A safety-optimized control method and control system for a mill operation

Abstract

The present application relates to mineral grinding equipment technical field, specifically provide a kind of safety optimization control method and control system of mill operation, comprising: acquisition mill in the process of operation Multi-source physical field signal and pre-processing is carried out to Multi-source physical field signal, obtain pre-processing signal, calculate the spatial motion state feature corresponding to time-frequency feature, establish the working condition identification model of mill to obtain the operating state of mill, generate dynamic safety constraint, real-time planning optimal operating point, control mill when meeting dynamic safety constraint, make the operating state of mill close to optimal operating point, mill operation produces Multi-source physical field signal is collected again, form closed-loop control.The present application is through multi-dimensional physical field perception technology, improves the transparency of mill internal operating state, uses dynamic constraint shrink mechanism, effectively avoids the occurrence of malignant mill blocking accident.

Description

Technical Field

[0001] This invention relates to the field of mineral grinding equipment technology, specifically providing a method and control system for optimizing the safety of mill operation. Background Technology

[0002] Grinding operations, as a critical link in mineral processing plants with the highest energy consumption and extremely stringent requirements for continuous operation, face numerous challenges. Due to frequent fluctuations in the hardness, particle size distribution, and moisture content of the ore fed into the mill during actual production, the optimal operating point of the mill constantly drifts, significantly increasing the difficulty of automatic control. Currently, grinding control technologies mostly rely on macroscopic signals such as electrical power and employ strategies like constant power control; however, these solutions have significant drawbacks. On the one hand, existing control systems cannot observe the spatial motion state of the grinding media inside the mill, making it difficult to distinguish between effective grinding and harmful operating conditions, resulting in overly conservative control strategies. On the other hand, the mill power has a non-linear relationship with the load; power may decrease in the initial stage of overload, easily misjudged as underload, increasing the risk of mill blockage accidents. Furthermore, fixed setpoints or static rules are difficult to adapt to changes in ore properties, failing to balance energy efficiency and production capacity. Therefore, developing an intelligent control method that can integrate internal physical state perception and dynamic optimization decision-making in a single operating mode to achieve continuous, adaptive, and safe operation of the grinding process is urgently needed. Summary of the Invention

[0003] To address the aforementioned problems, this invention provides a safe optimization control method and control system for mill operation. This invention utilizes multi-dimensional physical field sensing technology and a dynamic constraint contraction mechanism to enhance the transparency of the internal operating status, alleviate overload conditions, and effectively prevent the occurrence of severe mill blockage accidents.

[0004] The safety optimization control method for mill operation provided by this invention includes: S1: Collects multi-source physical field signals during the operation of the mill. The multi-source physical field signals include: mill cylinder vibration signal, acoustic emission signal, mill speed signal, mill angle signal, motor current signal, and motor power signal. S2: Preprocess the multi-source physical field signal to obtain the preprocessed signal; perform angle synchronization processing on the preprocessed signal to obtain the angle domain signal; extract the time-frequency features of the angle domain signal, and calculate the spatial motion state features corresponding to the time-frequency features. The spatial motion state features include the toe angle and shoulder angle of the grinding medium. S3: Establish the mill's operating condition feature vector based on the spatial motion state characteristics, and establish the mill's operating condition identification model based on the operating condition feature vector; the operating condition identification model is used to obtain the mill's operating status. S4: Generate dynamic safety constraints using the characteristic vector of the working condition and the multivariable derivative relationship; The multivariable derivative relationship is expressed as: ; Dynamic safety constraints are represented as: ; in, This represents the derivative of the rate of change of mill power with respect to load. An index representing the derivative relationship between the mill power change rate and the load change rate. Indicates control variables, Indicates comprehensive risk factors The lower boundary of the lower control variable. Indicates comprehensive risk factors The upper boundary of the lower control variable.

[0005] S5: Based on the mill's output, unit energy consumption, and comprehensive performance indicators, the optimal operating point is planned in real time; control commands are generated in real time using the mill's operating status and optimal operating point; when dynamic safety constraints are met, the operating status of the mill is controlled using control commands to make the mill's operating status close to the optimal operating point; the multi-source physical field signals generated by the mill's operation are collected according to the process of S1 to form closed-loop control.

[0006] Preferably, the preprocessing includes bandpass filtering and normalization, and the expression for the preprocessed signal is: ; in, Indicates the first Original signal of the road, This indicates the preprocessed signal. Indicates the filtering operator. This represents the mean of the corresponding signal. This represents the standard deviation of the corresponding signal.

[0007] Preferably, the angle synchronization processing method is: introducing the cylinder rotation angle. By using the mill rotation angle signal to establish a mapping relationship between time and rotation angle, the preprocessed time series signal is resampled to the rotation angle domain to obtain the angle domain signal corresponding to the spatial position of the mill.

[0008] Preferably, the expression for the time-frequency characteristics is: ; in, Represents the angular frequency variable. Represents a time variable. Indicates the first The time-frequency characteristic value of the signal in the time-frequency domain. Indicates the input signal channel number. This represents the time variable in the summation process. Represents the natural constant. Represents the window function. It represents the imaginary unit.

[0009] Preferably, the toe angle and shoulder angle The expressions are as follows: ; , ; in, Represents the impact energy density function. Indicates the threshold. This indicates the maximum effective turning angle.

[0010] Preferably, the expression for the operating condition feature vector is: ; in, This indicates the average power of the mill. This represents a load characterization metric. This represents the rate of change of power.

[0011] Preferably, in S4, dynamic safety constraints are generated using a comprehensive operational risk factor, which is expressed as: ; in, All represent weighting coefficients. Indicates continuous overload risk factor, This indicates the risk factor of media impact.

[0012] Preferably, a comprehensive operational risk factor is constructed based on the continuous overload risk factor and the medium shock risk factor; the expressions for the continuous overload risk factor and the medium shock risk factor are as follows: ; ; in, Represents the normalized mapping function. This indicates the preset safe threshold for shoulder angle. An index representing the derivative relationship between the mill power change rate and the load change rate. This represents the derivative of the mill load change rate with respect to time.

[0013] A safety optimization control system for mill operation is provided to realize a safety optimization control method for mill operation. The control system includes a multi-source sensing unit, an actuator unit, an edge computing and state reconstruction layer, a monitoring and dynamic constraint generation layer, a multi-source state detection module, a risk factor calculation module, a dynamic constraint and weight generation module, and a hierarchical optimization control layer. The multi-source sensing unit is used to collect multi-source physical field signals, and the actuator unit is used to control the mill speed, feeding, and water supply. The edge computing and state reconstruction layer is used to perform angle synchronization and spatial mapping of the multi-source physical field signals. The monitoring and dynamic constraint generation layer is used to assess the mill's operating risks and generate dynamic safety constraints. The multi-source state detection module is used to reflect the internal operating conditions of the mill. The risk factor calculation module is used to calculate continuous no-load risk factors. The dynamic constraint and weight generation module establishes a mapping function between no-load risk factors and the control feasible region. The hierarchical optimization control layer is used to generate comprehensive performance evaluation results.

[0014] Compared with the prior art, the present invention can achieve the following beneficial effects: This invention utilizes multi-dimensional physical field sensing technology to achieve visualization and mechanistic control of the spatial motion state of the grinding media inside the mill, significantly improving the transparency of the internal operating state. Employing a dynamic constraint contraction mechanism, it can smoothly and gradually release overload conditions, effectively preventing severe mill blockage accidents. Through hierarchical collaborative optimization of Extreme Value Search Control (ESC) and Model Predictive Control (MPC), it achieves online adaptive search and stable tracking of the optimal operating point, ensuring efficient and stable operation of the mill under complex conditions. While ensuring the inherent safety of the mill, this invention fully taps into the mill's production capacity potential, reduces unit energy consumption, and improves both economic and safety benefits. Attached Figure Description

[0015] Figure 1 This is a layout diagram of a safety optimization control system provided according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the active decoupling mechanism provided in an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and do not constitute a limitation thereof. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the invention. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, some operations related to the invention are not shown or described in the specification. This is to avoid obscuring the core parts of the invention with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0017] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined to form various implementations. Furthermore, the order of the steps or actions in the method description can be changed or adjusted in a manner readily apparent to those skilled in the art. Therefore, the various orders in the specification and drawings are merely for the clear description of a particular embodiment and do not imply a mandatory order, unless otherwise stated that a particular order must be followed.

[0018] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on this invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0019] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0020] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] This invention provides a method for optimizing and controlling the safety of mill operation, as detailed below: S1: Collects multi-source physical field signals during the operation of the mill. The multi-source physical field signals include: mill cylinder vibration signal, acoustic emission signal, mill speed signal, mill angle signal, motor current signal, and motor power signal.

[0022] During operation, mineral grinding equipment (hereinafter referred to as a mill) generates various physical field signals. These signals are related to the mill's operating state and can be used to provide feedback on its status. Multi-source sensors installed on the mill cylinder, bearing housing, and transmission system are used to collect these physical field signals related to mill operation in real time. The collected signals include mill cylinder vibration signals, acoustic emission signals, mill speed signals, mill rotation angle signals, motor current signals, and motor power signals. The collected signals are transmitted via multi-source sensing units to the edge computing and state reconstruction layer, providing data support for subsequent processing stages.

[0023] Among the multi-source physical field signals, the mill cylinder vibration signal can reflect the intensity of internal impact and collision and the mill load status; the acoustic emission signal can characterize the internal deformation and fracture of the ore and the wear condition of internal mill components; the mill speed signal and mill angle signal directly reflect the mill's operating speed and position, and can also provide a benchmark for synchronous analysis of multi-source signals; the motor current signal and motor power signal reflect the motor load and indirectly indicate the mill's operating efficiency. These multi-source physical field signals can be collected in real time using multi-source sensors installed at different parts of the mill, providing a basis for understanding the mill's operating status.

[0024] S2: Preprocess the multi-source physical field signal to obtain the preprocessed signal; extract the time-frequency features of the preprocessed signal, and calculate the spatial motion state features corresponding to the time-frequency features. The spatial motion state features include the toe angle and shoulder angle of the grinding medium.

[0025] The multi-source physical field signals obtained directly are diverse, and preprocessing is required before analysis. Preprocessing includes bandpass filtering and normalization of the multi-source physical field signals, angle synchronization processing of the mill cylinder vibration signal and acoustic emission signal using mill speed signal and mill angle signal, and mapping the time domain signal to the mill rotation angle domain.

[0026] The bandpass filtering and normalization processes are as follows: The embodiments of the present invention use Indicates the sequence number of the multi-source sensor, then the first... The raw discrete-time signals from multiple source sensors (the signals collected by each sensor are synchronized with a unified timestamp to form a discrete-time sequence) are represented as follows: , The expression for the preprocessed signal is: , in, Indicates the first Original signal of the road, This indicates the preprocessed signal. Indicates the filtering operator. This represents the mean of the corresponding signal. This represents the standard deviation of the corresponding signal. Bandpass filtering and normalization are performed on multi-source physical field signals to eliminate noise and dimensional differences.

[0027] During mill operation, the movement and state changes of its various components are closely related to the mill's rotation angle. The mill rotation angle signal accurately reflects the angular position of the mill's rotation, while the mill cylinder vibration signal and acoustic emission signal are physical quantity changes recorded in a time series during mill operation. The mill rotation angle signal is used to establish a mapping relationship from the time axis to the angle axis, serving as the reference for resampling vibration and acoustic emission signals into the angle domain, and is a detection signal used to acquire the cylinder rotation angle. Using the mill rotation angle signal as a reference, the mill cylinder vibration and acoustic emission signals, originally recorded in a time series, can be rearranged and correlated according to the mill rotation angle, achieving synchronization between multi-source physical field signals (mill cylinder vibration and acoustic emission signals) and the mill rotation angle. The mill rotation angle signal is decoded to obtain the cylinder rotation angle, which is then introduced into the mill rotation angle analysis. This constitutes the preprocessing signal and the cylinder rotation angle pair The angle synchronization processing method is as follows: introduce the cylinder rotation angle. By using the mill rotation angle signal to establish a mapping relationship between time and mill rotation angle, the preprocessed mill cylinder vibration signal and acoustic emission signal are resampled to the rotation angle domain to obtain the angle domain signal corresponding to the mill's spatial position.

[0028] In this embodiment of the invention, a key phase sensor (or photoelectric encoder) is installed on the pinion shaft or cylinder flange of the mill to obtain the real-time phase angle of the mill. The time-domain vibration signal is segmented according to the mill rotation period and mapped to an angle-domain signal. Subsequently, a short-time Fourier transform or continuous wavelet transform is performed on the angle-domain signal to construct a time-frequency feature spectrum based on polar coordinate representation.

[0029] Short-time Fourier transform was applied to the synchronized mill cylinder vibration signal and acoustic emission signal to obtain the time-frequency characteristics. The expression for the time-frequency characteristics is as follows: , in, Represents the angular frequency variable. Represents a time variable. Indicates the first The time-frequency characteristic value of the signal in the time-frequency domain. Indicates the input signal channel number. This represents the time variable in the summation process. Represents the natural constant. Represents the window function. It represents the imaginary unit.

[0030] Based on the aforementioned time-frequency characteristics, the corresponding spatial motion state characteristics are calculated. These characteristics include the toe angle and shoulder angle of the grinding medium. Specifically: Define an impact energy density function and calculate the energy density in the time-frequency characteristic spectrum. The impact energy density function is expressed as: , in, Indicates the mill rotation angle. This represents the impact energy density corresponding to the mill rotation angle. The lower angular frequency of the mill rotation is... The time-frequency characteristic value, This indicates the lower limit of the frequency integration interval. This indicates the upper limit of the frequency integration interval. This represents the integral.

[0031] Preset threshold When the energy density first exceeds the threshold At that time, the corresponding rotation angle is recorded as the toe angle, representing the position of the first impact of the abrasive medium. Threshold It is not a fixed constant; it can be initialized based on the structural parameters of the grinding equipment, the media filling rate, the liner type, and the ore properties. In this embodiment of the invention, the threshold... Offline calibration is performed using historical operating data, and the threshold value is determined based on the statistical distribution characteristics of vibration and acoustic signals under different operating conditions; in another implementation, the threshold value is... During system operation, online adaptive updates are performed based on feature statistics within the sliding time window to compensate for the effects of ore hardness variations, liner wear, and sensor drift on signal characteristics. Toe angle The expression is: , At the toe angle To the maximum effective turning angle The angle at which the energy density reaches its maximum value is defined as the shoulder angle. The expression is: , , in, This represents the impact energy density function.

[0032] S3: Establish the mill's operating condition feature vector based on the spatial motion state characteristics, and establish the mill's operating condition identification model based on the operating condition feature vector. The operating condition identification model is used to obtain the mill's operating status.

[0033] Based on the calculated spatial motion state characteristics, a working condition feature vector for the mill is established. The expression is: ; in, This indicates the average power of the mill. This represents a load characterization metric. This represents the rate of change of power.

[0034] A condition identification model for the mill is established based on the condition feature vector. The condition identification model is represented as follows: , in, This indicates the result of the working condition identification. This represents the identification model used to output the operating status of the mill.

[0035] In this embodiment of the invention, the operating condition identification model can be constructed using a multi-classification or regression model based on supervised learning. It takes operating condition feature vectors as input and outputs an evaluation result representing the mill's operating status under different operating conditions. The operating condition identification model can be a support vector machine (SVM), random forest, convolutional neural network (CNN), long short-term memory network (LSTM), or a combination thereof. It is trained offline using historical production operation data and updated online or its parameters are fine-tuned based on newly added operating condition samples during actual operation.

[0036] In another embodiment of the present invention, the operating condition identification model may also adopt a rule-based discrimination model, which sets multi-level threshold rules for toe angle, shoulder angle, high-frequency energy ratio, power change rate and load characterization indicators to achieve comprehensive evaluation of operating states such as underload, normal load, overload and abnormal impact.

[0037] The operating condition identification model outputs the evaluation results of the mill's operating status under different operating conditions. This evaluation result is presented as an operating condition confidence vector, which characterizes the mill's relative membership or probability distribution under various operating conditions. The operating condition confidence vector is not directly used to trigger discrete operating mode switching, but rather as input information for continuous adjustment of dynamic safety constraint parameters, optimization target weights, or control gains during subsequent control processes. When multiple operating conditions simultaneously have high confidence levels, the system comprehensively adjusts the control constraint boundaries or optimization preference parameters based on the confidence distribution, rather than making a hard determination of a single operating condition. This ensures smooth and adaptive optimization control of the mill under a single continuous operating mode.

[0038] S4: Generate dynamic safety constraints using the characteristic vector of the working condition and the multivariable derivative relationship.

[0039] Based on the operating status output in S3, potential risks during mill operation are continuously assessed. Specifically, by analyzing power variation trends, load variation trends, and their interrelationships, the risk states of overload, mill blockage, or ineffective grinding are identified. Simultaneously, considering the spatial impact characteristics of the grinding media, the risk level of direct impact of the grinding media on the liners is assessed. Based on the above risk assessment results, continuously changing operating risk factors are generated, and the corresponding safety constraint boundaries in the predictive control of the operating condition identification model are dynamically adjusted accordingly, including but not limited to: the upper limit of the feed rate, the safe range of mill speed, and the adjustment range of the water supply.

[0040] In S3, based on the operating condition feature vector, a continuous overload risk factor is constructed to characterize the operating risk of the mill, and dynamic safety constraints are generated accordingly. This invention embodiment uses... To indicate the mill load, use The power of the mill motor is represented by the power-load coupling change index, i.e., the multivariable derivative relationship, which is expressed as: , , in, This represents the derivative of the rate of change of mill power with respect to load. This index represents the derivative relationship between the mill power change rate and the load change rate. When there exists... If the time is right, it indicates that the mill is in a potential overload or clogging trend.

[0041] Based on the above physical criteria, a continuous overload risk factor is constructed. : , in, Represents the normalized mapping function. This represents the derivative of the mill load change rate with respect to time, used to map the level of risk to a preset range.

[0042] At the same time, based on the toe angle The expression is used to construct the medium shock risk factor: , in, This indicates the preset safe threshold for shoulder angle. This indicates the preset safe threshold for the toe angle.

[0043] By combining continuous overload risk factors and medium impact risk factors, a comprehensive operational risk factor is constructed: , in, All of these represent weighting coefficients.

[0044] Based on comprehensive operational risk factors, a dynamic safety constraint parameter set is generated. This is used to adjust the control variable constraint boundaries in subsequent operating condition identification model predictive control. The mathematical representation of dynamic safety constraints is: , in, Indicates control variables, Indicates comprehensive risk factors The lower boundary of the lower control variable. Indicates comprehensive risk factors The upper boundary of the lower control variable.

[0045] S5: Based on the mill's output, unit energy consumption, and comprehensive performance indicators, the optimal operating point is planned in real time; control commands are generated in real time using the mill's operating status and optimal operating point; when dynamic safety constraints are met, the operating status of the mill is controlled using the control commands to make the mill's operating status close to the optimal operating point.

[0046] At slower timescales, based on feedback from mill output, unit energy consumption, and overall performance indicators during mill operation, a steady-state optimization control method is employed to search for and update the mill's optimal operating point online. This method does not rely on pre-set fixed target values ​​but continuously adjusts the optimal operating point based on real-time operational feedback to adapt to changes in the properties of the grinding media. The updated optimal operating point serves as the input for dynamic adjustment control.

[0047] Among them, the function corresponding to the comprehensive performance index for: , in, This indicates the amount of grinding media processed or its equivalent characterization quantity. This indicates the mill's power consumption. All of these represent weighting coefficients used to balance output and energy consumption. This represents the mapping function for the comprehensive performance index. This represents the comprehensive performance index function value. Steady-state optimization control methods achieve optimal operating points. Apply small perturbation And observe the comprehensive performance index function. Based on the changing trend, estimate the performance gradient under the current operating conditions: , in, This represents the performance gradient of the comprehensive performance index function relative to the running point. Indicates the current running point. This represents a small disturbance applied to the operating point. When the overall performance index is detected to reach an extreme value as the optimal operating point changes, i.e. When the time is right, the corresponding operating point is determined as the optimal operating point under the current operating conditions. .

[0048] Based on the obtained optimal running point By combining the operating condition identification model to obtain the mill's operating status, a dynamic adjustment and control problem is constructed to coordinately optimize and adjust the mill's feed rate, rotational speed, and water supply. The operating status of the mill is indicated by... To represent control variables, use This indicates the optimal operating point. Within each control cycle, the controller aims to minimize the deviation between the operating state and the optimal operating point while suppressing drastic changes in the control input, thus solving for the optimal control command. The controller employs a rolling optimization method, updating the control command within each control cycle to ensure the mill smoothly approaches the optimal operating point while satisfying dynamic safety constraints. This enables continuous and stable closed-loop control.

[0049] In this embodiment of the invention, an actuator is provided. The optimal control command obtained from the solution is sent to the actuator, which then executes the control action via a frequency converter, electric valve, or feeding device. The mill operation process resulting from the execution is then used to acquire the multi-source physical field signals generated by the process in step S1, forming a complete closed-loop control.

[0050] Dynamic safety constraint parameters are input to the controller (MPC controller) as control constraints to achieve risk mitigation and safe operation of the mill in a single continuous operation mode.

[0051] like Figure 1 As shown, to achieve the above-mentioned safe optimization control method for mill operation, this embodiment of the invention also provides a safe optimization control system for mill operation, including a multi-source sensing unit, an actuator unit, an edge computing and state reconstruction layer, a monitoring and dynamic constraint generation layer, a multi-source state detection module, a risk factor calculation module, a dynamic constraint and weight generation module, and a hierarchical optimization control layer. The multi-source sensing unit is used to collect multi-source physical field signals during mill operation. The multi-source sensing unit includes an accelerometer, an acoustic emission sensor, an electrical power or current sensor, and a phase encoding module. During mill operation, the accelerometer, arranged on the outer wall of the mill cylinder, collects the mill cylinder vibration signal to characterize the motion state of the media in the ball mill; the acoustic emission sensor collects the acoustic emission signal generated by the impact of the grinding media to characterize the impact intensity of the ball mill; the electrical power or current sensor collects the motor current signal and motor power signal of the drive system to characterize the overall load state of the mill. The phase encoding module is installed on the mill pinion or end cover to collect the mill speed signal and mill angle signal in real time, providing a spatial reference for polar coordinate mapping.

[0052] The actuator unit is used to control the mill speed, feed (grinding media), and water supply. The actuator unit includes a frequency converter driver, a feeder frequency converter, and an electric water valve. The frequency converter driver is used to control the mill speed, the feeder frequency converter is used to control the feeder operating speed, the feeder includes a feeder belt scale, which is used to obtain the material flow rate entering the grinding equipment, and the electric water valve is used to control the amount of water entering the mill.

[0053] The edge computing and state reconstruction layer is used for angle synchronization and spatial mapping of multi-source physical field signals to reconstruct the spatial motion state characteristics of the grinding media. It includes a polar coordinate mapping module, a geometric feature calculation module, and a state vector construction module. Specifically, the polar coordinate mapping module receives mill cylinder vibration signals, acoustic emission signals, mill speed signals, mill rotation angle signals, motor current signals, and motor power signals from the multi-source sensing unit. It uses a synchronization algorithm to eliminate transmission delay and then a coordinate transformation algorithm to map the time-domain signals to the mill rotation angle coordinate system. The geometric feature calculation module constructs a polar coordinate map reflecting the distribution of physical impact energy of the grinding media inside the mill. Based on the polar coordinate map, it identifies energy density abrupt change points and calculates the toe angle and shoulder angle in real time. The toe angle and shoulder angle represent the key geometric features of the media's motion trajectory.

[0054] The monitoring and dynamic constraint generation layer includes a multi-source state detection module, a risk factor calculation module, and a dynamic constraint and weight generation module. Based on the spatial motion state characteristics and operating state change trends, it continuously assesses the mill's operating risks and generates dynamic safety constraints for optimized control. Specifically: The multi-source state detection module is used to comprehensively analyze multi-source state information during mill operation, forming multi-source state detection results reflecting the internal operating conditions of the mill. Multi-source state information includes state features extracted from multi-source physical field signals, as well as operating condition variables. The input to the multi-source state detection module includes at least the spatial motion state characteristic parameters of the grinding media output by the state reconstruction layer, and operating condition variables such as mill power, speed, and feed rate, along with their changing trends. The module achieves consistency detection, load-power coupling state detection, and operating state change trend detection through joint analysis of multi-variable information. Specifically, by analyzing the relative relationship and changing trend of the toe angle and shoulder angle, it determines whether the grinding media is in the effective grinding range; by jointly analyzing the relationship between the mill power change rate and the load change rate, it identifies the risks of overload, bulging, or mill blockage; and by analyzing the changing trends of state variables within a continuous time window, it distinguishes between short-term disturbances and continuous abnormal operating conditions. The output of the multi-source state detection module includes the detection results of the current operating state of the mill, the spatial motion state characteristics, and the corresponding state confidence information. The output results serve as the input information for the dynamic constraint and weight generation module to calculate risk factors and generate dynamic safety constraints, while also providing state criterion support for hierarchical optimization control.

[0055] The risk factor calculation module includes an overload calculation unit and an no-load calculation unit. The overload calculation is based on the derivative relationship between the power change rate and the load change rate to calculate the continuous overload risk factor. The no-load risk calculation unit calculates the continuous no-load risk factor (i.e., the impact risk factor) based on the deviation between the measured toe angle and the preset safe area (preset toe angle safety threshold range).

[0056] The dynamic constraint and weight generation module is used to establish a mapping function between the no-load risk factor (i.e., the impact risk factor) and the controllable feasible region. When the comprehensive risk factor (composed of a weighted average of the continuous overload risk factor and the no-load risk factor) increases, the constraint boundary of the controller is contracted in real time, thereby mathematically forcing the controller to generate load reduction or risk avoidance commands. The risk factor is a dimensionless index calculated by normalization and a continuous mapping function based on the degree of deviation of the spatial motion state characteristics from the preset safe range. The penalty weights in the objective function are then dynamically optimized and adjusted according to the risk level. For example, when an impact risk is detected, the penalty weight for "excessive" risk is increased exponentially.

[0057] The hierarchical optimization control layer includes a steady-state optimization module, a performance evaluation module, a working condition detection module, and a dynamic adjustment module. The steady-state optimization module employs an extreme value search control algorithm (ESC), injecting low-frequency detection signals into the system and determining the optimal power setpoint based on energy efficiency feedback under the current ore properties. Figure 2 As shown, to prevent the probe disturbance signal injected by the upper-level steady-state optimization module from being misjudged as an abnormal operating condition by the lower-level monitoring layer, this embodiment of the invention designs an active decoupling mechanism based on a synchronization identifier: When the steady-state optimization module enters the "apply disturbance" state (e.g., the sinusoidal micro-perturbation injection stage), it synchronously generates a high-level synchronization identifier (ID=1) and sends it to the edge computing and state reconstruction layer along with the control command. Identifier detection: The operating condition detection module checks the synchronization identifier in real time. Branch processing: Normal processing path (ID=0): If the synchronization identifier is low (steady-state control period), the monitoring layer executes normal feature extraction logic to maintain high sensitivity. Decoupling activation path (ID=1): If the synchronization identifier is high, the monitoring layer activates the decoupling filter. This filter sets a notch filter or band-stop filter for the disturbance frequency of ESC to filter out the signal components in this frequency band; or temporarily widens the dead time of fault judgment, thereby ignoring the controllable fluctuations caused by optimization probes. Effect: Through this logic, the system ensures optimization capabilities while avoiding unnecessary mode switching or shutdown protection triggered by "false signals." The performance evaluation module performs real-time or statistical analysis on output, energy consumption, and safety-related indicators during system operation, determines whether the current operating state meets the steady-state performance evaluation conditions, and generates a comprehensive performance evaluation result. The performance evaluation result guides the setpoint update of the steady-state optimization unit and coordinates the dynamic constraint adjustment process, thereby ensuring that steady-state optimization and dynamic adjustment are carried out collaboratively under safe and stable conditions. The dynamic adjustment module adopts the linear model predictive control (MPC) algorithm. It receives the optimal setpoint from the upper layer, dynamically performs time-domain optimization within the real-time safety constraint monitoring range planned by the "monitoring layer," calculates the optimal control sequence, and sends it to the actuator units in the physical layer.

[0058] In this embodiment of the invention, the modules interact with each other through a clearly defined data flow sequence. The multi-source physical field signals collected by the multi-source sensing unit at the physical layer are first transmitted to the edge computing and state reconstruction layer. After angle synchronization and feature extraction, a state vector representing the spatial motion state of the grinding medium is formed. The operating state feature vector is input to the dynamic constraint and weight generation module for operational risk assessment and dynamic safety constraint generation, and also serves as the system state input to the hierarchical optimization control layer. The risk factors and constraint boundary parameters generated by the dynamic constraint and weight generation module are transmitted to the dynamic adjustment controller as constraint inputs for online adjustment of the feasible solution space in model predictive control. The optimal operating point generated by the steady-state optimization controller based on the performance evaluation results is transmitted to the dynamic adjustment controller as the target input. The dynamic adjustment controller calculates the optimal control command by integrating the state input, target input, and constraint input, and sends the control command to the actuator. The execution result is fed back into the multi-source sensing unit via feedback signals, forming a closed-loop control. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

[0059] The specific embodiments of the present invention described above do not constitute a limitation on the scope of protection of the present invention. Any other corresponding changes and modifications made in accordance with the technical concept of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing and controlling the safety of mill operation, characterized in that, include: S1: Collect multi-source physical field signals during the operation of the mill. The multi-source physical field signals include: mill cylinder vibration signal, acoustic emission signal, mill speed signal, mill angle signal, motor current signal, and motor power signal. S2: Preprocess the multi-source physical field signal to obtain a preprocessed signal; perform angle synchronization processing on the preprocessed signal to obtain an angle domain signal; extract the time-frequency features of the angle domain signal, and calculate the spatial motion state features corresponding to the time-frequency features, wherein the spatial motion state features include the toe angle and shoulder angle of the grinding medium; S3: Based on the aforementioned spatial motion state characteristics, establish the mill's operating condition feature vector, the expression of which is: ； in, This indicates the average power of the mill. This represents a load characterization metric. Indicates the rate of change of power. Indicates the angle of the toe. Indicates the shoulder angle; A mill condition identification model is established based on the aforementioned operating condition feature vector; the operating condition identification model is used to obtain the operating status of the mill. S4: Generate dynamic safety constraints using the aforementioned working condition feature vectors and multivariable derivative relationships; The multivariable derivative relationship is expressed as follows: ； The dynamic security constraint is expressed as: ； in, This represents the derivative of the rate of change of mill power with respect to load. An index representing the derivative relationship between the mill power change rate and the load change rate. Indicates control variables, Indicates comprehensive risk factors The lower boundary of the lower control variable. Indicates comprehensive risk factors The upper boundary of the lower control variable; In step S4, the dynamic safety constraints are generated using a comprehensive operational risk factor, which is expressed as follows: ； in, All represent weighting coefficients. Indicates continuous overload risk factor, Indicates the risk factor of media impact; S5: Based on the mill's output, unit energy consumption, and comprehensive performance indicators, the optimal operating point is planned in real time; control commands are generated in real time using the mill's operating status and the optimal operating point; when the dynamic safety constraints are met, the operating status of the mill is controlled using the control commands to make the mill's operating status closer to the optimal operating point; the multi-source physical field signals generated by the mill's operation are collected according to the process in S1 to form a closed-loop control.

2. The method for optimizing and controlling the safety of mill operation according to claim 1, characterized in that, The preprocessing includes bandpass filtering and normalization. The expression for the preprocessed signal is: ； in, Indicates the first Original signal of the road, This indicates the preprocessed signal. Indicates the filtering operator. This represents the mean of the corresponding signal. This represents the standard deviation of the corresponding signal.

3. The method for optimizing and controlling the safety of mill operation according to claim 1, characterized in that, The angle synchronization processing method is as follows: introducing the cylinder rotation angle. The mill rotation angle signal is used to establish a mapping relationship between time and rotation angle. The preprocessed time series signal is resampled to the rotation angle domain to obtain the angle domain signal corresponding to the spatial position of the mill.

4. The method for optimizing and controlling the safety of mill operation according to claim 1, characterized in that, The expression for the time-frequency feature is: ； in, Represents the angular frequency variable. Represents a time variable. Indicates the first The time-frequency characteristic value of the signal in the time-frequency domain. Indicates the input signal channel number. This represents the time variable in the summation process. Represents the natural constant. Represents the window function. It represents the imaginary unit.

5. The method for optimizing and controlling the safety of mill operation according to claim 1, characterized in that, The toe angle and shoulder angle The expressions are as follows: ； ， ； in, Represents the impact energy density function. Indicates the threshold. This indicates the maximum effective turning angle.

6. The method for optimizing and controlling the safety of mill operation according to claim 1, characterized in that, Based on the continuous overload risk factor and the medium shock risk factor, the comprehensive operational risk factor is constructed; the expressions for the continuous overload risk factor and the medium shock risk factor are as follows: ； ； in, Represents the normalized mapping function. This indicates the preset safe threshold for shoulder angle. An index representing the derivative relationship between the mill power change rate and the load change rate. This represents the derivative of the mill load change rate with respect to time.

7. A safety optimization control system for mill operation, used to implement the safety optimization control method for mill operation as described in any one of claims 1-6, characterized in that, The control system includes a multi-source sensing unit, an actuator unit, an edge computing and state reconstruction layer, a monitoring and dynamic constraint generation layer, a multi-source state detection module, a risk factor calculation module, a dynamic constraint and weight generation module, and a hierarchical optimization control layer; The multi-source sensing unit is used to collect the multi-source physical field signals, the actuator unit is used to control the mill speed, feeding, and water supply; the edge computing and state reconstruction layer is used to perform angle synchronization and spatial mapping on the multi-source physical field signals, and the monitoring and dynamic constraint generation layer is used to assess the mill operation risk and generate dynamic safety constraints. The multi-source condition detection module is used to reflect the internal operating conditions of the mill. The risk factor calculation module is used to calculate continuous unloaded risk factors; The dynamic constraint and weight generation module establishes a mapping function between the unloaded risk factor and the controllable feasible region. The hierarchical optimization control layer is used to generate comprehensive performance evaluation results.

Images