Tower mill based on self-adaptive PID (Proportion Integration Differentiation) control

Abstract

The utility model relates to the technical field of mineral processing, in particular to a tower mill based on self-adaptive PID (Proportion Integration Differentiation) control, which comprises a detection module, a control module, an execution module and a man-machine interaction module, the detection module monitors in real time and transmits data to the control module in real time, traditional detection delay is eliminated, the control module integrates a fuzzy PID algorithm based on a PLC (Programmable Logic Controller), and the man-machine interaction module carries out man-machine interaction. PID parameters are dynamically adjusted, the rotating speed is preset, multivariable collaborative parameters are optimized in combination with a genetic algorithm, the rotating speed, the feeding amount and the filling rate are coordinated, and stability is ensured. The execution module accurately drives the tower mill through a frequency converter and a permanent magnet synchronous motor, and the response accuracy is improved through closed-loop control. The man-machine interaction module displays real-time data through an upper computer, parameter setting and communication configuration are supported, whole-process monitoring and optimization are achieved, and the system remarkably improves discharging stability, reduces energy consumption, prolongs the service life of equipment and adapts to complex working condition requirements through self-adaptive control and multi-module cooperation.

Description

Technical Field

[0001] This utility model relates to the field of mineral processing technology, specifically to a tower mill based on adaptive PID control. Background Technology

[0002] Tower mills are key equipment in mineral processing, using grinding media to grind ore to achieve target particle sizes. Traditional control methods often rely on fixed rotation speeds or empirical adjustments, which struggle to cope with fluctuations in ore hardness and changes in feed rate, resulting in poor particle size stability and high energy consumption. Furthermore, during operation, existing tower mills experience partial agglomeration of the slurry, adhering to the surface and bottom of the mill, increasing feed resistance and causing uneven feeding. Utility Model Content

[0003] To address the problems mentioned in the background section, this invention proposes a tower mill based on adaptive PID control.

[0004] A tower mill based on adaptive PID control includes a tower mill body, a slurry containing coarse particles fed from the bottom of the tower mill, and fine particles ground from the top of the tower mill outputting from the output pipe. It also includes a detection module that outputs an electrical signal to the control module. The detection module includes an online laser particle size analyzer installed at the outlet pipe of the tower mill. The online laser particle size analyzer detects the particle size distribution of the ore in real time, using the D50 value (median particle size) as the core indicator, and simultaneously outputs both analog and digital signals.

[0005] A tower mill based on adaptive PID control includes a detection module, a control module, and an execution module. The detection module includes an online laser particle size analyzer, which is laterally embedded in the center of the side wall of the tower mill's discharge port pipe.

[0006] The detection window is not completely vertical or horizontal, but forms a 30° angle with the axis of the discharge pipe. The tilt direction matches the material flow direction, allowing the laser beam to penetrate the material flow at an angle, thus enhancing the ability to capture dynamic particles.

[0007] It also includes a feed rate sensor, which is installed inside the flange of the feed inlet pipe of the tower mill and is fixed coaxially with the feed pipe;

[0008] It also includes a media filling rate detection module, which includes a vibration sensor and a pressure sensor; the vibration sensor is installed in a ring array on the upper middle part of the outer wall of the tower mill cylinder.

[0009] The detection window of the online laser particle size analyzer is inclined at a 30° angle to the axis of the discharge pipe. The circular array of vibration sensors is distributed at a spacing of at least 1 / 8 of the circumference of the tower mill, completely covering the tower mill, and the sensor orientation is consistent with the tangential direction of the agitator.

[0010] The vibration sensor installed on the outer wall of the tower mill cylinder is covered with a wear-resistant rubber sleeve, and the surface of the wear-resistant rubber sleeve is equipped with heat dissipation fins.

[0011] The pressure sensors are installed at the bottom of the tower mill. The pressure sensors are arranged in a uniform ring around the center of the bearing seat, and the sensing surface of the pressure sensors is parallel to the bottom surface of the filter plate.

[0012] The filter plates are installed at the bottom of the mixing chamber of the tower mill. The filter plates include at least four sector-shaped filter plates with sealing rings between them. The mill also includes a mixing shaft. The sector-shaped filter plates are connected to the mixing shaft by flange bolts. The slurry enters the tower mill from the feed pipe. As the slurry continues to enter, it gradually floats to the top of the mixing chamber for further grinding.

[0013] During the grinding process, some lumps will adhere to the cylinder wall or fall onto the filter plate surface under gravity. The stirring shaft drives the scraper to rotate, stirring the slurry at the bottom while scraping the surface of the filter plate. Stirring the slurry at the bottom causes some of the lumps adhering to the inner wall of the cylinder to fall off. When the scraper scrapes the surface of the filter plate, it rubs and squeezes some of the lumps, breaking them down. At the same time, it scrapes the through holes to prevent them from becoming blocked. The shape of the bottom of the scraper needs to be selected according to the specific composition of the slurry.

[0014] During cleaning, the filter plate divides the cylinder into two chambers, and a cleaning nozzle is installed at the bottom. The cleaning nozzle can directly rinse the bottom of the filter plate and the bottom of the chamber. At the same time, the filter plate intercepts some of the clumps, reducing the adhesion of the grinding balls to the surface.

[0015] When the grinding balls need to be replaced or repaired, the split-type filter plates are easy to disassemble and assemble, and some fan-shaped filter plates that are severely worn can be replaced individually.

[0016] To ensure that most of the mortar can pass through the through holes, the diameter of the through holes should be 4 to 10 times the maximum particle size to avoid clogging and poor flow due to excessively small hole diameter.

[0017] The fan-shaped filter plates are tightly connected by sealing rings and sealing grooves, and the flange bolts ensure the overall structure is airtight, preventing liquid or waste leakage during cleaning and ensuring stable system operation.

[0018] The filter plate has a central mounting hole with a sealing groove around its edge. A single-frequency vibration motor is installed at the flange connection of the filter plate. The filter plate surface has tapered through-holes; this tapered design combines gravity and water flow for easy cleaning. The single-frequency vibration motor starts during cleaning, further loosening stubborn materials through vibration, significantly improving cleaning efficiency and reducing downtime.

[0019] A scraper is connected to the bottom of the stirring shaft, and the bottom of the scraper contacts the filter plate. The scraper is installed at the bottom of the stirring shaft and contacts the surface of the filter plate. During the rotation of the stirring shaft, the scraper continuously scrapes the surface of the filter plate, rubbing and squeezing out any clumps, effectively removing residual material adhering to the area around the through holes, preventing blockages, and ensuring the filter plate remains unobstructed for a long time.

[0020] The control module uses a PLC controller; the PLC controller includes a hardware module and a software control module.

[0021] The hardware module includes an analog input module, which is electrically connected to the online laser particle size analyzer and the feed rate sensor.

[0022] The hardware module also includes a high-speed counter module and an analog output module. The high-speed counter module collects the electrical signal output by the material quantity sensor, and the analog output module is electrically connected to the execution module.

[0023] The hardware module also includes a communication module, which supports PROFIBUS / DP and PROFINET protocols and is electrically connected to the human-machine interface module and the execution module.

[0024] The execution module includes a frequency converter and a permanent magnet synchronous motor. The frequency converter 9 supports input of analog or digital commands. The permanent magnet synchronous motor has a built-in encoder. The frequency converter 9 outputs electrical signals to connect to the permanent magnet synchronous motor.

[0025] The human-machine interaction module uses a host computer, which is equipped with a communication interface to synchronize data with the PLC in real time via PROFINET or Ethernet.

[0026] Furthermore, the analog signal range is 4–20 mA, corresponding to different granularity ranges, such as 0–100 μm. The digital signal adopts the Modbus RTU / TCP protocol, supports RS-485 or Ethernet interface, and has a sampling frequency of ≥1 Hz to ensure the real-time performance of granular data.

[0027] Furthermore, compared to traditional offline detection, which is delayed by several minutes, or low-frequency online detection, such as once per minute, the data update speed is increased by 60 times, significantly shortening the data feedback cycle.

[0028] Furthermore, the detection module also includes a feed rate sensor, which uses a belt scale or electromagnetic flow meter to monitor the ore input rate. The signal output is a pulse signal or an analog signal. When a pulse signal is output, it is acquired by the PLC high-speed counter HSC. When an analog signal is output, the analog signal range is 4 to 20 mA, corresponding to different flow ranges, such as 0 to 50 t / h.

[0029] Furthermore, analog signals use shielded twisted-pair cables, while digital signals are transmitted via the PROFIBUS DIP bus to avoid electromagnetic interference.

[0030] Furthermore, the detection module also includes a medium filling rate detection module; the medium filling rate detection module includes a vibration sensor and a pressure sensor;

[0031] Furthermore, vibration sensors are used to monitor the vibration spectrum of the tower mill cylinder and indirectly calculate the filling rate of steel balls / ceramic balls;

[0032] Furthermore, a pressure sensor is installed at the bottom of the tower mill to monitor the pressure at the bottom of the mill and calculate the filling rate by combining it with the media density;

[0033] Furthermore, an analog signal is output, with a range of 4–20 mA and a fill rate of 0–100%.

[0034] Furthermore, vibration and pressure sensors need to be manually calibrated periodically to prevent sensor misalignment and inaccurate data.

[0035] Furthermore, the control module is responsible for data processing, algorithm calculation, and instruction output. The control module outputs electrical signals to connect to the execution module, and the control module is bidirectionally electrically connected to the human-machine interaction module.

[0036] Furthermore, the control module adopts a PLC controller; the PLC controller includes a hardware module and a software control module.

[0037] Furthermore, the hardware module includes an analog input module, which receives 4-20mA signals output from the online laser particle size analyzer and the medium filling rate detection module;

[0038] Furthermore, it also includes a high-speed counter module, which collects the pulse signal output by the material quantity sensor;

[0039] Furthermore, it also includes an analog output module, which outputs a 4-20mA speed command to the frequency converter to adjust the speed.

[0040] Furthermore, the hardware module also includes a communication module that supports PROFIBUS / DP and PROFINET protocols and connects to the host computer and the frequency converter.

[0041] Furthermore, the software control module includes a fuzzy PID algorithm module, which has a built-in adaptive PID parameter tuning algorithm, receives granularity deviation signals, and dynamically calculates the speed adjustment amount.

[0042] Furthermore, the adaptive PID parameter tuning algorithm dynamically adjusts the ratio (K) by integrating fuzzy logic reasoning and genetic algorithm optimization. p ), integral (K) i ), differential (K) d ) parameters to address the nonlinear, time-varying, and multivariable coupling characteristics of the tower mill system, including the following steps:

[0043] Start the fuzzy PID controller and load the preset initial values ​​of the PID parameters (K). p0 ),(K i0 ),(K d0 ).

[0044] Set target granularity D50 target For example, 45μm.

[0045] Real-time data acquisition is performed, and the current discharge particle size D50 is obtained through a particle size analyzer. current .

[0046] The self-tuning process for fuzzy PID parameters is as follows:

[0047] 1. Perform online parameter adjustment for fuzzy logic:

[0048] (1) Input variable definition;

[0049] Particle size deviation: e(t)=D50 current -D50 target

[0050] Deviation change rate:

[0051] (2) Fuzzification process: Fuzzy set partitioning: Fuzzy set of granularity deviation e(t):

[0052] {Negative Large (NB), Negative Small (NS), Zero (ZO), Positive Small (PS), Positive Large (PB)}

[0053] The fuzzy set of the rate of change of deviation Δe(t):

[0054] {Rapid Decrease (FD), Slow Decrease (SD), Steady (ST), Slow Increase (SI), Rapid Increase (FI)}

[0055] Membership function: Use trigonometric function or Gaussian function;

[0056] For example: e(t) = positive (PB): when e(t) > 10μm, the membership degree is 1, and it decreases linearly as e(t) decreases.

[0057] For example: Δe(t) = Rapidly increasing (FI): when Δe(t) > 2μm / s, the membership degree is 1.

[0058] (3) Fuzzy rule base design: Rule form:

[0059] IFe(t)ISA i ANDΔe(t)IS B j THENΔK p =C k ΔK i =D l ΔK d =E m

[0060] Typical rule example: If e(t) = positive (PB) and Δe(t) = rapidly increasing (FI), then K increases significantly. p Slight increase in K i Suppress K d .

[0061] Furthermore, if e(t) = zero (ZO) and Δe(t) = stable (ST), then fine-tune K. i Eliminate steady-state error.

[0062] (4) Defuzzification output and parameter update:

[0063] The defuzzification method uses the centroid method (COG) to convert the fuzzy output into precise PID parameter correction values.

[0064]

[0065] Dynamic update of PID parameters: K p =K p0 +ΔK p

[0066] K i =K i0 +ΔK i

[0067] K d =K d0 +ΔK d

[0068] Furthermore, it is necessary to limit the range of parameters, such as K. p ∈[0,10], to avoid integral saturation or differential noise amplification.

[0069] Furthermore, it also includes a feedforward compensation module and a multivariate optimization module;

[0070] Furthermore, the feedforward compensation module combines the feed rate sensor data to pre-adjust the rotation speed and reduce the lag effect. Its working principle is as follows: The feedforward compensation module receives the ore input rate monitored by the belt scale or electromagnetic flow meter in real time, and constructs a disturbance prediction model based on the mapping relationship between feed rate change and particle size disturbance established based on historical data.

[0071] The compensation logic of the feedforward compensation module is as follows: Based on the change in feed rate ΔQ, the production speed is pre-adjusted according to a preset ratio:

[0072] Δn 前馈 =K f ·ΔQ

[0073] Among them, K f This is the feedforward gain coefficient, which needs to be obtained through experimental calibration or model calculation.

[0074] The feedforward compensation is added to the PID output to form the total speed command:

[0075] n 总 =n PID +Δn 前馈

[0076] Furthermore, if the system detects that the granularity deviation is still large after feedforward compensation, K can be dynamically corrected using fuzzy logic. f .

[0077] Furthermore, the multivariate optimization module establishes a dynamic relationship matrix of rotational speed, feed rate, and medium filling rate through a genetic algorithm, calls the parameter library generated by the genetic algorithm, coordinates the relationship between rotational speed, feed rate, and filling rate, and realizes a multivariate collaborative mechanism to avoid system instability or increased energy consumption caused by adjusting a single parameter.

[0078] Simultaneously, a genetic algorithm is used to train the optimal control parameter library offline and match operating conditions online in real time. The working principle is as follows: the optimization objective is to minimize granularity error and energy consumption, and the fitness function is defined as follows:

[0079] F=α·RMSE(D50)+β·Energy+γ·Wear

[0080] Where: α, β, γ are weighting coefficients, RMSE is the root mean square error of particle size, Energy is the unit energy consumption, and Wear is the media wear rate.

[0081] The genetic manipulation procedure is as follows;

[0082] a. Population initialization: Randomly generate multiple sets of PID combination parameters and multivariate collaborative parameters;

[0083] b. Selection: Use a roulette wheel selection based on fitness to retain high-quality individuals.

[0084] c. Crossover and mutation: Crossover: exchanging some parameters between different individuals;

[0085] Mutation: Increasing diversity by randomly perturbing parameter values ​​with low probability;

[0086] D. Iterative convergence: Repeat selection, crossover, and mutation until the fitness function converges or the maximum number of iterations is reached.

[0087] A coupled model of rotation speed, feed rate, and medium filling rate is established, and multi-parameter collaborative optimization is achieved through PLC.

[0088] Furthermore, a parameter library is constructed, and the optimized parameter combinations are stored in the PLC data port and classified according to working conditions. During online operation, the PLC calls the matching parameters according to the real-time working conditions to achieve multi-variable collaborative optimization, while updating the dynamic relationship matrix based on real-time data.

[0089] Furthermore, when the speed increases and causes the motor to overload, the feed rate or filling rate is automatically reduced to maintain system stability.

[0090] Furthermore, the execution module includes a frequency converter and a permanent magnet synchronous motor. The frequency converter supports input of analog or digital commands and outputs a three-phase frequency converter to control the speed of the permanent magnet synchronous motor. The permanent magnet synchronous motor has a built-in encoder.

[0091] Furthermore, the human-machine interaction module adopts a host computer, which is equipped with a communication interface to synchronize data with the PLC in real time via PROFINET or Ethernet.

[0092] Furthermore, the host computer displays real-time data including particle size curves, rotational speed, feed rate, energy consumption statistics, etc., and also provides a parameter setting interface including target particle size setting, PID self-tuning switch, alarm threshold, etc.

[0093] Furthermore, a closed loop of "particle size deviation - speed adjustment" is constructed, and the speed of the tower mill drive motor is dynamically adjusted through a frequency converter.

[0094] The closed-loop control process is as follows;

[0095] 1. Particle size detection and deviation calculation: The particle size analyzer collects the output D50 in real time and transmits the data to the PLC; the calculation of particle size deviation e(t) is the same as the process in the fuzzy PID parameter self-tuning above.

[0096] 2. Dynamic adjustment of fuzzy PID parameters: The input is fuzzified, and the specific steps are the same as those in the self-tuning of fuzzy PID parameters above;

[0097] Rule base application: If |e(t)|>10μm, then K is significantly improved. p With a rapid response;

[0098] If |e(t)| < 5 μm, then enhance K i To eliminate steady-state error;

[0099] If Δe(t) changes abruptly, then K is activated. d Suppress overshoot;

[0100] Perform defuzzification output; calculate ΔK using the centroid method (COG). p ΔK i ΔK d The PID parameters are updated dynamically.

[0101] 3. Generation and execution of speed commands;

[0102] PID controller output speed correction:

[0103]

[0104] The frequency converter receives commands and adjusts the motor speed, which directly affects the grinding intensity and the output particle size.

[0105] 4. Closed-loop feedback verification: The updated granular data re-enters the detection layer, forming a closed loop of "detection → calculation → adjustment → re-detection".

[0106] Furthermore, a dual closed-loop control structure is constructed; the outer loop granular control uses the D50 deviation as input and adjusts the speed through fuzzy PID.

[0107] Furthermore, the inner loop speed control: the frequency converter adjusts the motor speed in a closed loop to ensure execution accuracy. Attached Figure Description

[0108] To more clearly illustrate the technical solutions in the embodiments of this utility model or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0109] Figure 1 This is a diagram of the overall system architecture.

[0110] Figure 2 The flowchart for fuzzy PID parameter self-tuning is shown below.

[0111] Figure 3 This is a flowchart of the closed-loop control process.

[0112] Figure 4 This is a schematic diagram showing the distribution and connection of sensors in a tower mill.

[0113] Figure 5 for Figure 4 Enlarged view of section A in the middle;

[0114] Figure 6 This is a 3D structural diagram of the filter plate.

[0115] In the diagram: 1. Tower mill; 2. Vibration sensor; 3. Feed pipe; 4. Filter plate; 401. Fan-shaped filter plate; 402. Through hole; 403. Mounting hole; 404. Sealing groove; 5. Feed rate sensor; 6. Pressure sensor; 7. Online laser particle size analyzer; 8. PLC controller; 9. Frequency converter; 10. Permanent magnet synchronous motor; 12. Single-frequency vibration motor; 13. Scraper; 14. Cleaning nozzle. Detailed Implementation

[0116] To make the objectives, technical solutions, and advantages of this utility model clearer, the following detailed description is provided in conjunction with embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this utility model.

[0117] The application principle of this utility model will be further described below with reference to the accompanying drawings and specific embodiments.

[0118] like Figure 1-3 As shown, a tower mill based on adaptive PID control includes a tower mill body 1 and a detection module. The detection module outputs an electrical signal to connect to a control module. The detection module includes an online laser particle size analyzer 7, which is installed at the discharge port pipe of the tower mill 1. The online laser particle size analyzer 7 detects the particle size distribution of the ore in real time, using the D50 value, i.e., the median particle size, as the core indicator, and simultaneously outputs analog and digital signals.

[0119] The detection module includes an online laser particle size analyzer 7, which is horizontally embedded in the center of the side wall of the discharge pipe of the tower mill 1.

[0120] It also includes a feed rate sensor 5, which is installed inside the flange of the feed pipe 3 of the tower mill 1 and is coaxially fixed with the feed pipe 3;

[0121] It also includes a media filling rate detection module, which includes a vibration sensor 2 and a pressure sensor 6; the vibration sensor 2 is installed in a ring array on the upper middle part of the outer wall of the tower mill 1 cylinder.

[0122] The detection window of the online laser particle size analyzer 7 is inclined at a 30° angle to the axis of the discharge pipe. The circular array of vibration sensors 2 is spaced at least 1 / 8 of the circumference of the tower mill 1, and the sensor orientation is consistent with the tangential direction of the spiral agitator.

[0123] The vibration sensor 2, which is installed on the outer wall of the tower mill 1, is covered with a wear-resistant rubber sleeve. The thickness of the wear-resistant rubber sleeve is 1:3 of the height of the sensor, and the surface of the wear-resistant rubber sleeve is provided with heat dissipation fins.

[0124] The pressure sensor 6 is mounted on the surface of the filter plate 4. The pressure sensor 6 is arranged in a uniform ring around the center of the bearing seat, and the sensing surface of the pressure sensor 6 is parallel to the surface of the filter plate 4.

[0125] The filter plate 4 is installed at the bottom of the stirring chamber of the tower mill 1. The filter plate 4 includes at least four fan-shaped filter plates 401, and a sealing ring is provided between the fan-shaped filter plates 401. It also includes a stirring shaft, and the fan-shaped filter plates 401 are connected to the stirring shaft by flange bolts.

[0126] The fan-shaped filter plates 401 are tightly connected by sealing rings and sealing grooves 403. A cleaning nozzle 14 is installed on the bottom wall of the mixing chamber. The flange bolts are fixed to ensure the overall structure is airtight, avoid leakage of liquid or waste during the cleaning process, and ensure the stable operation of the system.

[0127] The filter plate 4 has a central mounting hole 404 and a sealing groove 403 around its edge. A single-frequency vibration motor 12 is installed at the flange connection of the filter plate 4. The surface of the filter plate 4 has a through hole 402, which is tapered. The tapered through hole 402 design of the filter plate combines gravity and water flow to accelerate waste discharge. The single-frequency vibration motor 12 starts during cleaning, further loosening stubborn materials through vibration, significantly improving cleaning efficiency and shortening downtime.

[0128] A scraper 13 is connected to the bottom of the stirring shaft, and the bottom of the scraper 13 contacts the filter plate 4. The scraper 13 is installed at the bottom of the stirring shaft and contacts the surface of the filter plate 4. During the rotation of the stirring shaft, the scraper continuously scrapes the surface of the filter plate, effectively removing residual material adhering to the area around the through hole 402, preventing blockage, and ensuring that the filter plate remains unobstructed for a long time.

[0129] The control module adopts a PLC controller 8; the PLC controller 8 includes a hardware module and a software control module.

[0130] The hardware module includes an analog input module, which is electrically connected to the online laser particle size analyzer 7 and the feed rate sensor 5.

[0131] The hardware module also includes a high-speed counter module and an analog output module. The high-speed counter module collects the electrical signal output by the material quantity sensor, and the analog output module is electrically connected to the execution module.

[0132] The hardware module also includes a communication module that supports PROFIBUSDP and PROFINET protocols and is electrically connected to the human-machine interaction module and the execution module.

[0133] The execution module includes a frequency converter 9 and a permanent magnet synchronous motor 10. The frequency converter 9 supports input of analog or digital commands. The permanent magnet synchronous motor 10 has a built-in encoder. The frequency converter 9 outputs an electrical signal to connect to the permanent magnet synchronous motor 10.

[0134] The human-machine interaction module adopts a host computer, which is equipped with a communication interface to synchronize data with the PLC in real time via PROFINET or Ethernet.

[0135] The analog signal range is 4–20 mA, corresponding to different granularity ranges, such as 0–100 μm. The digital signal adopts the Modbus RTU / TCP protocol, supports RS-485 or Ethernet interface, and has a sampling frequency of ≥1 Hz to ensure the real-time performance of granular data.

[0136] Compared to traditional offline detection, which has a delay of several minutes, or low-frequency online detection, such as once per minute, the data update speed is increased by 60 times, significantly shortening the data feedback cycle.

[0137] The detection module also includes a feed rate sensor 5, which uses a belt scale or an electromagnetic flow meter to monitor the ore input rate. The signal output is a pulse signal or an analog signal. When a pulse signal is output, it is acquired by a PLC high-speed counter HSC. When an analog signal is output, the analog signal range is 4 to 20 mA, corresponding to different flow ranges, such as 0 to 50 t / h.

[0138] The detection module also includes a medium filling rate detection module; the medium filling rate detection module includes a vibration sensor 2 and a pressure sensor 6.

[0139] Vibration sensor 2 is used to monitor the vibration spectrum of the tower mill 1 cylinder and indirectly calculate the filling rate of steel balls / ceramic balls;

[0140] Pressure sensor 6 is installed at the bottom of tower mill 1 to monitor the bottom pressure and calculate the filling rate by combining the medium density.

[0141] It outputs an analog signal with a range of 4–20 mA and a fill rate of 0–100%.

[0142] Vibration sensor 2 and pressure sensor 6 need to be manually calibrated regularly to prevent sensor misalignment and inaccurate data.

[0143] The control module is responsible for data processing, algorithm calculation and instruction output. The control module outputs electrical signals to connect to the execution module, and the control module is bidirectionally electrically connected to the human-machine interaction module.

[0144] The control module uses a PLC controller; the PLC controller includes a hardware module and a software control module.

[0145] The hardware module includes an analog input module, which receives 4-20mA signals output from the online laser particle size analyzer 7 and the medium filling rate detection module;

[0146] It also includes a high-speed counter module, which collects the pulse signal output by the material quantity sensor;

[0147] It also includes an analog output module, which outputs a 4-20mA speed command to the frequency converter 9 to adjust the speed.

[0148] The hardware module also includes a communication module that supports PROFIBUSDP and PROFINET protocols and connects to the host computer and the frequency converter 9.

[0149] The software control module includes a fuzzy PID algorithm module, which has a built-in adaptive PID parameter tuning algorithm, receives granularity deviation signals, and dynamically calculates the speed adjustment amount.

[0150] The adaptive PID parameter tuning algorithm dynamically adjusts the ratio (K) by integrating fuzzy logic reasoning and genetic algorithm optimization. p ), integral (K) i ), differential (K) d ) parameters to address the nonlinear, time-varying, and multivariable coupling characteristics of the tower mill 1 system, including the following steps:

[0151] Start the fuzzy PID controller and load the preset initial values ​​of the PID parameters (K). p0 ),(K i0 ),(K d0 ).

[0152] Set target granularity D50 target For example, 45μm.

[0153] Real-time data acquisition is performed, and the current discharge particle size D50 is obtained through a particle size analyzer. current .

[0154] 1. Perform online parameter adjustment for fuzzy logic:

[0155] (1) Input variable definition;

[0156] Particle size deviation: e(t)=D50 current -D50 target

[0157] Deviation change rate:

[0158] (5) Fuzzification process: Fuzzy set partitioning: Fuzzy set of granularity deviation e(t):

[0159] {Negative Large (NB), Negative Small (NS), Zero (ZO), Positive Small (PS), Positive Large (PB)}

[0160] The fuzzy set of the rate of change of deviation Δe(t):

[0161] {Rapid Decrease (FD), Slow Decrease (SD), Steady (ST), Slow Increase (SI), Rapid Increase (FI)}

[0162] Membership function: Use trigonometric function or Gaussian function;

[0163] For example: e(t) = positive (PB): when e(t) > 10μm, the membership degree is 1, and it decreases linearly as e(t) decreases.

[0164] For example: Δe(t) = Rapidly increasing (FI): when Δe(t) > 2μm / s, the membership degree is 1.

[0165] (6) Fuzzy rule base design: Rule format:

[0166] IFe(t)ISA i ANDΔe(t)IS B j THENΔK p =C k ΔK i =D l ΔK d =E m

[0167] Typical rule example: If e(t) = positive (PB) and Δe(t) = rapidly increasing (FI), then K increases significantly. p Slight increase in K i Suppress K d .

[0168] If e(t) = zero (ZO) and Δe(t) = stable (ST), then fine-tune K. i Eliminate steady-state error.

[0169] (7) Defuzzification output and parameter update:

[0170] The defuzzification method uses the centroid method (COG) to convert the fuzzy output into precise PID parameter correction values.

[0171]

[0172] Dynamic update of PID parameters: K p =K p0 +ΔK p

[0173] K i =K i0 +ΔK i

[0174] K d =K d0 +ΔK d

[0175] The parameter range needs to be limited, for example, K. p ∈[0,10], to avoid integral saturation or differential noise amplification.

[0176] It also includes a feedforward compensation module and a multivariate optimization module;

[0177] The feedforward compensation module combines the data from the feed rate sensor 5 to pre-adjust the rotation speed and reduce the lag effect. Its working principle is as follows: The feedforward compensation module receives the ore input rate monitored by the belt scale or electromagnetic flow meter in real time, and constructs a disturbance prediction model based on the mapping relationship between feed rate change and particle size disturbance established by historical data.

[0178] The compensation logic of the feedforward compensation module is as follows: Based on the change in feed rate ΔQ, the production speed is pre-adjusted according to a preset ratio:

[0179] Δn 前馈 =K f ·ΔQ

[0180] Among them, K f This is the feedforward gain coefficient, which needs to be obtained through experimental calibration or model calculation.

[0181] The feedforward compensation is added to the PID output to form the total speed command:

[0182] n 总 =n PID +Δn 前馈

[0183] If the system detects that the granularity deviation is still large after feedforward compensation, K can be dynamically corrected using fuzzy logic. f .

[0184] The multivariate optimization module establishes a dynamic relationship matrix of rotation speed, feed rate, and medium filling rate through a genetic algorithm, calls the parameter library generated by the genetic algorithm, coordinates the relationship between rotation speed, feed rate, and filling rate, and realizes a multivariate collaborative mechanism to avoid system instability or increased energy consumption caused by adjusting a single parameter.

[0185] Simultaneously, a genetic algorithm is used to train the optimal control parameter library offline and match operating conditions online in real time. The working principle is as follows: the optimization objective is to minimize granularity error and energy consumption, and the fitness function is defined as follows:

[0186] F=α·RMSE(D50)+β·Energy+γ·Wear

[0187] Where: α, β, γ are weighting coefficients, RMSE is the root mean square error of particle size, Energy is the unit energy consumption, and Wear is the media wear rate.

[0188] The genetic manipulation procedure is as follows;

[0189] d. Population initialization: Randomly generate multiple sets of PID combination parameters and multivariate cooperative parameters;

[0190] e. Selection: Use a roulette wheel selection based on fitness to retain high-quality individuals.

[0191] f. Crossover and Mutation: Crossover: Exchanging some parameters between different individuals;

[0192] Mutation: Increasing diversity by randomly perturbing parameter values ​​with low probability;

[0193] E. Iterative convergence: Repeat selection, crossover, and mutation until the fitness function converges or the maximum number of iterations is reached.

[0194] A coupled model of rotation speed, feed rate, and medium filling rate is established, and multi-parameter collaborative optimization is achieved through PLC.

[0195] A parameter library is constructed, and the optimized parameter combinations are stored in the PLC data port and classified according to the working conditions. When running online, the PLC calls the matching parameters according to the real-time working conditions to achieve multi-variable collaborative optimization, and updates the dynamic relationship matrix based on real-time data.

[0196] When the speed increases and causes the motor to overload, the feed rate or filling rate is automatically reduced to maintain system stability.

[0197] The execution module includes a frequency converter 9 and a permanent magnet synchronous motor 10. The frequency converter 9 supports input of analog or digital commands, and the permanent magnet synchronous motor 10 has a built-in encoder.

[0198] The human-machine interaction module adopts a host computer, which is equipped with a communication interface to synchronize data with the PLC in real time via PROFINET or Ethernet.

[0199] A closed loop of "particle size deviation - speed adjustment" is constructed, and the speed of the drive motor of the tower mill 1 is dynamically adjusted through the frequency converter 9.

[0200] The closed-loop control process is as follows;

[0201] 5. Particle size detection and deviation calculation: The particle size analyzer collects the output D50 in real time and transmits the data to the PLC; the calculation of particle size deviation e(t) is the same as the process in the fuzzy PID parameter self-tuning above.

[0202] 6. Dynamic adjustment of fuzzy PID parameters: The input is fuzzified, and the specific steps are the same as those in the self-tuning of fuzzy PID parameters above.

[0203] Rule base application: If |e(t)|>10μm, then K is significantly improved. p With a rapid response;

[0204] If |e(t)| < 5 μm, then enhance K i To eliminate steady-state error;

[0205] If Δe(t) changes abruptly, then K is activated. d Suppress overshoot;

[0206] Perform defuzzification output; calculate ΔK using the centroid method (COG). p ΔK i ΔK d The PID parameters are updated dynamically.

[0207] 7. Generation and execution of speed commands;

[0208] PID controller output speed correction:

[0209]

[0210] The frequency converter 9 receives instructions and adjusts the motor speed, which directly affects the grinding intensity and the output particle size.

[0211] 4. Closed-loop feedback verification: The updated granular data re-enters the detection layer, forming a closed loop of "detection → calculation → adjustment → re-detection".

[0212] It will be apparent to those skilled in the art that this invention is not limited to the details of the exemplary embodiments described above, and that it can be implemented in other specific forms without departing from the spirit or essential characteristics of this invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of this invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within this invention.

[0213] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A tower mill based on adaptive PID control, comprising a detection module, a control module, and an execution module installed on the tower mill (1), characterized in that: The detection module includes a medium filling rate detection module, which includes a vibration sensor (2) and a pressure sensor (6). The vibration sensor (2) is installed in a ring array on the upper middle part of the outer wall of the tower mill (1); it also includes a filter plate (4), the pressure sensor (6) is installed on the surface of the filter plate (4), and the sensing surface of the pressure sensor (6) is parallel to the bottom surface of the tower mill (1). The pressure sensors (6) are arranged in a uniform ring around the center of the bearing housing.

2. A tower mill based on adaptive PID control according to claim 1, characterized in that: The spacing of the ring array of vibration sensors (2) is at least 1 / 8 of the circumference of the tower mill (1) shell, and the sensor orientation is consistent with the tangential direction of the spiral stirrer.

3. A tower mill based on adaptive PID control according to claim 2, characterized in that: The surface of the vibration sensor (2) installed on the outer wall of the tower mill (1) is covered with a wear-resistant rubber sleeve.

4. A tower mill based on adaptive PID control according to claim 2, characterized in that: The detection module includes an online laser particle size analyzer (7), which is horizontally embedded in the center of the side wall of the discharge pipe of the tower mill (1); the detection window of the online laser particle size analyzer (7) is inclined at a 30° angle to the axis of the discharge pipe.

5. A tower mill based on adaptive PID control according to claim 4, characterized in that: The detection module also includes a feed rate sensor (5), which is installed inside the flange of the feed pipe (3) of the tower mill (1) and is coaxially fixed with the feed pipe (3).

6. A tower mill based on adaptive PID control according to claim 1, characterized in that: The control module adopts a PLC controller (8); the PLC controller (8) includes a hardware module and a software control module.

7. A tower mill based on adaptive PID control according to claim 1, characterized in that: The filter plate (4) is installed at the bottom of the stirring chamber of the tower mill (1). The filter plate (4) includes at least 4 fan-shaped filter plates (401), and a sealing ring is provided between the fan-shaped filter plates (401). It also includes a stirring shaft. The fan-shaped filter plates (401) are connected to the stirring shaft by flange bolts.

8. A tower mill based on adaptive PID control according to claim 7, characterized in that: The filter plate (4) has a mounting hole (404) in the center and a sealing groove (403) on the edge of the mounting hole (404). A single-frequency vibration motor (12) is installed at the flange connection of the filter plate (4). The filter plate (4) has a through hole (402) on its surface. The through hole (402) is a tapered through hole.

9. A tower mill based on adaptive PID control according to claim 7, characterized in that: A scraper (13) is connected to the bottom of the stirring shaft, and the bottom of the scraper (13) is in contact with the filter plate (4).

Images