocol.

3. The system according to claim 2, wherein In the dynamic scheduling module, the improved genetic algorithm specifically includes the following steps: Generate an initial population based on process priority, where the chromosome coding uses a hybrid coding of binary and real numbers; Introduce a dynamic penalty function to handle equipment load overrun and process conflict constraints; Use the simulated annealing strategy to adaptively adjust the crossover and mutation probabilities. The calculation formula is: P c = P c0 ·e -α·t , Among them, P c is the crossover probability after dynamic adjustment, P c0 is the initial crossover probability, P m is the mutation probability after dynamic adjustment, P m0 is the initial mutation probability, α and β are attenuation coefficients, and t is the number of iterations.

4. The system according to claim 1, characterized in that, In the energy consumption prediction module, the deep spatio-temporal convolutional neural network includes the following structure: The input layer receives time-series data and extracts local time-series features through a one-dimensional convolutional layer; The spatio-temporal attention mechanism layer dynamically assigns weights to different sensor data; The dilated convolutional layer expands the receptive field to capture long-term dependency relationships; The output layer predicts the energy consumption value through a fully connected network regression.

5. The system according to claim 1, wherein In the anomaly detection module, the improvement of the isolation forest algorithm includes: Divide data subsets based on a sliding window, and each subset independently constructs an isolation tree; Introduce the Mahalanobis distance to replace the Euclidean distance to eliminate the dimension difference of sensor data; Dynamically update the anomaly threshold. The calculation formula is: T th = μ score + k·σ score Among them, T th is the dynamically updated anomaly determination threshold, μ score is the mean of historical anomaly scores, σ score is the standard deviation, and k is the dynamic adjustment coefficient.

6. The system according to claim 1, wherein In the optimization feedback module, the reward function of the reinforcement learning framework is designed as: Among them, R is the reinforcement learning reward value, E actual is the actual energy consumption, E pred is the predicted energy consumption, T idle is the device idle time, Δ violate is the constraint violation amount, T total is the total scheduling time, and w1, w2, w3 are weight coefficients.

7. The system according to claim 3, wherein The dynamic scheduling module also includes a conflict resolution sub-module, which is used to generate an alternative scheduling plan using a fuzzy logic rule base when there is a process conflict. The rule base includes rolling force thresholds, temperature tolerance intervals and equipment priority levels.

8. The system according to claim 4, characterized in that, The energy consumption prediction module also includes a data augmentation sub-module, which is used to process the input time-series data as follows: Generate extended data with noise through time warping technology; Use a generative adversarial network to synthesize virtual data under extreme working conditions; Apply wavelet transform to remove high-frequency noise and retain low-frequency trend components.

9. The system according to claim 5, wherein The anomaly detection module also includes a false alarm filtering sub-module, which is used to smooth the anomaly scores output by the isolation forest through a Kalman filter and eliminate physically infeasible anomaly points in combination with the process knowledge base.

10. The system according to claim 1, wherein The system further includes a visual interaction module for real-time displaying the rolling process status through a three-dimensional virtual twin model. The model integrates sensor data, scheduling schemes, and energy consumption prediction results, and supports multi-dimensional data drilling and analysis.

Images