Intelligent mill based on data driving and control method thereof

Abstract

The application relates to a data-driven intelligent mill and a control method thereof, and belongs to the field of mine intelligent equipment. The intelligent mill comprises an intelligent sensing module, which is used for collecting key parameters of a grinding process in real time; an intelligent analysis and decision module, which is used for receiving data collected by the intelligent sensing module, generating a mill operation parameter optimization instruction through an intelligent decision algorithm; and an autonomous precise execution module, which is used for receiving and precisely executing the mill operation parameter optimization instruction, and optimizing and controlling variables in the grinding process. The intelligent sensing module, the intelligent analysis and decision module and the autonomous precise execution module form a closed-loop control loop of "sensing-decision-execution-re-sensing". The intelligent mill can greatly reduce human factor interference, improve grinding efficiency and reduce operation cost.

Description

Technical Field

[0001] This invention belongs to the field of intelligent mining equipment, specifically relating to a data-driven intelligent mill and its control method. Background Technology

[0002] Grinding is a core step in the mineral processing flow, accounting for 40% to 60% of the total energy consumption of a mineral processing plant. Furthermore, the particle size and quality of the grinding products directly affect the performance of subsequent processes such as flotation and magnetic separation. With the increasing depletion, refinement, and complexity of mineral resources, higher demands are being placed on the precision and optimization of the grinding process.

[0003] Traditional mill control relies primarily on operator experience for parameter adjustments, which presents several significant problems: It only uses basic sensors for temperature, pressure, and flow; equipment operating status and key process indicators (such as feed particle size and material filling rate) are mainly determined manually, leading to high labor intensity, safety hazards, and the inability to monitor online in real time. Furthermore, manual judgment is inherently delayed, making it difficult to respond promptly to fluctuations in ore quantity and changes in ore properties, resulting in significant fluctuations in grinding production, high energy consumption, and impacting the overall production indicators and economic benefits of the concentrator.

[0004] To address these issues, the industry has proposed a variety of intelligent grinding control solutions.

[0005] For example, Chinese patent application CN121050360A discloses a mill collaborative optimization control method and system based on digital twins. It constructs a digital twin model of the target mill (including a three-dimensional mechanical structure model and a physical process model), combines this with multi-source operating data collected in real time by a sensor array to simulate the mill's operating state, and uses a particle swarm optimization algorithm to solve a multi-objective optimization mathematical model to generate a mill control strategy. However, this approach focuses on constructing a digital twin through three-dimensional modeling for state simulation. Its core lies in the virtual mapping between geometric structure and physical processes, but it does not establish an energy consumption mechanism model, making it difficult to achieve precise collaborative optimization of energy consumption and particle size during the grinding process.

[0006] Chinese patent application CN121386617A discloses a digital twin monitoring and prediction method and system for ball mills, focusing on the specific monitoring of "bloating" risk. It constructs a multi-physics coupled digital twin model (integrating mechanical dynamics, CFD-DEM simulation, and LSTM neural network correction), calculates a bloating risk level index using a multi-modal decision fusion algorithm, and utilizes reinforcement learning to simulate and optimize within the digital twin model when the risk exceeds limits, outputting coordinated control commands. While this solution achieves real-time simulation and risk warning of the mill's internal state, its core focus is on fault prevention and risk control. It lacks an independent energy consumption constraint model and does not treat energy minimization as an optimization objective alongside product quality for multi-objective coordinated optimization, making it difficult to meet the comprehensive needs of mineral processing plants for "quality improvement, efficiency enhancement, energy saving, and consumption reduction" in the grinding process.

[0007] Chinese patent application CN121338884A discloses a grinding efficiency optimization method and system based on ore quantity measurement. It uses a high-precision ore quantity measurement device (such as a nuclear belt scale) and multiple sensors to collect data on ore quantity, ore hardness, and particle size in real time. Based on machine learning algorithms (such as random forests), it constructs a grinding efficiency optimization model to optimize parameters such as mill speed and water supply. While this scheme achieves data-driven parameter optimization, it employs a purely data-driven "black box" modeling method, lacking deep coupling with the grinding process mechanism. When ore properties change drastically or exceed the historical data distribution range, the model's generalization ability and prediction accuracy are difficult to guarantee.

[0008] In summary, existing technical solutions either focus on digital twin virtual mapping, single risk warning, or pure data-driven optimization, and have not yet formed a comprehensive intelligent mill solution that can sense the mill's operating status, deeply integrate mechanism models and data models, achieve granular control and energy consumption minimization through collaborative optimization, and coordinate with upstream and downstream equipment. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides a data-driven intelligent mill and its control method, which enables the perception of the grinding equipment status, process, and key parameters. Through intelligent decision-making driven by both data and models, the operating parameters are optimized. Combined with precise execution and upstream and downstream collaborative control, the goal of stable particle size, minimum energy consumption, and high efficiency throughout the entire grinding process is ultimately achieved.

[0010] To achieve the above objectives, the technical solution adopted by this invention is: a data-driven intelligent mill, comprising: The intelligent sensing module is used to collect key parameters of the grinding process in real time; wherein, the key parameters of the grinding process include one or more of the following: mill power, rotation speed, feed rate, feed particle size, filling rate, water replenishment, pump frequency, and hydrocyclone overflow particle size; The intelligent analysis and decision-making module is used to receive the data collected by the intelligent sensing module, and based on the energy consumption constraint model, generate an optimized scheme for mill operation parameters under preset constraints by using an intelligent decision-making algorithm with the dual objectives of product granularity control and energy consumption minimization. The autonomous precision execution module is connected to the intelligent analysis and decision-making module. It is used to receive and precisely execute the mill operation parameter optimization instructions, and to optimize and control at least one variable among water volume, ore volume and rotation speed in the grinding process to ensure that the grinding process is stable under optimal conditions. The intelligent sensing module, intelligent analysis and decision-making module, and autonomous precision execution module form a closed-loop control circuit of "perception-decision-execution-re-perception".

[0011] Furthermore, the method for constructing the energy consumption constraint model includes: (1) Dataset establishment: Collect data on different ore hardness and feed particle size F. 80 The process parameters and energy consumption data under the condition of liner wear are used to establish a dataset of process parameter-energy consumption mapping relationship; the process parameters include at least feed rate, mill speed and grinding concentration. (2) Energy consumption calculation and prediction: Based on Bond crushing theory, a unit energy consumption calculation formula is constructed. Combined with the process parameter-energy consumption mapping relationship dataset, the predicted energy consumption under different working conditions is calculated to obtain the energy consumption constraint model.

[0012] Furthermore, the formula for calculating the unit energy consumption is as follows: Where K is the correction factor, Wi is the Bond power index, and P 80 For product particle size, F 80 This refers to the feed particle size.

[0013] Furthermore, in the method for constructing the energy consumption constraint model, the process parameter-energy consumption mapping relationship dataset is divided into a training set and a test set. The model is trained using the training set, and the model's prediction accuracy is verified using the test set.

[0014] This invention also proposes a mill control method, wherein the mill is the data-driven intelligent mill described above, comprising the following steps: S1 collects key parameters of the grinding process in real time, and after preprocessing, uploads them to the intelligent analysis and decision-making module in real time through the industrial communication network. S2, the intelligent analysis and decision-making module, is based on the energy consumption constraint model and uses intelligent decision-making algorithms to analyze, calculate and predict the collected data, and generate mill operation parameter optimization instructions; S3, the autonomous precision execution module receives and executes the mill operation parameter optimization instruction, and optimizes and controls at least one variable among the water supply, feed rate, and mill speed in the grinding process through the execution mechanism, so that the grinding equipment and process are in the best state; The execution effect of the S4 autonomous precision execution module is reflected in the operating data of the mill equipment and process. It is then re-collected by the sensing module and fed back to the intelligent analysis and decision-making module for continuous optimization, forming a closed-loop control of "perception-decision-execution-re-perception".

[0015] The beneficial effects of this invention are as follows: The data-driven intelligent mill and its control method described in this invention include an intelligent sensing module, an intelligent analysis and decision-making module, and an autonomous and precise execution module. It senses key operating parameters of the equipment and process during the milling process, analyzes the operating status of the equipment and process, and provides optimized adjustment variable parameters through intelligent analysis and calculation. This enables autonomous and precise execution of commands, avoiding production fluctuations caused by the lag in manual adjustments, while simultaneously optimizing production indicators and improving grinding production efficiency. Attached Figure Description

[0016] Figure 1 This is a control flowchart of the intelligent mill in an embodiment of the present invention. Detailed Implementation

[0017] The present invention will be further described in detail below with reference to the embodiments, but this should not be construed as limiting the invention in any way.

[0018] This embodiment takes the modification of a Φ3.6×6.84m ball mill in a copper-zinc ore beneficiation plant as an example to illustrate the specific implementation of the data-driven intelligent mill described in this invention.

[0019] The data-driven intelligent mill includes an intelligent sensing module, an intelligent analysis and decision-making module, and an autonomous precision execution module. These modules work together to form a closed-loop control circuit of "perception-decision-execution-re-perception". The specific technical solution is as follows.

[0020] 1. Intelligent Sensing Module The intelligent sensing module can realize online sensing of key parameters in the grinding process, such as mill power, speed, feed rate, feed particle size, filling rate, water replenishment, pump frequency, and hydrocyclone overflow particle size.

[0021] By installing basic sensors such as current, power, pressure, and vibration sensors on the mill body and auxiliary equipment, data such as motor current, power, hydraulic system pressure, and spindle speed are collected in real time. Utilizing process monitoring instruments such as belt scales, electromagnetic flow meters, concentration meters, and online particle size analyzers, process parameters such as feed rate, water flow rate, pump pool level, slurry pump frequency, classifier pressure, slurry concentration, and particle size are acquired in real time. Dedicated intelligent instruments are used to perceive key parameters that are traditionally difficult to detect online, including but not limited to: installing an online ore particle size analyzer on the feed belt to analyze ore particle size distribution in real time; and installing acoustic sensors near the mill cylinder to infer the material filling rate inside the mill in real time through vibration signal fusion analysis.

[0022] 2. Intelligent Analysis and Decision-Making Module The intelligent analysis and decision-making module, as the core intelligent engine, receives data collected by the intelligent sensing module and generates mill operation parameter optimization instructions through an intelligent decision-making algorithm; the intelligent analysis and decision-making module includes: (1) Energy consumption constraint model, constructed based on mill energy consumption mechanism and historical operating data, is used to predict mill energy consumption levels under different operating conditions in real time. The specific construction method is as follows: (1-1) Dataset establishment: Collect data on different ore hardness and feed particle size F of the copper-zinc mine. 80 The process parameters and energy consumption data under the condition of liner wear were used to establish a process parameter-energy consumption mapping dataset. The process parameters include at least feed rate, mill speed and grinding concentration. A total of 106,750 sets of valid data were collected and divided into training set and test set in a 7:3 ratio. The model was trained using the training set and the model prediction accuracy was verified using the test set.

[0023] (1-2) Energy Consumption Calculation and Prediction: Based on Bond's crushing theory, a unit energy consumption calculation formula is established. Combined with the process parameter-energy consumption mapping dataset, the predicted energy consumption values ​​under different operating conditions are calculated to obtain the energy consumption constraint model. The unit energy consumption calculation formula is as follows: Where K is the correction factor, Wi is the Bond work index (tested to be 17.72 kWh / t), and P 80 F represents the particle size at which 80% of the product's mass passes through the sieve. 80 This refers to the particle size at which 80% of the mass of the feed passes through the screen.

[0024] It should be noted that the training and testing methods for the training and testing sets in this invention are as follows: (a) using ore hardness and feed particle size F 80(a) The liner wear, feed rate, rotation speed, and concentration are used as input features, and unit energy consumption is used as the output label. The process parameters-energy consumption mapping relationship is fitted by combining the Bond crushing theory formula, and the core parameters such as the power index Wi are optimized to complete the model training; (b) The energy consumption prediction value is obtained by substituting the input features of the test set into the trained model, and compared with the actual energy consumption. The deviation rate and mean square error are calculated to verify the accuracy; if the threshold is not reached, the deviation data of the test set is sent back to the training set, and the model parameters are iterated and optimized again until the real-time prediction requirements of grinding energy consumption are met.

[0025] The intelligent decision-making algorithm, coupled with the energy consumption constraint model, is used to generate an optimization scheme for mill operating parameters under preset constraints, with the dual objectives of product granularity control and energy consumption minimization.

[0026] The intelligent decision-making algorithm uses the granularity control objective function min F1 = |D 80 - D 80 * The objective function min F2 = E, which minimizes energy consumption, is a dual objective. Under constraints including feed rate, mill power, mill speed, and product particle size, a multivariate optimization scheme is generated, incorporating mill speed, feed rate, and water flow. Where D... 80 For product particle size, D 80 * Let D be the target product particle size (0.074 mm in this embodiment), and E be the unit product energy consumption. Constraints: feed rate Q ∈ [70, 105] t / h, mill power P ≤ 0.9 × Pmax (Pmax = 1300 kW), rotational speed N ∈ [15.6, 19] r / min (critical rotational speed Nc ≈ 22.3 r / min), and product particle size D. 80 ∈[0.06, 0.09]mm.

[0027] 3. Autonomous and precise execution module The autonomous precision execution module is connected to the intelligent analysis and decision-making module. It is used to receive and precisely execute the mill operation parameter optimization instructions, and optimize and control at least one variable among water volume, ore volume and rotation speed in the grinding process to ensure that the grinding process is stable under optimal conditions.

[0028] In this embodiment, precise execution is achieved through the following methods: frequency conversion speed regulation technology is used to control the speed of the belt feeder, and the feed rate set value calculated by the intelligent decision algorithm is used to achieve precise feeding through PID control; a high-voltage frequency converter is used to control the mill speed dynamically according to energy consumption constraints and product particle size requirements; and an electric regulating valve is used to control the water supply, and the water supply is automatically adjusted according to the slurry concentration detection value.

[0029] 4. Closed-loop control and coordination mechanism The intelligent sensing module collects real-time data on the control effect of the autonomous precision execution module (such as product granularity and actual energy consumption) and feeds it back to the intelligent analysis and decision-making module. The intelligent analysis and decision-making module compares the optimization target with the actual effect, performs deviation analysis and parameter correction, and adaptively fine-tunes the model parameters (such as Wi in the energy consumption model) to realize online learning and optimization of the model, forming a closed-loop control of continuous improvement of "perception-decision-execution-re-perception".

[0030] The control process of the data-driven intelligent mill proposed in this invention is as follows: Figure 1 As shown, the specific steps include the following: S1, the intelligent sensing module collects relevant data on the mill equipment and process from the basic sensors of the equipment, the basic instruments of the grinding process section, and the intelligent sensing dedicated instruments, senses the operating status of the equipment and process, and feeds it back to the intelligent analysis and decision-making module; the data is collected and acquired through the equipment PLC, the OPC server of the DCS system, and the intelligent control system server; S2, after the intelligent analysis and decision-making module obtains comprehensive data on the mill equipment and process, it combines the energy consumption constraint model and uses intelligent control algorithms to analyze, calculate and predict the data, calculate the optimal adjustment parameters of the grinding system such as feed rate, water rate, and rotation speed, evaluate the equipment performance based on the above information, and send relevant instructions to the autonomous precision execution module. S3, after receiving instructions from the intelligent analysis and decision-making module, the autonomous precision execution module optimizes and controls variables such as water volume, ore volume, and rotation speed in the grinding process through actuators such as frequency converters, valves, and frequency converter feeders, so that the grinding equipment and process are in the best state. The execution effect of the S4 autonomous precision execution module is reflected in the operating data of the mill equipment and process, which is then sensed by the intelligent sensing module. The sensed data needs to be fed back to the intelligent analysis and decision-making module for continuous optimization, forming a closed-loop control.

[0031] Through the above implementation methods, the present invention successfully realizes the transformation of mill control from isolated single-machine control to intelligent agent control integrated into the process, which significantly improves grinding efficiency, stabilizes product quality and reduces overall energy consumption.

[0032] When applied to the Φ3.6×6.84m ball mill in this copper-zinc ore beneficiation plant, this invention achieved fully automated and intelligent operation of the grinding process. Production capacity increased from 72.95 t / h to 83.97 t / h, an increase of 15.11%; unit grinding power consumption decreased from 19.06 kWh / t to 17.7 kWh / t, a decrease of 7.13%; the overflow particle size qualification rate of the hydrocyclone increased from 76.92% to 96.26%, an increase of 19.34%; and the overflow concentration qualification rate increased from 81.12% to 95.95%, an increase of 18.28%. This significantly reduced the labor intensity of operators and the operating costs of equipment, achieving remarkable economic and social benefits.

[0033] The parts of this invention not described in detail are prior art.

[0034] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the specific implementation of the present invention with reference to the above embodiments. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention are within the protection scope of the pending claims.

Claims

1. A data-driven based intelligent grinding mill, characterized in that, include: The intelligent sensing module is used to collect key parameters of the grinding process in real time; wherein, the key parameters of the grinding process include one or more of the following: mill power, rotation speed, feed rate, feed particle size, filling rate, water replenishment, pump frequency, and hydrocyclone overflow particle size; The intelligent analysis and decision-making module is used to receive the data collected by the intelligent sensing module, and based on the energy consumption constraint model, generate an optimized scheme for mill operation parameters under preset constraints by using an intelligent decision-making algorithm with the dual objectives of product granularity control and energy consumption minimization. The autonomous precision execution module is connected to the intelligent analysis and decision-making module. It is used to receive and precisely execute the mill operation parameter optimization instructions, and to optimize and control at least one variable among water volume, ore volume and rotation speed in the grinding process to ensure that the grinding process is stable under optimal conditions. The intelligent sensing module, intelligent analysis and decision-making module, and autonomous precision execution module form a closed-loop control circuit of perception-decision-execution-re-perception; The method for constructing the energy consumption constraint model includes: (1) Dataset establishment: Collect data on different ore hardness and feed particle size F. 80 The process parameters and energy consumption data under the condition of liner wear are used to establish a dataset of process parameter-energy consumption mapping relationship; the process parameters include at least feed rate, mill speed and grinding concentration. (2) Energy consumption calculation and prediction: Based on Bond crushing theory, a unit energy consumption calculation formula is constructed. Combined with the process parameter-energy consumption mapping relationship dataset, the predicted energy consumption under different working conditions is calculated to obtain the energy consumption constraint model.

2. The data-driven based smart grinding machine of claim 1, wherein, The unit energy consumption calculation formula is: Wherein, K is a correction coefficient, Wi is a Bond work index, P 80 is a product particle size, F 80 is a feed particle size.

3. The data-driven, intelligent grinding mill of claim 1, wherein, In the method for constructing the energy consumption constraint model, the process parameter-energy consumption mapping relationship dataset is divided into a training set and a test set. The model is trained using the training set, and the model's prediction accuracy is verified using the test set.

4. A mill control method, characterized by, The mill is the data-driven intelligent mill according to any one of claims 1-3, comprising the following steps: S1 collects key parameters of the grinding process in real time, and after preprocessing, uploads them to the intelligent analysis and decision-making module in real time through the industrial communication network. S2, the intelligent analysis and decision-making module, is based on the energy consumption constraint model and uses intelligent decision-making algorithms to analyze, calculate and predict the collected data, and generate mill operation parameter optimization instructions; S3, the autonomous precision execution module receives and executes the mill operation parameter optimization instruction, and optimizes and controls at least one variable among the water supply, feed rate, and mill speed in the grinding process through the execution mechanism, so that the grinding equipment and process are in the best state; The execution effect of the S4 autonomous precision execution module is reflected in the operating data of the mill equipment and process. It is then re-collected by the sensing module and fed back to the intelligent analysis and decision-making module for continuous optimization, forming a closed-loop control of perception-decision-execution-re-sensing.

Images