Iron phosphate preparation energy-saving control system based on energy consumption scheduling model

Abstract

The invention belongs to the technical field of iron phosphate preparation, and discloses an energy-saving control system for iron phosphate preparation based on an energy consumption scheduling model. The system is composed of a data acquisition module, an energy consumption sensing module, a preparation process modeling module, an energy consumption prediction module, an energy-saving scheduling module, an intelligent execution module, a feedback correction module, a man-machine interaction module and a remote operation and maintenance module. The energy consumption sensing module intelligently senses an energy consumption state, the preparation process modeling and energy consumption prediction module accurately predicts energy consumption, the energy-saving scheduling module generates an optimal scheduling strategy, the intelligent execution module accurately executes an instruction, and the feedback correction module realizes closed-loop adaptive regulation and control; all the modules cooperatively operate, process parameters are adjusted in real time according to actual working conditions of iron phosphate preparation, energy consumption in the preparation process is remarkably reduced, the energy utilization rate is increased, and energy-saving optimization of iron phosphate preparation is achieved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of iron phosphate preparation, and particularly relates to an iron phosphate preparation energy-saving control system based on an energy consumption scheduling model. BACKGROUND

[0002] In the preparation process of iron phosphate, the energy consumption problem has been an important factor restricting the development of the industry. The traditional iron phosphate preparation system often adopts a rough energy consumption management mode, lacks precise regulation and optimization of energy consumption, and causes serious energy waste. According to industry statistics, the energy utilization rate of the traditional preparation process is 15%-25% lower than the advanced level, which not only causes a large amount of energy consumption, but also makes the production cost high, seriously affecting the economic benefit and market competitiveness of enterprises. At the same time, the traditional system has obvious deficiencies in real-time acquisition of process parameters. The sensors used have low precision and low sampling frequency, and it is difficult to realize real-time high-precision acquisition of key process parameters such as power, temperature, and flow, resulting in large errors and hysteresis in the acquired process data. In terms of intelligent perception of energy consumption, the traditional system lacks effective data fusion and feature extraction means, and cannot accurately perceive the change trend and abnormal situation of the energy consumption state, making it difficult to comprehensively and deeply analyze the energy consumption. In terms of modeling and prediction of the preparation process, the traditional system mostly uses simple linear models or empirical models, which cannot accurately describe the complex nonlinear dynamic characteristics in the preparation process of iron phosphate, and has low prediction accuracy for long-term energy consumption trends and short-term fluctuations, making it difficult to meet the needs of actual production. In terms of energy-saving scheduling, the traditional system often relies on manual experience for scheduling decisions, lacks scientific optimization algorithms and global scheduling strategies, and is difficult to achieve optimal balance between unit energy consumption, reaction time, and material yield, resulting in low efficiency and high energy consumption of the preparation process. SUMMARY

[0003] The purpose of the present application is to provide an iron phosphate preparation energy-saving control system based on an energy consumption scheduling model to solve the problems raised in the background.

[0004] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme: an iron phosphate preparation energy-saving control system based on an energy consumption scheduling model, which is composed of a data acquisition module, an energy consumption perception module, a preparation process modeling module, an energy consumption prediction module, an energy-saving scheduling module, an intelligent execution module, a feedback correction module, a man-machine interaction module, and a remote operation and maintenance module. The data acquisition module: through multi-channel terminal integration of power, temperature, flow pressure sensors and reaction kettle visual acquisition devices, high-frequency synchronous sampling, time stamp encoding and Modbus communication are used to realize precise synchronous data acquisition. Energy Consumption Perception Module: This module normalizes and preprocesses multi-source data, encodes features to construct multi-dimensional vectors, uses an attention mechanism to perceive dynamic trends, and uses autocorrelation to identify and locate abnormal process conditions. Preparation process modeling module: Adopting a dual-channel strategy, the gray model predicts long-term trends and the LSTM network captures short-term fluctuations. After dynamic weight fusion, the model is generalized with the help of rolling data updates. Energy consumption prediction module: Combined with modeling results, it uses the reinforcement learning strategy library to simulate multi-strategy energy consumption, evaluates the impact of strategies on unit energy consumption and process parameters, and improves prediction accuracy and adaptability through disturbance compensation; Energy-saving scheduling module: The energy-saving scheduling module uses the NSGA-II algorithm, with the goal of "minimizing unit energy consumption, optimizing reaction time, and maximizing material yield". It combines process constraints to construct a solution space and generates an optimized scheduling sequence to guide equipment operation; Intelligent execution module: The edge computing unit compiles scheduling instructions into PLC code, controls devices via Ethernet, and sets an abnormal feedback path. When the deviation exceeds the limit, it initiates secondary issuance or policy replacement to ensure stable operation. Feedback correction module: Compares actual and predicted energy consumption. If the deviation exceeds the limit for three consecutive cycles, the LSTM residual update fine-tunes the model parameters and updates the working condition mapping model to achieve closed-loop correction to ensure system stability and energy saving. Human-computer interaction module: provides a real-time energy consumption trend panel, process stage view, and strategy parameter configuration interface, supports strategy status display, history rollback, and security lock, improving operability and safety; Remote operation and maintenance module: Leveraging 5G to synchronize data with cloud platforms and implement policy linkage, it supports remote monitoring, early warning, updates, model deployment, fault diagnosis, etc., improving maintenance efficiency and system flexibility.

[0005] Preferably, the data acquisition module includes: (1) Module composition and function: The data acquisition module serves as the basis of the energy-saving control system and is used to collect key equipment and process parameters for ferric phosphate preparation in real time. It includes a multi-channel data acquisition terminal that integrates three power, four temperature, a set of flow and pressure sensor sub-units, and a reactor visual acquisition device to provide data support for system operation; (2) Functions of each subunit: The three-way power subunit uses high-precision sensors to monitor the power of the equipment; the four-way temperature subunit uses high-temperature resistant sensors to monitor the temperature of key parts; the flow and pressure subunit measures the material flow and reactor pressure; the visual device uses industrial cameras and image recognition to analyze the reaction status. Each unit works together to ensure data accuracy; (3) Data acquisition and communication: the module has high-frequency synchronous sampling and timestamp unified coding capability, and communicates with the central control unit through the Modbus bus. At the same time, the data standardization formula is adopted to normalize the multi-source data stream, realize cross-calculation characteristic unified analysis, and ensure high-time efficiency and high-precision synchronous data acquisition; Data standardization formula: ; In the formula: represents the original acquisition value of the process parameter of the type (such as temperature, pressure, power, etc.) at time ; represents the historical average value of the type parameter; represents the standard deviation of the type parameter; represents the normalized parameter value, which is used for subsequent perception and modeling processing.

[0006] Preferably, the energy consumption perception module comprises: (1) Data processing and vector construction: the energy consumption perception module fuses and processes multi-source data, first eliminates the dimensional difference through normalization preprocessing, so that the data has comparability, and lays a foundation for subsequent feature extraction. Then, key features are extracted through feature coding, and an energy consumption state vector construction formula is used to integrate power, temperature, flow, image and other features into a unified input format for use by the modeling module, fully expressing the comprehensive state of system energy consumption and physical working conditions; Energy consumption state vector construction formula: ; In the formula: is a multi-dimensional energy consumption state vector at time ; is the power consumption of the device; is the temperature vector of the reaction system; is the flow and pressure feature vector; is the stirring or sedimentation state coding vector based on image processing.

[0007] (2) Trend perception and abnormality positioning: the module introduces an attention mechanism, gives different weights to time series data, strengthens key change point identification, realizes dynamic energy consumption trend perception, and analyzes data correlation with the aid of autocorrelation identification function, perceives energy consumption abnormalities in advance and positions the process state, and provides a basis for parameter adjustment.

[0008] Preferably, the preparation process modeling module comprises: (1) Double-channel modeling strategy: The process modeling module is based on the actual acquisition of working conditions and energy consumption data, adopts a double-channel modeling strategy to build a dynamic response model, fully combines the advantages of gray prediction model and LSTM deep network, and lays a foundation for energy consumption prediction; (2) Long and short trends are modeled separately: The long-term trend is established by a gray prediction model (GM(1, N)), which is suitable for small sample and poor information problems, and can predict long-term energy consumption trends; The short-term fluctuations are modeled by an LSTM deep network, which uses its nonlinear fitting and memory capacity to capture short-term complex energy consumption fluctuation characteristics; (3) Model fusion and dynamic update: The outputs of the two models are fused through a dynamic weight coefficient a, and the weight is automatically adjusted according to the working conditions and prediction requirements to form a robust model. With the help of the hybrid prediction model fusion formula, the energy consumption trend is extracted at different time scales, and the model robustness and responsiveness are improved by combining rolling historical data for dynamic updating.

[0009] Hybrid prediction model fusion formula: ; In the formula: is the fused unit time energy consumption prediction value; is the short-term prediction result based on the LSTM model; is the long-term trend prediction based on the gray model GM(1, N); is the fusion coefficient, , the error is dynamically adjusted Preferably, the energy consumption prediction module comprises: (1) Multi-strategy simulation and evaluation: The energy consumption prediction module combines the modeling results, integrates a reinforcement learning strategy library, and supports multi-strategy feedforward simulation. By simulating and evaluating multiple process control strategies in the library, the unit energy consumption and process parameter response relationship under different strategy combinations are analyzed according to the energy consumption function, and the optimal strategy that adapts to the current working conditions is selected; (2) Disturbance correction and adaptive improvement: The module is provided with a disturbance factor compensation mechanism, which monitors external voltage fluctuations, device temperature lag and other nonlinear disturbances in real time, analyzes and corrects the disturbances with the help of the disturbance correction formula, effectively improves the accuracy and adaptability of the energy consumption prediction results, and ensures the stable operation of the system in complex industrial environments.

[0010] Energy consumption function and disturbance correction formula ; In the formula: is the predicted unit time energy consumption; is the energy consumption response function input about the current operation strategy (such as heating rate, stirring speed, etc.); For the disturbance correction term, consider the influence of external factors such as voltage fluctuation, load change, etc.

[0011] Preferably, the energy-saving scheduling module comprises: (1) Multi-objective optimization strategy construction: Based on the energy consumption prediction results, the energy-saving scheduling module adopts NSGA-II algorithm, takes "minimum unit energy consumption, optimal reaction time, and maximum material yield" as the multi-objective optimization objective function, combines with process constraints such as stirring speed range, and constructs the global scheduling solution space, laying the foundation for generating optimization strategy; Multi-objective optimization objective function (NSGA-II): ; In the formula: is the cumulative value of unit energy consumption (the smaller the better); is the reaction process time (the shorter the better); is the product yield (the higher the better); is the total prediction period; is the effective iron phosphate mass output; is the raw material mass input; (2) Optimization instruction sequence generation: search for optimization in the solution space through NSGA-II algorithm, generate non-dominated multi-objective scheduling strategy based on Pareto frontier, and get the optimized scheduling instruction sequence S. This sequence balances the three objectives under the constraint conditions, and is called by the control execution module to guide the equipment operation, realizing energy-saving optimization of the preparation process.

[0012] Preferably, the intelligent execution module comprises: (1) Instruction compilation and device control: The intelligent execution module compiles the optimized scheduling instructions into PLC compatible languages such as ST and LD through the built-in edge computing control unit, interacts with devices such as temperature controllers and mixers through Ethernet, and realizes precise control of process equipment; (2) Deviation feedback and strategy adjustment: The module has an abnormal feedback path, which uses the execution deviation feedback judgment formula to determine whether the execution action conforms to the scheduling strategy. If the control response deviates from the target beyond the threshold, it automatically starts secondary issuance or strategy replacement, and monitors in real time to ensure the stability and reliability of the preparation process.

[0013] Execution deviation feedback judgment formula: ; In the formula: is the current process execution deviation; is the actual execution result (such as stirring rate, temperature); is the scheduling instruction daily target value; is the allowed deviation threshold.

[0014] Preferably, the feedback correction module comprises: (1) Closed-loop control mechanism: the feedback correction module realizes the closed-loop adaptive control of the system. By comparing the actual and predicted energy consumption, if the deviation of three consecutive periods is out of limit, the model fine-tuning program is started, the parameters are updated by means of the model residual self-learning formula, and the working condition mapping model is optimized. Model residual self-learning formula: ; In the formula: is the energy consumption deviation between the model prediction and the actual value; is the current model parameter; is the learning rate; is the gradient of the residual with respect to the model parameter; (2) System stability guarantee: through continuous feedback correction, the LSTM or gray model parameters are updated by means of the residual self-learning mechanism, so that the system can adapt to the changes in the preparation process, guarantee the stability and energy saving of long-term operation, and improve the energy consumption prediction and process control precision.

[0015] Preferably, the human-computer interaction module comprises: (1) Interactive interface function division: the human-computer interaction module provides multi-level interactive interfaces, including an energy consumption real-time trend panel, a process stage view, and a strategy parameter configuration interface, which are respectively used for intuitively displaying the energy consumption trend, comprehensively presenting the preparation process stage and parameters, and meeting the adjustment of strategy parameters set according to different production needs. (2) Strategy management and safety mechanism: the module supports graphical display of the current control strategy execution state, and by means of the strategy rollback criterion formula, the historical strategy can be manually selected for rollback, so as to improve the safety fault tolerance capability of the system, and at the same time, the operation safety locking mechanism is provided, which is convenient for manual confirmation and optimization of the scheduling strategy result, and enhances the operability and safety.

[0016] Strategy rollback criterion formula: ; In the formula: is the current execution strategy; is the historical strategy at the moment; is the energy consumption deviation of the current strategy; is the maximum acceptable deviation, and the historical strategy can be manually selected for rollback, so as to improve the safety fault tolerance capability of the control system.

[0017] Preferably, the remote operation and maintenance module comprises: (1) 5G communication and remote monitoring: The remote operation module realizes data synchronization and policy linkage with the cloud control platform through the 5G communication module, supports remote monitoring of system operation state, pushing of early warning information and updating of strategy. The operator can remotely and real-timely check the system state, and timely handle the abnormality, thereby guaranteeing the timeliness of system monitoring; (2) Cloud cooperation and maintenance optimization: In cooperation with the industrial cloud platform, the energy consumption prediction model and scheduling algorithm can be remotely deployed, and remote one-key fault diagnosis, intelligent log analysis and versioned system upgrade are supported. In combination with the cloud control synchronization delay fault tolerance criterion, stable control can still be realized when the network is unstable, thereby improving the maintenance efficiency, reducing the maintenance cost, enhancing the system maintainability and reliability; Cloud control synchronization delay fault tolerance criterion: , else fallback to local control In the formula: is the current network delay; is the maximum network delay threshold that can be tolerated; fallback is switching to the local edge control mode if the delay exceeds the threshold.

[0018] The beneficial effects of the present application are as follows: 1. The present application realizes complete energy consumption optimization control closed loop by constructing: the data acquisition module realizes real-time and accurate acquisition of energy consumption data, the energy consumption perception module intelligently perceives the energy consumption state, the preparation process modeling and energy consumption prediction module accurately predicts the energy consumption, the energy-saving scheduling module generates the optimal scheduling strategy, the intelligent execution module accurately executes the instructions, and the feedback correction module realizes closed-loop adaptive regulation and control; the modules are cooperatively operated, the process parameters are real-timely adjusted according to the actual working conditions of the iron phosphate preparation, the system is operated in the state of the lowest energy consumption, the energy consumption in the preparation process is significantly reduced, the energy utilization rate is improved, the production cost is reduced, and the energy-saving optimization of the iron phosphate preparation is realized.

[0019] 2. The present application realizes real-time and high-precision acquisition of key process parameters such as power and temperature through the data acquisition module, provides accurate parameter information for the system, the energy consumption perception and preparation process modeling module deeply analyzes and models, helps the system master the preparation process law, the energy-saving scheduling module generates the optimal strategy, optimizes the process parameters while ensuring the product quality and reducing the fluctuation, the intelligent execution module accurately controls the process equipment, and the feedback correction module ensures the stable execution of the parameters through closed-loop regulation and control, thereby improving the preparation process stability together, the stable process conditions guarantee the product quality and consistency of the iron phosphate product, and effectively improve the product market competitiveness.

[0020] 3. The present invention utilizes a remote operation and maintenance module, leveraging 5G to synchronize data with cloud platforms and implement policy linkage, supporting remote monitoring, early warning push notifications, policy updates, fault diagnosis, and system upgrades. Operators can remotely manage and maintain systems, improving efficiency and reducing costs. The system's intelligent forecasting and optimized scheduling enable advance production planning, reducing downtime and adjustment time, and improving production efficiency. The user-friendly interface and convenient operation of the human-computer interaction module facilitate parameter monitoring and policy adjustments, further improving production management efficiency. Through the collaborative efforts of various modules, this energy-saving control system achieves energy-saving optimization, process stability, and efficient production in ferric phosphate preparation, with significant economic and social benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The figure is a flow chart of the energy-saving control system for iron phosphate preparation based on the energy consumption scheduling model of the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] like Figure 1 As shown, an embodiment of the present invention provides an energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model, which consists of a data acquisition module, an energy consumption perception module, a preparation process modeling module, an energy consumption prediction module, an energy-saving scheduling module, an intelligent execution module, a feedback correction module, a human-computer interaction module, and a remote operation and maintenance module; Data acquisition module: Through the multi-channel terminal, it integrates power, temperature, flow and pressure sensors and reactor visual acquisition devices, and realizes accurate and synchronous data acquisition through high-frequency synchronous sampling, time stamp encoding and Modbus communication; Energy Consumption Perception Module: This module normalizes and preprocesses multi-source data, encodes features to construct multi-dimensional vectors, uses an attention mechanism to perceive dynamic trends, and uses autocorrelation to identify and locate abnormal process conditions. Preparation process modeling module: Adopting a dual-channel strategy, the gray model predicts long-term trends and the LSTM network captures short-term fluctuations. After dynamic weight fusion, the model is generalized with the help of rolling data updates. Energy consumption prediction module: Combined with modeling results, it uses the reinforcement learning strategy library to simulate multi-strategy energy consumption, evaluates the impact of strategies on unit energy consumption and process parameters, and improves prediction accuracy and adaptability through disturbance compensation; Energy-saving scheduling module: The energy-saving scheduling module uses the NSGA-II algorithm, with the goal of minimizing unit energy consumption, optimizing reaction time, and maximizing material yield, to construct the solution space in combination with process constraints and generate an optimized scheduling sequence to guide device operation. Intelligent execution module: The intelligent execution module compiles scheduling instructions into PLC code through the edge computing unit, controls the device via Ethernet, and sets up an abnormal feedback path. If the deviation exceeds the limit, secondary issuance or strategy replacement is started to ensure stable operation. Feedback correction module: Compare the actual and predicted energy consumption. If the deviation exceeds the limit for three consecutive periods, start the LSTM residual to update the model parameters and update the working condition mapping model to achieve closed-loop correction and ensure system stability and energy saving. Human-computer interaction module: Provides real-time trend panels, process stage views, and strategy parameter configuration interfaces, supports strategy state display, historical rollback, and security locking, and improves operability and security. Remote operation and maintenance module: With the help of 5G and cloud platform data synchronization and strategy linkage, remote monitoring, early warning, updating, model deployment, fault diagnosis, and other functions are supported to improve maintenance efficiency and system flexibility.

[0024] In the embodiments of the present application, a complete energy consumption optimization control closed loop is constructed. The data acquisition module accurately and timely acquires energy consumption data, the energy consumption perception module intelligently perceives the energy consumption state, the preparation process modeling and energy consumption prediction module accurately predicts energy consumption, the energy-saving scheduling module generates the optimal scheduling strategy, the intelligent execution module accurately executes the instructions, and the feedback correction module realizes closed-loop adaptive regulation and control. The modules work together to adjust the process parameters in real time based on the actual working conditions of the iron phosphate preparation, so that the system operates at the lowest energy consumption, significantly reduces the energy consumption of the preparation process, improves the energy utilization rate, and reduces the production cost. At the same time, the data acquisition module provides accurate parameter information for the system. Through in-depth analysis, modeling, generation of optimized strategies, accurate control of devices, and closed-loop regulation and control, the stability of the preparation process is improved, the product quality and consistency are guaranteed, and the market competitiveness is improved. In addition, the remote operation and maintenance module cooperates with the 5G and cloud platform to support remote management and maintenance, improve efficiency and reduce cost, and the intelligent prediction and optimization scheduling reduces downtime and adjustment time. The human-computer interaction module facilitates parameter monitoring and strategy adjustment. Through the cooperation of the modules, the system realizes energy-saving optimization, process stability, and efficient production of iron phosphate preparation, and has significant economic and social benefits.

[0025] The data acquisition module is the basis of the energy-saving control system and is responsible for real-time acquisition of key equipment and process parameters in the iron phosphate preparation process. The multi-channel data acquisition terminal integrates three power channels, four temperature channels, one set of flow and pressure sensing subunits, and a reaction kettle vision acquisition device based on image recognition, providing reliable data support for subsequent energy consumption analysis and process control of the system. The three-way power subunit adopts a power sensor with an accuracy of 0.1W to monitor the equipment power, the four-way temperature subunit selects a temperature sensor resistant to 200℃ with an error of ±0.5℃ to monitor the key positions, the flow and pressure subunit measures the material flow and reactor pressure through the sensor, and the visual device analyzes the reaction state by means of industrial camera and image recognition technology, and each unit cooperates to ensure the accuracy of data; The module has the ability of 100Hz high-frequency synchronous sampling and timestamp unified coding, communicates through Modbus bus, normalizes multi-source data stream by using data standardization formula, realizes cross-scale feature unified analysis, and ensures high-time-efficiency and high-precision data collection synchronization; Data standardization formula: ; In the formula: represents the original collection value of the process parameter of the first type (such as temperature, pressure, power, etc.) at time ; represents the historical average value of the parameter of the first type; represents the standard deviation of the parameter of the first type; represents the normalized parameter value, which is used for subsequent perception and modeling processing.

[0026] The energy consumption perception module is responsible for fusion processing and intelligent perception of multi-source data. First, the normalization preprocessing is performed, such as Z-score standardization, to eliminate the dimensional difference, so that different types and ranges of data have comparability, laying a foundation for feature extraction; Then, the key features such as power, temperature, flow, and image are extracted from the preprocessed data through feature coding, and the energy consumption state vector construction formula is used to integrate them into a unified input format for the modeling module, so as to comprehensively express the system energy consumption and physical working condition comprehensive state; The energy consumption perception module introduces the attention mechanism to give different weights to time series data, and strengthens the identification of key change points to realize dynamic energy consumption trend perception. At the same time, with the help of autocorrelation identification function, the data correlation is analyzed, the energy consumption anomaly is perceived in advance, and the process state is located, which provides a scientific basis for process parameter adjustment; Energy consumption state vector construction formula: ; In the formula: is a multi-dimensional energy consumption state vector at time ; is the equipment power consumption; is the temperature vector of the reaction system; is the flow and pressure feature vector; ​​​An image processing-based stirring or sedimentation state encoding vector.

[0027] The preparation process modeling module is based on the actual collected working conditions and energy consumption data, adopts a double-channel modeling strategy to construct a dynamic response model, fully combines the advantages of a gray prediction model and an LSTM deep network, and lays a foundation for energy consumption prediction. The long-term trend is established by a gray prediction model (GM (1, N)) suitable for small sample and poor information problems, and the long-term energy consumption trend in the preparation process of iron phosphate can be predicted and analyzed; the short-term fluctuation is modeled by an LSTM deep network with strong nonlinear fitting and memory capacity, so as to capture the short-term complex energy consumption fluctuation characteristics. The two model outputs are fused through a dynamic weight coefficient α, the weight is automatically adjusted according to different working conditions and prediction requirements, a robust model is formed, the energy consumption trend is extracted at different time scales by means of a hybrid prediction model fusion formula, and the model is dynamically updated combined with rolling historical data, so as to effectively improve the robustness and responsiveness of the model.

[0028] Hybrid prediction model fusion formula ; In the formula: is the fused unit time energy consumption prediction value; is a short-term prediction result based on an LSTM model; is a long-term trend prediction based on a gray model GM (1, N); is a fusion coefficient, is dynamically adjusted according to the error The energy consumption prediction module combines the modeling results, integrates a reinforcement learning strategy library, supports multi-strategy feedforward simulation, analyzes the unit energy consumption and process parameter response relationship under different strategy combinations according to the energy consumption function, and selects the optimal strategy adapted to the current working condition. At the same time, a disturbance factor compensation mechanism is provided in the module, which monitors nonlinear disturbances such as external voltage fluctuation and equipment temperature lag in real time, analyzes and corrects the disturbances by means of a disturbance correction formula, effectively improves the energy consumption prediction accuracy and adaptability, and ensures the stable operation of the system in complex industrial environments.

[0029] Energy consumption function and disturbance correction formula ; In the formula: is the predicted unit time energy consumption; is an energy consumption response function input about the current operation strategy (such as heating rate, stirring speed, etc.); is a disturbance correction term, which considers the influence of external factors such as voltage fluctuation and load change.

[0030] The energy-saving scheduling module adopts an NSGA-II multi-objective optimization algorithm to construct a global scheduling solution space based on the energy consumption prediction result, with "unit energy consumption minimization, reaction time optimization, and material yield maximization" as target functions, and in combination with process constraint conditions such as stirring speed range; The NSGA-II algorithm is used to search and optimize the solution space, a non-dominated multi-objective scheduling strategy is generated based on the Pareto front, and an optimized scheduling instruction sequence is obtained, which realizes the optimal balance of the three objectives under the condition of meeting the constraint conditions, and is called by the control execution module to guide the operation of the process equipment, so that the energy-saving optimization of the ferric phosphate preparation process is realized; Multi-objective optimization target function (NSGA-II): ; In the formula: is the cumulative value of unit time energy consumption (the smaller the better); is the reaction process time (the shorter the better); is the product yield (the higher the better); is the total prediction period; is the mass of effective ferric phosphate output; is the mass of raw material input.

[0031] The intelligent execution module compiles the optimized scheduling instructions into PLC compatible languages such as ST and LD through the built-in edge computing control unit, interacts with devices such as temperature controllers and stirrers through Ethernet, and realizes precise control of process equipment. The edge computing architecture supports local real-time processing, reduces data transmission delay, and improves control response speed; The intelligent execution module has an abnormal feedback path, which continuously monitors the matching degree of execution action and scheduling strategy by means of an execution deviation feedback judgment formula. If the control response deviates from the target beyond the limit, a secondary issuance or strategy replacement mechanism is automatically triggered to form a closed-loop control loop. Through real-time deviation detection and dynamic adjustment, the module effectively guarantees the stability and reliability of the preparation process, and ensures that the process parameters strictly follow the optimized scheduling instructions.

[0032] Execution deviation feedback judgment formula: ; In the formula: is the current process execution deviation; is the actual execution result (such as stirring rate, temperature); is the target value of the scheduling instruction; is the allowable deviation threshold.

[0033] The feedback correction module implements closed-loop adaptive control of the system. By comparing the energy consumption collected during actual execution with the predicted energy consumption, if the deviation exceeds the standard for three consecutive cycles, the model fine-tuning program is activated. The parameters of the LSTM or gray model are updated with the help of the model residual self-learning formula, thereby optimizing the working condition mapping model. By continuously providing feedback and making corrections, and utilizing a residual self-learning mechanism to continuously update model parameters, the system can better adapt to various changes in the ferric phosphate preparation process, effectively ensuring the long-term stability and energy efficiency of the system, and further improving the accuracy of energy consumption prediction and process control. Model residual self-learning formula: ; Where: is the energy consumption deviation between the model prediction and the actual value; is the current model parameter; is the learning rate; is the gradient of the residual with respect to the model parameters.

[0034] The human-computer interaction module provides a multi-level interactive interface, including a real-time energy consumption trend panel, a process stage view, and a strategy parameter configuration interface. The real-time energy consumption trend panel graphically displays energy consumption trends, the process stage view comprehensively presents the process parameters and execution status of each stage of preparation, and the strategy parameter configuration interface supports operators to set and adjust scheduling strategy parameters according to different production needs. The human-computer interaction module supports graphical display of the current control strategy execution status. With the help of the strategy rollback criterion formula, operators can manually select historical strategy rollback to improve the system's safety and fault tolerance. At the same time, through the operation of the safety locking mechanism, it is convenient to manually confirm and optimize the scheduling strategy results, thereby enhancing the system's operability and safety.

[0035] Policy rollback criterion formula ; Where: is the current execution strategy; For the history Moment strategy; is the energy consumption deviation of the current strategy; To ensure the maximum acceptable deviation, historical strategies can be manually selected for rollback to improve the safety and fault tolerance of the control system.

[0036] The remote operation and maintenance module uses a 5G communication module to achieve data synchronization and policy linkage with the cloud control platform, supporting remote monitoring of system operating status, pushing early warning information and updating policies. Operators can remotely view system status in real time and handle abnormalities in a timely manner to ensure the timeliness of system monitoring. At the same time, cooperate with the industrial cloud platform, can remotely deploy energy consumption prediction model and scheduling algorithm, support remote one-key fault diagnosis, intelligent log analysis and version system upgrade, combined with cloud control synchronization delay fault tolerance criterion, when the network is unstable, it can still be stable control, effectively improve the maintenance efficiency, reduce the maintenance cost, enhance the maintainability and reliability of the system; Cloud control synchronization delay fault tolerance criterion: , else fallback to local control In the formula: is the current network delay; is the maximum network delay threshold that can be tolerated; fallback is to switch to local edge control mode if the delay is out of limit.

[0037] It should be noted that in this text, relationship 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 order between the entities or operations. Moreover, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device.

[0038] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. An energy-saving control system for iron phosphate preparation based on an energy consumption scheduling model, characterized in that: The system consists of a data acquisition module, an energy consumption perception module, a preparation process modeling module, an energy consumption prediction module, an energy-saving scheduling module, an intelligent execution module, a feedback correction module, a human-computer interaction module, and a remote operation and maintenance module. Data acquisition module: This module uses a multi-channel terminal to synchronously collect power, temperature, flow, pressure, and image characteristics during the ferric phosphate preparation process at high frequency, and uses timestamp coding and the Modbus bus to achieve real-time data transmission. Energy Consumption Perception Module: This module performs normalization preprocessing and feature encoding on collected data, constructs a multi-dimensional energy consumption state vector, and combines attention mechanisms with autocorrelation analysis to identify dynamic trends and locate anomalies. Preparation process modeling module: used to model long-term trends and short-term fluctuations based on the grey model and LSTM neural network, and improve the generalization ability of the prediction model through weighted fusion and rolling update mechanism; Energy consumption prediction module: used to perform multi-strategy energy consumption simulation based on modeling results combined with the reinforcement learning strategy library, and use disturbance compensation algorithm to correct external interference to improve prediction accuracy; Energy-saving scheduling module: used to construct a scheduling strategy sequence with unit energy consumption, reaction time and productivity as optimization targets based on the NSGA-II multi-objective optimization algorithm; Intelligent execution module: used to translate the optimized scheduling strategy into PLC control instructions through the edge computing unit, and has deviation feedback and strategy replacement mechanisms; Feedback correction module: It is used to dynamically update modeling parameters based on the difference between actual and predicted energy consumption using a residual self-learning algorithm to achieve closed-loop control; Human-computer interaction module: used to provide energy consumption trends, process stages and strategy configuration interfaces, and supports strategy status monitoring, history rollback and security lock; Remote operation and maintenance module: With the help of 5G and cloud platform data synchronization and policy linkage, it supports remote monitoring, early warning, update, model deployment, and fault diagnosis.

2. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The data acquisition module includes: (1) Module composition and function: The data acquisition module serves as the basis of the energy-saving control system and is used to collect real-time data of key equipment and process parameters for ferric phosphate preparation. It includes a multi-channel data acquisition terminal that integrates three power, four temperature, a set of flow and pressure sensor sub-units, and a reactor visual acquisition device. (2) Functions of each subunit: The three-way power subunit uses high-precision sensors to monitor the power of the equipment; the four-way temperature subunit uses high-temperature resistant sensors to monitor the temperature of key parts; the flow and pressure subunit measures the material flow and reactor pressure; the visual device uses industrial cameras and image recognition to analyze the reaction status; (3) Data acquisition and communication: The module has the ability of high-frequency synchronous sampling and unified timestamp encoding, and communicates with the central control unit through the Modbus bus. At the same time, it uses data standardization formulas to normalize multi-source data streams and realize unified analysis of cross-dimensional features; Data normalization formula: ； Where: To express the time Moment Original collection values ​​of process parameters; To indicate the Historical mean of class parameters; To indicate the standard deviation of class parameters; represents the normalized parameter value.

3. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The energy consumption sensing module includes: (1) Data processing and vector construction: Multi-source data is fused and processed. Normalization preprocessing is performed to eliminate dimensional differences. Key features are then extracted through feature coding. The energy consumption state vector is used to construct a formula to integrate power, temperature, flow, and image features into a unified input format. Energy consumption state vector construction formula: ； Where: For in time Multi-dimensional energy consumption state vector; is the power consumption of the device; is the temperature vector of the reaction system; are the flow and pressure characteristic vectors; Encoding vectors for stirring or settling states based on image processing; (2) Trend perception and anomaly location: The module introduces an attention mechanism to assign different weights to time series data. At the same time, it uses the autocorrelation recognition function to analyze data associations, perceive energy consumption anomalies in advance, and locate process status.

4. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The preparation process modeling module includes: (1) Dual-channel modeling strategy: Based on the actual collected working conditions and energy consumption data, a dual-channel modeling strategy is used to build a dynamic response model, fully combining the advantages of the grey prediction model and the LSTM deep network; (2) Modeling long and short trends separately: the long-term trend is established by the grey prediction model (GM(1, N)), and the short-term fluctuation is modeled by the LSTM deep network, which uses its nonlinear fitting and memory capabilities to capture the short-term complex energy consumption fluctuation characteristics; (3) Model fusion and dynamic update: The outputs of the two models are fused through the dynamic weight coefficient α, and the weight is automatically adjusted according to the working conditions and forecast demand. With the help of the hybrid forecast model fusion formula, energy consumption trends are extracted at different time scales and dynamically updated in combination with rolling historical data; Hybrid prediction model fusion formula: ； Where: is the predicted value of energy consumption per unit time after fusion; is the short-term prediction result based on the LSTM model; It is the long-term trend forecast based on the grey model GM(1,N); is the fusion coefficient, .

5. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The energy consumption prediction module includes: (1) Multi-strategy simulation and evaluation: Combined with the modeling results, the reinforcement learning strategy library is integrated. Through simulation and evaluation of various process control strategies in the library, the relationship between unit energy consumption and process parameter response under different strategy combinations is analyzed based on the energy consumption function, and the optimal strategy adapted to the current working conditions is selected; (2) Disturbance correction and adaptability improvement: A built-in disturbance factor compensation mechanism monitors external voltage fluctuations and equipment temperature rise lag nonlinear interference in real time, and uses disturbance correction formulas to analyze and correct interference; Energy consumption function and disturbance correction formula ； Where: is the predicted energy consumption per unit time; For current operating strategies Input energy consumption response function; is the disturbance correction term.

6. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The energy-saving scheduling module includes: (1) Construction of multi-objective optimization strategy: Based on the energy consumption prediction results, the NSGA-II algorithm is used, with "minimizing unit energy consumption, optimizing reaction time, and maximizing material yield" as the multi-objective optimization objective function, combined with the process constraints of the stirring speed range, to construct the global scheduling solution space; Multi-objective optimization objective function (NSGA-II): ； Where: is the cumulative value of energy consumption per unit time; is the reaction process time; is the product yield; is the total forecast period; To produce effective iron phosphate quality; Input for raw material quality; (2) Optimized instruction sequence generation: The solution space search is optimized through the NSGA-II algorithm, and a non-dominated multi-objective scheduling strategy is generated based on the Pareto front to obtain an optimized scheduling instruction sequence. This sequence balances the three objectives under the constraints and is called by the control execution module to guide the operation of the equipment.

7. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The intelligent execution module includes: (1) Instruction compilation and equipment control: The intelligent execution module compiles the optimized scheduling instructions into ST and LDPLC compatible languages ​​through the built-in edge computing control unit, and interacts with the temperature controller and agitator equipment via Ethernet; (2) Deviation feedback and strategy adjustment: The intelligent execution module has an abnormal feedback path. It uses the execution deviation feedback judgment formula to determine whether the execution action complies with the scheduling strategy. If the deviation between the control response and the target exceeds the limit, it automatically initiates a secondary release or strategy replacement. Execution deviation feedback judgment formula: ； Where: Implement deviations for the current process; The actual execution results; is the daily standard value of the dispatch instruction; is the allowable deviation threshold.

8. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The feedback correction module includes: (1) Closed-loop control mechanism: Implement closed-loop adaptive control of the system, compare actual and predicted energy consumption, and if the deviation exceeds the limit for three consecutive cycles, start the model fine-tuning program and update the parameters with the help of the model residual self-learning formula; Model residual self-learning formula: ； ； Where: is the energy consumption deviation between the model prediction and the actual value; is the current model parameter; is the learning rate; is the gradient of the residual with respect to the model parameters; (2) System stability assurance: Through continuous feedback correction, the residual self-learning mechanism is used to update the LSTM or gray model parameters so that the system can adapt to changes in the preparation process.

9. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The human-computer interaction module includes: (1) Functional division of interactive interface: The human-computer interaction module provides a multi-level interactive interface, including a real-time trend panel for energy consumption, a process stage view, and a strategy parameter configuration interface to meet the strategic parameter setting and adjustment requirements of different production needs; (2) Strategy management and safety mechanism: The human-computer interaction module supports graphical display of the current control strategy execution status, and uses the strategy rollback criterion formula to select historical strategy rollback, while also having an operation safety locking mechanism; Policy rollback criterion formula: ； Where: is the current execution strategy; For the history Moment strategy; is the energy consumption deviation of the current strategy; is the maximum acceptable deviation.

10. The energy-saving control system for ferrous phosphate preparation based on an energy consumption scheduling model according to claim 1, characterized in that: The remote operation and maintenance module includes: (1) 5G communication and remote monitoring: With the help of 5G communication modules, data synchronization and policy linkage with the cloud control platform are achieved, supporting remote monitoring of system operation status, push of early warning information and update of policies; (2) Cloud collaboration and maintenance optimization: Cooperate with the industrial cloud platform to remotely deploy energy consumption prediction models and scheduling algorithms, support remote one-click fault diagnosis, intelligent log analysis and versioned system upgrades, and combine cloud control synchronization delay fault tolerance criteria to ensure stable control even when the network is unstable; Cloud control synchronization delay fault tolerance criteria: ,else fallback to local control Where: is the current network delay; is the maximum tolerable network delay threshold.

Images