A high-precision aluminum smelting raw material quantitative feeding device and method

Abstract

The application provides a high-precision aluminum smelting raw material quantitative feeding device and method, and belongs to the technical field of aluminum smelting raw material feeding. The device comprises a multi-channel raw material storage and pretreatment unit, a high-precision dynamic weighing and flow sensing unit, a core intelligent control and data processing unit, an actuator driving unit and a system state monitoring and self-correction unit. The multi-channel raw material storage and pretreatment unit is designed with a 65° bin cone angle, a high-frequency pneumatic arch breaking device and a variable-pitch double-spiral feeder, and ensures stable flow of raw materials based on the principle of material mechanics; the high-precision dynamic weighing and flow sensing unit is configured with a combination of a weighing sensor and a Coriolis mass flowmeter, and realizes multi-dimensional data fusion and error self-correction by combining a temperature compensation mechanism. The application improves the feeding precision and fundamentally solves the stability and precision problems of aluminum smelting raw material quantitative feeding.

Description

Technical Field

[0001] This invention relates to the field of aluminum smelting raw material feeding technology, specifically to a high-precision aluminum smelting raw material quantitative feeding device and method. Background Technology

[0002] Aluminum smelting, as a core link in the non-ferrous metals industry, places extremely high demands on the precision and stability of the raw material (mainly including alumina, fluoride salts, and additives) ratios. The accuracy of raw material feeding directly determines the thermal balance and current efficiency of the electrolytic cell, ultimately affecting the purity and overall energy consumption of the final aluminum product. In modern aluminum industry automation systems, the quantitative feeding system is a key node connecting raw material storage and the electrolysis process; its performance directly determines the controllability and economy of the entire smelting process.

[0003] To achieve stable and efficient smelting production, the aluminum smelting process places extremely stringent requirements on the control of raw material flow fluctuations, typically requiring the fluctuation range to be controlled within ±0.5%. In existing technologies, aluminum smelting raw material quantitative feeding systems usually employ a solution based on belt scales or a single screw feeder, combined with weighing sensors and a simple PID (proportional-integral-derivative) control algorithm, in order to achieve the metering and conveying of powdered or granular raw materials.

[0004] The invention patent with announcement number CN103171877B discloses a double-helix differential quantitative conveying system and its control method. The system consists of a double-helix conveyor and a double-helix differential quantitative control device. The double-helix conveyor includes a body, an inlet, an outlet, a main helix, a main helix motor, an auxiliary helix, and an auxiliary helix motor. The double-helix differential quantitative control device includes a main control unit, a weighing sensor, a human-machine interface device, a switch input / output module, a main helix driver, and an auxiliary helix driver. The control method performs digital filtering and feature analysis on the collected weight values ​​to extract weight information and impact information respectively. The impact amount introduced by the descent of bulk materials is filtered out from the weight information. Then, the estimated hang weight value is added to this information to perform estimated hang weight compensation.

[0005] However, existing technical solutions have the following inherent drawbacks when dealing with the complex object of aluminum smelting raw materials due to their variable physical properties and strong dynamic coupling of the system:

[0006] First, the existing hopper design often fails to fundamentally address the material flowability issue. When the moisture content of the raw material (such as alumina) changes or its particle size is uneven, bridging or adhesion can easily occur at the bottom of the hopper, leading to interruptions or fluctuations in material feeding. Traditional screw feeders experience a significant decrease in conveying stability when dealing with such situations, causing metering drift and making it difficult to maintain high precision over the long term.

[0007] Secondly, existing systems mostly rely on a single weighing sensor, whose measurement signal is easily affected by factors such as changes in ambient temperature and mechanical vibration. Furthermore, they lack an effective real-time self-compensation mechanism, which leads to the continuous accumulation of measurement errors during the production process, making it impossible to meet the requirements for long-term stable operation.

[0008] Third, most existing systems employ open-loop or simple PID control. As a linear controller, the PID controller struggles to effectively compensate for nonlinear and time-varying feeding deviations caused by silo pressure fluctuations, material bridging, or equipment mechanical wear. Furthermore, in scenarios involving multi-raw material co-feeding, dynamic coupling interference exists between different feeding channels, and existing control strategies lack effective decoupling methods. This means that adjusting one channel can affect the stability of other channels, making it difficult to maintain overall proportioning accuracy synchronously.

[0009] The aforementioned defects are particularly prominent in large-scale, continuous aluminum smelting production. They can easily cause fluctuations in the composition of the electrolytic cell, increase energy consumption, and may even lead to abnormal operation of the electrolytic cell, resulting in significant economic losses. Summary of the Invention

[0010] The purpose of this invention is to provide a high-precision quantitative feeding device and method for aluminum smelting raw materials, which can effectively solve the above-mentioned technical problems.

[0011] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0012] A high-precision quantitative feeding device for aluminum smelting raw materials includes:

[0013] A multi-channel raw material storage and pretreatment unit is used to store raw materials and perform physical pretreatment on the raw materials. The multi-channel raw material storage and pretreatment unit includes at least 3 independent silos, and each silo outlet is connected to a twin-screw feeder.

[0014] A high-precision dynamic weighing and flow sensing unit is connected to the output end of the twin-screw feeder to detect and output the instantaneous mass flow rate and cumulative feed amount of each raw material channel in real time.

[0015] The core intelligent control and data processing unit is communicatively connected to the high-precision dynamic weighing and flow sensing unit, and is used to receive sensing data, execute control algorithms and output control commands.

[0016] The actuator drive unit receives instructions from the core intelligent control and data processing unit and drives the twin-helix feeder to precisely adjust the feeding rate.

[0017] The system status monitoring and self-calibration unit is integrated into key nodes to continuously monitor operating parameters and trigger self-calibration procedures.

[0018] In one embodiment of the present invention, in the multi-channel raw material storage and pretreatment unit, the inner wall of each silo is lined with ultra-high molecular weight polyethylene wear-resistant lining plate, the bottom cone angle of the silo is designed to be 65° and equipped with a high-frequency pneumatic arch-breaking device, the working frequency of the arch-breaking device is dynamically adjusted according to the material repose angle, and the adjustment strategy is based on the comparison result of the material humidity sensor feedback value and the preset threshold; the main and auxiliary screws of the twin-screw feeder adopt a variable pitch design, the pitch of the main screw increases along the discharge direction, the pitch of the auxiliary screw is constant, and the speed ratio of the two is controlled within the range of 1:1.5 to 1:2.2.

[0019] In one embodiment of this invention, the high-precision dynamic weighing and flow sensing unit employs a combination of a strain-based weighing sensor and a Coriolis mass flow meter. The weighing sensor is located on the support structure of the twin-screw feeder, with a measurement range covering 0 to 500 kg, a resolution of 0.01%FS, and a built-in temperature compensation module with a compensation coefficient of: ;in, This is the temperature compensation coefficient for the weighing sensor, used to correct the impact of temperature changes on measurement accuracy. This represents the change in resistance of the sensor strain gauge caused by temperature changes. The initial resistance of the strain gauge in the sensor; The measurement is based on changes in ambient temperature; the Coriolis mass flow meter is installed in the downstream pipeline of the feeder and outputs material density parameters in real time.

[0020] In one embodiment of the present invention, the core intelligent control and data processing unit integrates an adaptive predictive controller and a multivariable decoupling compensator; the adaptive predictive controller uses a recursive least squares method with a forgetting factor to identify the model parameters of the controlled object online, and the objective function is:

[0021] ;

[0022] : is the objective function of the Adaptive Predictive Controller (APC);

[0023] , : These are the summation indices, corresponding to discrete moments in the prediction and control time domains, respectively;

[0024] : This is the forgetting factor, with an initial value of 0.98, used to weight recent data;

[0025] The control weighting coefficient, ranging from 0.1 to 1.0, balances control accuracy and system stability.

[0026] For the time domain prediction, the main dynamic response time of the system is covered, and the value is taken as 15 to 25 sampling periods;

[0027] To control the time domain, take 1 / 3 to 1 / 2 of the prediction time domain;

[0028] :for The actual output of the controlled object at all times, i.e., the instantaneous mass flow rate of the raw materials;

[0029] :for Always Predicted output at time step;

[0030] :for The control increment at each moment, the difference between control values ​​at adjacent moments, corresponds to the change in motor speed of the twin-helix feeder;

[0031] The multivariable decoupling compensator is based on the feedforward compensation principle and uses the inverse model of the coupling channel to counteract the dynamic coupling effect between the raw material feeding channels.

[0032] In one embodiment of the present invention, the actuator drive unit uses a vector control frequency converter to drive a permanent magnet synchronous motor. The carrier frequency of the frequency converter is set to be adjustable from 8 kHz to 12 kHz. The permanent magnet synchronous motor is connected to the main shaft of the twin-helix feeder through a planetary gear reducer. The reduction ratio is determined based on the ratio of the rated torque of the twin-helix feeder to the peak torque of the permanent magnet synchronous motor, and the value range is from 15:1 to 25:1. The rotor position of the permanent magnet synchronous motor is fed back to the frequency converter through a 23-bit absolute encoder, forming a fully closed-loop speed control loop.

[0033] In one embodiment of the present invention, the system status monitoring and self-calibration unit includes a vibration acceleration sensor, an acoustic emission sensor, and an infrared thermal imager. The vibration acceleration sensor is installed on the bearing housing of the twin-screw feeder, and the monitoring frequency range is 10 Hz to 1000 Hz. When the vibration amplitude exceeds the threshold for 2 consecutive seconds, an early warning is triggered. The acoustic emission sensor is arranged on the weighing sensor mounting base of the high-precision dynamic weighing and flow sensing unit, and is used to capture stress wave signals generated by microcracks in the mechanical structure. The infrared thermal imager is aimed at the motor housing and reducer surface of the actuator drive unit, with a sampling period of 5 seconds, and the overheating risk of the equipment is judged by monitoring the temperature rise gradient.

[0034] This invention also discloses a high-precision method for quantitative feeding of raw materials in aluminum smelting, comprising the following steps:

[0035] Step S110: System initialization and parameter self-tuning, loading the physical property parameters of each raw material, the target feed formula and the initial values ​​of the control parameters, and executing the equipment no-load operation to calibrate the sensor zero position;

[0036] Step S120: Multimodal data synchronous acquisition and preprocessing. Raw data from weighing sensors, Coriolis mass flow meters, vibration sensors and temperature sensors are acquired in parallel. The data are then filtered by moving average and outliers are removed.

[0037] Step S130: The feed rate is adaptively controlled. Based on the deviation between the real-time flow feedback value and the target set value, the control output is calculated by the adaptive predictive controller, and independent speed control commands for each channel are generated by the multivariable decoupling compensator.

[0038] Step S140: Real-time system status assessment and fault diagnosis, comprehensively analyze vibration spectrum characteristics, acoustic emission signal energy and temperature distribution, assess the health status of mechanical components and identify potential fault modes.

[0039] Step S150: Periodic online calibration and parameter update. After every 4 hours of continuous operation, the static calibration program is automatically executed to update the sensor calibration coefficients and correct the control model parameters.

[0040] Step S160: Multi-channel coordinated proportioning control. Based on the preset raw material proportioning matrix, the feeding rate of each channel is dynamically coordinated to ensure that the total proportioning error is less than 0.3%.

[0041] In one embodiment of the present invention, the data preprocessing in step S120 specifically includes: performing noise reduction processing on the weighing sensor signal using a 4th-order Butterworth low-pass filter with a cutoff frequency of 50 Hz; using the 3σ criterion to remove instantaneous abnormal values ​​caused by bubble or particle impact from the density reading output by the Coriolis mass flow meter; and performing a 1024-point fast Fourier transform on the vibration signal after windowing through a Hanning window to extract the amplitude values ​​of the 1st, 2nd, and 3rd harmonics as feature quantities.

[0042] In one embodiment of the present invention, the adaptive predictive controller parameter tuning process in step S130 is as follows: the initial value of the forgetting factor λ is set to 0.98, and the control weight coefficient ρ is set between 0.1 and 1.0 according to the response speed requirements; the prediction time domain N covers the main dynamic response time and is set to 15 to 25 sampling periods; the control time domain M is set to 1 / 3 to 1 / 2 of the prediction time domain; and the controller output limit is set according to the maximum safe feeding rate of the twin-helix feeder.

[0043] In one embodiment of the present invention, the online calibration procedure in step S150 includes two stages: static weight calibration and dynamic material calibration. Static weight calibration is performed under no-load conditions on the twin-screw feeder, where five standard weights are applied sequentially, covering 20%, 40%, 60%, 80%, and 100% of the range. The sensor output is recorded and a linear regression equation is fitted. Dynamic material calibration uses standard materials of known mass and feeds them at three different feeding rates. The cumulative feeding mass is compared with the weighing sensor reading to correct the system's dynamic error coefficient.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] This invention addresses the fundamental problem of quantitative feeding of raw materials in aluminum smelting. It tackles the root cause of unstable material flow due to variations in humidity and particle size. The multi-channel raw material storage and pretreatment unit, through optimized silo cone angle design, a high-frequency pneumatic anti-bridging device, and an innovative variable-pitch twin-screw feeder working in tandem, effectively eliminates material bridging and adhesion, enhancing the system's adaptability to changes in raw material physical properties and laying a stable foundation for high-precision feeding.

[0046] Meanwhile, the high-precision dynamic weighing and flow sensing unit adopts a combination of weighing sensor and Coriolis mass flow meter, and combined with the built-in real-time temperature compensation mechanism, it realizes multi-dimensional data fusion and environmental error self-correction from the sensing and measurement level, and the system has stable accuracy in long-term operation.

[0047] More importantly, the core intelligent control and data processing unit solves the core problems of dynamic response lag and multi-channel coupling interference in traditional control algorithms through the deep integration of an adaptive predictive controller and a multivariable decoupling compensator. The adaptive predictive controller identifies the system model online and dynamically optimizes the control output, while the multivariable decoupling compensator cancels the dynamic coupling between channels in real time. This significantly shortens the device adjustment time and ensures a smaller overall proportion error in multi-raw material co-feeding scenarios. It also enables early warning of mechanical failures and periodic online calibration, significantly improving the system's reliability and long-term operational stability from both preventative maintenance and self-calibration perspectives. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0049] Figure 1 This is a flowchart of the control algorithm fusion processing of the present invention.

[0050] Figure 2 This is a flowchart of the online calibration procedure of the present invention.

[0051] Figure 3 This is a simplified overall diagram of the feeding device of the present invention. Detailed Implementation

[0052] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0053] The following is in conjunction with the appendix Figure 1 - Figure 3 The embodiments of the present invention will be described in detail below.

[0054] Example 1:

[0055] This embodiment discloses a high-precision quantitative feeding device for aluminum smelting raw materials, including:

[0056] A multi-channel raw material storage and pretreatment unit is used to store raw materials and perform physical pretreatment on the raw materials. The multi-channel raw material storage and pretreatment unit includes at least 3 independent silos, and each silo outlet is connected to a twin-screw feeder.

[0057] A high-precision dynamic weighing and flow sensing unit is connected to the output end of the twin-screw feeder to detect and output the instantaneous mass flow rate and cumulative feed amount of each raw material channel in real time.

[0058] The core intelligent control and data processing unit is communicatively connected to the high-precision dynamic weighing and flow sensing unit, and is used to receive sensing data, execute control algorithms and output control commands.

[0059] The actuator drive unit receives instructions from the core intelligent control and data processing unit and drives the twin-helix feeder to precisely adjust the feeding rate.

[0060] The system status monitoring and self-calibration unit is integrated into key nodes to continuously monitor operating parameters and trigger self-calibration procedures.

[0061] By dividing the high-precision aluminum smelting raw material quantitative feeding device into five functionally distinct and collaboratively operating units, the fundamental problem of insufficient overall accuracy and stability caused by the isolated functions and lack of coordination of components in existing technologies is solved. The multi-channel raw material storage and pretreatment unit ensures a stable supply of materials; the high-precision dynamic weighing and flow sensing unit provides accurate feedback information; the core intelligent control and data processing unit acts as the brain for intelligent decision-making; the actuator drive unit precisely executes control commands; and the system status monitoring and self-calibration unit ensures the long-term reliable operation of the high-precision aluminum smelting raw material quantitative feeding device. This enables the high-precision aluminum smelting raw material quantitative feeding device to form a complete closed loop of perception, decision-making, execution, and calibration, achieving high precision and high reliability throughout the entire process from raw material supply to rate control.

[0062] Furthermore, in the multi-channel raw material storage and pretreatment unit, the inner wall of each silo is lined with ultra-high molecular weight polyethylene wear-resistant lining plate, the bottom cone angle of the silo is designed to be 65° and equipped with a high-frequency pneumatic arch-breaking device. The working frequency of the arch-breaking device is dynamically adjusted according to the material repose angle, and the adjustment strategy is based on the comparison result of the material humidity sensor feedback value and the preset threshold. The main and auxiliary screws of the twin-screw feeder adopt a variable pitch design. The pitch of the main screw increases along the discharge direction, while the pitch of the auxiliary screw is constant. The speed ratio between the two is controlled within the range of 1:1.5 to 1:2.2.

[0063] By defining the specific structures of the silos and the twin-screw feeder in the multi-channel raw material storage and pretreatment unit, the problems of "bridging," adhesion, and uneven conveying caused by changes in the physical properties of materials are directly solved. The 65° cone angle of the silo, combined with a high-frequency pneumatic anti-bridging device, optimizes the flow path of materials within the silo and dynamically adjusts the anti-bridging frequency based on feedback from the material humidity sensor, thus fundamentally disrupting the mechanical conditions for material arching. The variable pitch design and specific speed ratio of the twin-screw feeder achieve gradual compression and stable pushing of materials through the gradually increasing pitch of the main screw, while the secondary screw plays a role in clearing blockages and assisting in conveying. Together, they ensure a uniform and stable outflow of materials from the discharge port, effectively suppressing feeding fluctuations caused by changes in the physical state of the materials. This creates stable preconditions for the high-precision metering of the subsequent high-precision dynamic weighing and flow sensing unit.

[0064] Furthermore, the high-precision dynamic weighing and flow sensing unit adopts a combination configuration of a strain-based weighing sensor and a Coriolis mass flow meter; the weighing sensor is arranged at the support structure of the twin-screw feeder, its measurement range covers 0 to 500 kg, its resolution reaches 0.01%FS, and it has a built-in temperature compensation module with a compensation coefficient of: ;in, This is the temperature compensation coefficient for the weighing sensor, used to correct the impact of temperature changes on measurement accuracy. This represents the change in resistance of the sensor strain gauge caused by temperature changes. The initial resistance of the strain gauge in the sensor; The measurement is based on changes in ambient temperature; the Coriolis mass flow meter is installed in the downstream pipeline of the feeder and outputs material density parameters in real time.

[0065] By defining the specific configuration and compensation mechanism of the high-precision dynamic weighing and flow sensing unit, the technical problems of single sensors being susceptible to interference and experiencing long-term accuracy degradation are solved. The combination of a weighing sensor and a Coriolis mass flow meter enables redundant measurement and mutual verification of mass flow. The material density parameter provided by the Coriolis mass flow meter also provides an additional process variable for the control algorithm of the core intelligent control and data processing unit. The built-in temperature compensation module of the weighing sensor and its explicit compensation coefficient formula, based on the temperature effect principle of strain gauge resistance, correct for sensor zero-point drift and sensitivity changes caused by ambient temperature variations in real time. This proactively combats major interference sources at the sensing principle level, ensuring the long-term accuracy and stability of measurement data and providing a reliable data foundation for achieving ±0.3% high-precision control.

[0066] The core intelligent control and data processing unit integrates an adaptive predictive controller and a multivariable decoupling compensator. The adaptive predictive controller uses a recursive least squares method with a forgetting factor to identify the controlled object model parameters online. The objective function is:

[0067] ;

[0068] : is the objective function of the Adaptive Predictive Controller (APC);

[0069] , : These are the summation indices, corresponding to discrete moments in the prediction and control time domains, respectively;

[0070] : This is the forgetting factor, with an initial value of 0.98, used to weight recent data;

[0071] The control weighting coefficient, ranging from 0.1 to 1.0, balances control accuracy and system stability.

[0072] For the time domain prediction, the main dynamic response time of the system is covered, and the value is taken as 15 to 25 sampling periods;

[0073] To control the time domain, take 1 / 3 to 1 / 2 of the prediction time domain;

[0074] :for The actual output of the controlled object at all times, i.e., the instantaneous mass flow rate of the raw materials;

[0075] :for Always Predicted output at time step;

[0076] :for The control increment at each moment, the difference between control values ​​at adjacent moments, corresponds to the change in motor speed of the twin-helix feeder;

[0077] The multivariable decoupling compensator is based on the feedforward compensation principle and uses the inverse model of the coupling channel to counteract the dynamic coupling effect between the raw material feeding channels.

[0078] The control algorithm in the core intelligent control and data processing unit directly addresses the pain points of traditional PID control, which cannot handle system nonlinearity, time-varying characteristics, and multi-channel coupling. The adaptive predictive controller employs a recursive least squares method with a forgetting factor to identify the controlled object model parameters online. Its objective function simultaneously optimizes output error and control increment, enabling the adaptive predictive controller to track system dynamic changes in real time and make predictive adjustments for future states, thus solving the response lag problem. The multivariable decoupling compensator, based on the feedforward compensation principle, actively and proactively cancels the dynamic coupling interference between raw material feeding channels by constructing the inverse model of the coupled channels. The fusion of these two algorithms allows the core intelligent control and data processing unit to handle the complex dynamics of a single channel while coordinating the mutual influence between multiple channels, thereby ensuring the consistency of the proportioning accuracy of each channel even under complex operating conditions.

[0079] The actuator drive unit uses a vector control frequency converter to drive the permanent magnet synchronous motor. The carrier frequency of the frequency converter is adjustable from 8 kHz to 12 kHz. The permanent magnet synchronous motor is connected to the main shaft of the twin-helix feeder through a planetary gear reducer. The reduction ratio is determined based on the ratio of the rated torque of the twin-helix feeder to the peak torque of the permanent magnet synchronous motor, with a value range of 15:1 to 25:1. The rotor position of the permanent magnet synchronous motor is fed back to the frequency converter through a 23-bit absolute encoder, forming a fully closed-loop speed control circuit.

[0080] This solution addresses the issues of slow response and low control precision in traditional drive systems. By employing a vector control frequency converter to drive the permanent magnet synchronous motor, precise decoupling control of motor torque and magnetic field is achieved, resulting in excellent speed regulation performance. Combined with a fully closed-loop speed control circuit formed by the 23-bit absolute encoder, the rotor position of the permanent magnet synchronous motor can be fed back in real time with high precision, enabling the vector control frequency converter to perform millisecond-level speed regulation. The planetary gear reducer and its specific reduction ratio range ensure that the high speed and low torque of the permanent magnet synchronous motor are converted into the low speed and high torque required by the twin-helix feeder. Together, these factors ensure that the feeding rate can quickly and accurately track control commands, achieving a millisecond-level dynamic response in the feeding rate.

[0081] Furthermore, the system status monitoring and self-calibration unit includes a vibration acceleration sensor, an acoustic emission sensor, and an infrared thermal imager. The vibration acceleration sensor is installed in the bearing housing of the twin-screw feeder, and the monitoring frequency range is 10 Hz to 1000 Hz. When the vibration amplitude exceeds the threshold for 2 consecutive seconds, an early warning is triggered. The acoustic emission sensor is arranged on the weighing sensor mounting base of the high-precision dynamic weighing and flow sensing unit to capture stress wave signals generated by microcracks in the mechanical structure. The infrared thermal imager is aimed at the motor housing and reducer surface of the actuator drive unit, with a sampling period of 5 seconds, and judges the overheating risk of the equipment by monitoring the temperature rise gradient.

[0082] By defining the specific sensors and their deployment within the system status monitoring and self-calibration unit, the problem of the "black box" mechanical status and the inability to provide early warning of faults in the high-precision aluminum smelting raw material quantitative feeding device is solved. The vibration acceleration sensor monitors the mechanical status of rotating components such as bearings of the twin-helix feeder, with a specific frequency range covering common fault characteristic frequencies. The acoustic emission sensor is highly sensitive to stress wave signals generated by the propagation of microcracks inside mechanical structures such as the weighing sensor mounting base, enabling early damage diagnosis. The infrared thermal imager monitors the temperature rise gradient on the surface of the motor housing and reducer, promptly detecting overheating risks caused by overload or poor lubrication of the actuator drive unit. These three sensors construct a comprehensive status monitoring network for the high-precision aluminum smelting raw material quantitative feeding device from different physical dimensions, significantly improving the reliability and availability of the device.

[0083] In addition, this embodiment also discloses a high-precision method for quantitative feeding of aluminum smelting raw materials, including the following steps:

[0084] Step S110: System initialization and parameter self-tuning, loading the physical property parameters of each raw material, the target feed formula and the initial values ​​of the control parameters, and executing the equipment no-load operation to calibrate the sensor zero position;

[0085] Step S120: Multimodal data synchronous acquisition and preprocessing. Raw data from weighing sensors, Coriolis mass flow meters, vibration sensors and temperature sensors are acquired in parallel. The data are then filtered by moving average and outliers are removed.

[0086] Step S130: The feed rate is adaptively controlled. Based on the deviation between the real-time flow feedback value and the target set value, the control output is calculated by the adaptive predictive controller, and independent speed control commands for each channel are generated by the multivariable decoupling compensator.

[0087] Step S140: Real-time system status assessment and fault diagnosis, comprehensively analyze vibration spectrum characteristics, acoustic emission signal energy and temperature distribution, assess the health status of mechanical components and identify potential fault modes.

[0088] Step S150: Periodic online calibration and parameter update. After every 4 hours of continuous operation, the static calibration program is automatically executed to update the sensor calibration coefficients and correct the control model parameters.

[0089] Step S160: Multi-channel coordinated proportioning control. Based on the preset raw material proportioning matrix, the feeding rate of each channel is dynamically coordinated to ensure that the total proportioning error is less than 0.3%.

[0090] The high-precision feeding process is standardized into a logically rigorous and interconnected sequence of steps. From step S110, system initialization and parameter self-tuning ensure accurate starting points; to step S120, multi-modal data synchronous acquisition and preprocessing guarantee information quality; then to step S130, adaptive control of the feeding rate, and step S140, real-time system status assessment and fault diagnosis, achieving a balance between control and safety through parallel execution; finally, step S150, periodic online calibration and parameter updates maintain long-term accuracy; and step S160, multi-channel collaborative proportioning control ensures global optimization. This addresses the shortcomings of various aspects in the background technology. Through process standardization and intelligentization, the repeatability and high efficiency of the method are ensured, ultimately achieving excellent performance in which the high-precision aluminum smelting raw material quantitative feeding device maintains long-term stable accuracy within ±0.3% and a total proportioning error of less than 0.3%.

[0091] Specifically, the data preprocessing in step S120 includes: denoising the weighing sensor signal using a 4th-order Butterworth low-pass filter with a cutoff frequency of 50 Hz; removing instantaneous outliers caused by bubble or particle impact from the density reading output by the Coriolis mass flow meter using the 3σ criterion; and performing a 1024-point fast Fourier transform on the vibration signal after windowing using a Hanning window to extract the 1st, 2nd, and 3rd harmonic amplitudes as feature quantities.

[0092] The specific algorithm for data preprocessing in step S120 solves the problem of invalid information such as noise and impact interference in the original sensor signals affecting control accuracy and fault diagnosis accuracy. A fourth-order Butterworth low-pass filter effectively filters out high-frequency mechanical vibration noise from the weighing sensor signal, retaining the true changes in feed flow rate. Outliers are removed from the density readings output by the Coriolis mass flow meter using the 3σ criterion, an effective data cleaning method based on statistical principles, preventing instantaneous erroneous data caused by bubble or particle impact from entering the core intelligent control and data processing unit. Vibration signals are processed using a Hanning window and a fast Fourier transform, accurately extracting the characteristic frequency amplitudes related to equipment faults. The preprocessing steps provide a clean and reliable data foundation for subsequent precise control and reliable diagnosis.

[0093] The adaptive predictive controller parameter tuning process in step S130 is as follows: the initial value of the forgetting factor λ is set to 0.98, and the control weight coefficient ρ is set between 0.1 and 1.0 according to the system response speed requirements; the prediction time domain N covers the main dynamic response time of the system and is set to 15 to 25 sampling periods; the control time domain M is set to 1 / 3 to 1 / 2 of the prediction time domain; and the controller output limit is set according to the maximum safe feeding rate of the twin-helix feeder.

[0094] The tuning range and strategy for the key parameters of the adaptive predictive controller described in step S130 solve the problem of blindly tuning controller parameters and failing to achieve optimal algorithm performance. Setting the initial value of the forgetting factor λ to 0.98 achieves a good balance between tracking system dynamic changes and suppressing data noise. The prediction time domain N covers the main dynamic response time of the system, with a value ranging from 15 to 25 sampling periods. The control time domain M is taken as 1 / 3 to 1 / 2 of the prediction time domain, ensuring the foresight and computational efficiency of the control optimization. The control weight coefficient ρ is set between 0.1 and 1.0 according to the system response speed requirements, achieving a flexible trade-off between control strength and system stability. The controller output limit is set according to the maximum safe feeding rate of the twin-screw feeder, directly related to equipment safety and preventing overshoot damage to the equipment.

[0095] Furthermore, the online calibration procedure in step S150 includes two stages: static weight calibration and dynamic material calibration. Static weight calibration is performed under no-load conditions on the twin-screw feeder, where five standard weights are applied sequentially, covering 20%, 40%, 60%, 80%, and 100% of the weighing range. The sensor output is recorded, and a linear regression equation is fitted. Dynamic material calibration uses standard materials of known mass and feeds them at three different feeding rates. The cumulative feeding mass is compared with the weighing sensor readings to correct the system's dynamic error coefficient.

[0096] The online calibration procedure in step S150 solves the technical problems of sensor calibration coefficient drift and system dynamic error accumulation. The static weight calibration, by applying five standard weights covering 20% ​​to 100% of the weighing range under no-load conditions on the twin-screw feeder and fitting a linear regression equation, can accurately calibrate the static characteristics of the weighing sensor and correct its linearity error. The dynamic material calibration uses standard materials of known mass for verification during actual feeding. By comparing the cumulative feeding mass with the weighing sensor reading, the system dynamic error coefficient caused by factors such as material impact can be identified and corrected. This combined static and dynamic calibration strategy forms a complete accuracy assurance closed loop, ensuring the long-term metering accuracy of the high-precision aluminum smelting raw material quantitative feeding device across the entire weighing range and at various operating rates.

[0097] To facilitate a better understanding of the present invention by those skilled in the art, the present invention will be further described below in conjunction with specific implementation processes.

[0098] This document details the fusion and collaborative operation of the control algorithms for a high-precision aluminum smelting raw material quantitative feeding device. Specifically, it describes the fusion and collaborative operation of the Adaptive Predictive Controller (APC) and Multivariable Decoupling Compensator (MDC) within the core intelligent control and data processing unit during the device's operation. Two feeding channels (Channel 1 and Channel 2) are preset, each corresponding to an independent hopper and a twin-screw feeder. The following steps are executed at sampling time k, with the algorithms interacting via data flow.

[0099] 1. Data Acquisition and Preprocessing:

[0100] Objective: To acquire raw data from sensors, perform filtering, outlier removal, and feature extraction to provide cleaned data for model parameter updates and control calculations.

[0101] Sensor data:

[0102] The load cell outputs the raw mass flow rate signal. and (Unit: kg / s).

[0103] Coriolis mass flow meter outputs raw density reading (Unit: kg / m³).

[0104] The vibration acceleration sensor outputs the raw vibration signal. (Unit: m / s²)

[0105] Preprocessing steps:

[0106] Mass flow signal filtering: and A fourth-order Butterworth low-pass filter with a cutoff frequency of 50 Hz is applied to obtain the filtered signal. .

[0107] Filter transfer function:

[0108] ;

[0109] Among them, coefficient Based on the Butterworth design method, the cutoff frequency is determined to be 50 Hz and the sampling frequency is 1 kHz.

[0110] Output: .

[0111] Outlier removal: For Apply the 3σ criterion to calculate the mean within the window (e.g., 10 sampling points). and standard deviation Eliminate those that meet the requirements The points were then replaced with linear interpolation to obtain the mass flow rate after cleaning. .

[0112] Density reading processing: Similarly, the 3σ criterion is applied to remove outliers, resulting in the cleaned density. .

[0113] Vibration feature extraction: for After applying the Hanning window, a 1024-point Fast Fourier Transform (FFT) is performed to extract the amplitude values ​​of the first, second, and third harmonics. (Unit: m / s²), used for condition monitoring.

[0114] Hanning window function:

[0115] ,in, .

[0116] FFT output: .

[0117] Feature extraction: ,in These are the 1st, 2nd, and 3rd harmonic frequencies.

[0118] Pretreatment output: Mass flow rate after cleaning Used for control; vibration characteristics Used for condition monitoring.

[0119] 2. Model parameter update (Recursive Least Squares, RLS);

[0120] Objective: To estimate the parameters of a multivariate ARX model online, providing model parameters for APC and MDC. The model parameters are updated based on preprocessed data, reflecting adaptive characteristics.

[0121] Model building:

[0122] The system dynamics are described using a multivariate ARX model:

[0123]

[0124]

[0125] in,

[0126] For the multivariate ARX model parameters of the i-th output channel, the k-th lag term (unitless, representing the coupling relationship between outputs) corresponds to the j-th output variable. For the multivariate ARX model parameters of the i-th output channel, the k-th lag term corresponding to the j-th control input (unitless, representing the influence coefficient of the input on the output) (e.g.: Output for channel 1 For its own first-order lag term coefficient, Output for channel 1 Control input for channel 2 ; The instantaneous mass flow rate of the i-th raw material channel is derived from the preprocessed data of the high-precision dynamic weighing unit. This is the control input for the i-th channel, i.e., the speed setting value of the permanent magnet synchronous motor of the actuator drive unit; The modeling error for the i-th channel is caused by measurement noise, material property fluctuations, etc., and follows a normal distribution. .

[0127] Meaning of parameter subscripts:

[0128] for equation, yes coefficient, yes coefficient, coefficient, yes The coefficient;

[0129] for equation, yes coefficient, coefficient, coefficient, yes The coefficient.

[0130] Parameter vector definition:

[0131] For channel 1: .

[0132] For channel 2: .

[0133] Definition of regression vector:

[0134] .

[0135] .

[0136] Model training (RLS algorithm):

[0137] For each channel Update parameters using RLS with a forgetting factor:

[0138] ;

[0139] ;

[0140] ;

[0141] ;

[0142] in,

[0143] In the formula, The model parameter vector for the i-th channel (containing Parameters, dimension n is determined by the model order, and the parameter estimate at time k).

[0144] Let i be the RLS gain vector of the i-th channel (not a gain matrix, with dimensions equal to...). Consistent);

[0145] The parameter covariance matrix of the i-th channel (initial value) (where I is an n-order identity matrix, and the initial value is set to a large matrix to ensure flexibility in the initial parameter estimation).

[0146] Let be the forgetting factor for the i-th channel (initial value 0.98).

[0147] The RLS residual of the i-th channel (unitless, the difference between the actual output and the model's predicted output);

[0148] Let i be the regression vector for the i-th channel, specifically in the form of: (Includes all output lag terms and input lag terms).

[0149] Interactive processing: Forgetting factor Adjustments can be made based on condition monitoring (e.g., vibration characteristics). When the threshold is exceeded, decrease To accelerate adaptation).

[0150] Model Application:

[0151] Output estimated parameters: and .

[0152] These parameters are used for APC control calculations and MDC decoupling compensation.

[0153] 3. Adaptive Predictive Controller (APC) Control Calculation:

[0154] Objective: To calculate control variables based on a univariate prediction model. This allows the feed rate to track the reference value. APC uses parameters estimated by RLS to demonstrate its adaptive characteristics.

[0155] Model building:

[0156] For each channel i, a univariate ARX model is used for prediction:

[0157] ;

[0158] in:

[0159] Channel i in Predicted output at time step.

[0160] : Control quantity calculated by APC (before decoupling).

[0161] and From RLS estimation: for channel 1, For channel 2, (Note: For channel 2, the univariate model is used) and (i.e., the coefficients of its own output and input).

[0162] Control objective function:

[0163] ;

[0164] in:

[0165] Reference feed rate (from target formulation);

[0166] : Control weighting coefficient, set between 0.1 and 1.0 according to system response requirements (based on the manuscript).

[0167] Control quantity calculation:

[0168] minimize ,right Take the derivative and set it to zero:

[0169] ;

[0170] Solving for:

[0171] ;

[0172] Interactive processing: APC uses RLS output, where and , This directly affects the calculation of control quantities. If... Then use the control quantity from the previous moment. As a backup.

[0173] Output: APC control input and It is passed to MDC for decoupling.

[0174] 4. Multivariable Decoupling Compensator (MDC) for decoupling compensation:

[0175] Objective: To compensate for dynamic coupling between channels and make the control variables of each channel independent. MDC uses coupling parameters estimated by RLS to reflect the integration with APC.

[0176] Model building:

[0177] Based on the multivariate ARX model, the coupling effect is determined by... Parameter descriptions, etc.

[0178] Decoupling coefficient calculation:

[0179] ;

[0180] in,

[0181] The decoupling coefficient (unitless, characterizing the dynamic coupling strength between the two channels) is the factor by which the i-th channel is affected by the control input of the j-th channel; for example: For the second channel control input Output to channel 1 The coupling coefficient, For the control input of channel 1 Output to channel 2 Coupling coefficient; when When this occurs, it indicates that the j-th channel has no coupling effect on the i-th channel.

[0182] For channel 1 equation The coefficient of ) (from );

[0183] For channel 1 equation coefficient (from) );

[0184] For Channel 2 equation coefficient (from) );

[0185] For Channel 2 equation coefficient (from) ).

[0186] Decoupling control quantity calculation:

[0187] Calculate the actual control quantity based on the feedforward compensation principle. and :

[0188] ;

[0189] ;

[0190] Interactive processing: MDC uses APC output. and and RLS output If the denominator Then use As a backup.

[0191] Output: Decoupled control quantity and Send to the actuator drive unit.

[0192] 5. Control Execution:

[0193] Objective: To convert the decoupled control quantity into a motor speed setpoint to drive the twin-screw feeder.

[0194] Execution process:

[0195] and The set value is sent to the vector control frequency converter to drive the permanent magnet synchronous motor.

[0196] The motor is connected to the main shaft of the twin-helix feeder via a planetary gear reducer to adjust the feeding rate.

[0197] The encoder provides feedback on the rotor position, forming a closed-loop control.

[0198] 6. System status monitoring and self-calibration:

[0199] Objective: To monitor the system status based on multi-sensor data, trigger a self-calibration program, and ensure long-term accuracy.

[0200] Status monitoring:

[0201] Using vibration characteristics Calculate health indicators :

[0202]

[0203] if If the speed exceeds the threshold (e.g., 10 m / s²) for two consecutive seconds, an alert will be triggered.

[0204] The motor temperature is monitored using an infrared thermal imager. If the temperature rise gradient exceeds 5°C / s, an overheat warning is triggered.

[0205] Self-correction:

[0206] After every 4 hours of continuous operation, execute the online calibration procedure:

[0207] Static weight calibration: With the twin-screw feeder under no-load conditions, apply five standard weights (20%, 40%, 60%, 80%, and 100% of the range), record the sensor output, and fit a linear regression equation: ,

[0208] in, The output signal of the i-th channel load cell corresponds to the electrical signal conversion value of the load.

[0209] The load of the standard weight;

[0210] The calibration coefficient of the i-th channel represents the change in sensor output per unit load;

[0211] is the zero-point offset of the i-th channel.

[0212] Dynamic material calibration: Using standard materials of known mass, actual feeding is performed at three feeding rates (low speed, medium speed, and high speed). The cumulative feeding mass is compared with the weighing sensor reading, and the dynamic error coefficient is corrected. .

[0213] Model parameter reset: After calibration, reset the RLS parameters. Covariance Matrix Initial values ​​to accommodate the new calibration.

[0214] Algorithm Fusion Summary

[0215] Data flow interaction:

[0216] Preprocessed output Used for RLS model updates.

[0217] RLS output Used for APC control calculations and MDC decoupling coefficient calculations.

[0218] APC output Used for MDC decoupling compensation.

[0219] MDC output Used to control execution and fed back to the RLS regression vector.

[0220] Model training and application:

[0221] RLS is used to train a multivariate model online, which is then applied to APC and MDC for control and decoupling.

[0222] State monitoring adjusts the RLS forgetting factor to reflect adaptive fusion.

[0223] Synergistic effect: Through algorithm fusion, high-precision feeding is achieved (total proportion error is less than 0.3%), and the channel coupling problem is overcome.

[0224] This invention decomposes the quantitative feeding process of aluminum smelting raw materials into basic technical elements such as material flow stability, measurement accuracy, dynamic coupling of control, and reliability. Through a multi-channel raw material storage and pretreatment unit, the cone angle of the silo and the high-frequency pneumatic anti-bridging device are optimized based on the principles of material mechanics, fundamentally eliminating bridging phenomena. The variable-pitch twin-screw feeder, based on solid conveying dynamics, ensures uniform material output. A high-precision dynamic weighing and flow sensing unit integrates strain measurement and the Coriolis principle to achieve redundant detection and real-time temperature compensation, improving data reliability. The core intelligent control and data processing unit employs adaptive predictive control and a multivariable decoupling compensator, based on identification and feedforward compensation principles, overcoming nonlinearity and time-varying coupling. The actuator drive unit ensures millisecond-level dynamic response through fully closed-loop vector control. The system status monitoring and self-calibration unit integrates multi-physical quantity sensing and periodic calibration to maintain long-term accuracy from a preventative maintenance perspective.

[0225] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0226] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-precision quantitative feeding device for aluminum smelting raw materials, characterized in that, include: A multi-channel raw material storage and pretreatment unit is used to store raw materials and perform physical pretreatment on the raw materials. The multi-channel raw material storage and pretreatment unit includes at least 3 independent silos, and each silo outlet is connected to a twin-screw feeder. A high-precision dynamic weighing and flow sensing unit is connected to the output end of the twin-screw feeder to detect and output the instantaneous mass flow rate and cumulative feed amount of each raw material channel in real time. The core intelligent control and data processing unit is communicatively connected to the high-precision dynamic weighing and flow sensing unit, and is used to receive sensing data, execute control algorithms and output control commands. The actuator drive unit receives instructions from the core intelligent control and data processing unit and drives the twin-helix feeder to precisely adjust the feeding rate. The system status monitoring and self-correction unit is integrated into key nodes to continuously monitor operating parameters and trigger self-correction procedures; The high-precision dynamic weighing and flow sensing unit adopts a combination of a strain-based weighing sensor and a Coriolis mass flow meter; the weighing sensor is arranged at the support structure of the twin-screw feeder and has a built-in temperature compensation module with a compensation coefficient of: ; in, This is the temperature compensation coefficient for the weighing sensor, used to correct the impact of temperature changes on measurement accuracy. This represents the change in resistance of the sensor strain gauge caused by temperature changes. The initial resistance of the strain gauge in the sensor; This refers to the change in ambient temperature. The Coriolis mass flow meter is installed in the downstream pipeline of the feeder to output material density parameters in real time. The core intelligent control and data processing unit integrates an adaptive predictive controller and a multivariable decoupling compensator; the adaptive predictive controller uses a recursive least squares method with a forgetting factor to identify the model parameters of the controlled object online. The multivariable decoupling compensator is based on the feedforward compensation principle and uses the inverse model of the coupling channel to counteract the dynamic coupling effect between the raw material feeding channels.

2. The high-precision aluminum smelting raw material quantitative feeding device according to claim 1, characterized in that: In the multi-channel raw material storage and pretreatment unit, the inner wall of each silo is lined with ultra-high molecular weight polyethylene wear-resistant lining plate. The main and auxiliary screws of the twin-screw feeder adopt a variable pitch design. The pitch of the main screw increases along the discharge direction, while the pitch of the auxiliary screw is constant. The speed ratio between the two is controlled within the range of 1:1.5 to 1:2.

2.

3. The high-precision aluminum smelting raw material quantitative feeding device according to claim 1, characterized in that: The actuator drive unit uses a vector control frequency converter to drive the permanent magnet synchronous motor. The carrier frequency of the frequency converter is adjustable from 8 kHz to 12 kHz. The permanent magnet synchronous motor is connected to the main shaft of the twin-helix feeder through a planetary gear reducer. The reduction ratio is determined based on the ratio of the rated torque of the twin-helix feeder to the peak torque of the permanent magnet synchronous motor, with a value range of 15:1 to 25:

1. The rotor position of the permanent magnet synchronous motor is fed back to the frequency converter through a 23-bit absolute encoder, forming a fully closed-loop speed control loop.

4. The high-precision aluminum smelting raw material quantitative feeding device according to claim 1, characterized in that: The system status monitoring and self-correction unit includes a vibration acceleration sensor, an acoustic emission sensor, and an infrared thermal imager. The vibration acceleration sensor is installed in the bearing housing of the twin-screw feeder, and the monitoring frequency range is 10 Hz to 1000 Hz. When the vibration amplitude exceeds the threshold for 2 consecutive seconds, an early warning is triggered. The acoustic emission sensor is arranged on the weighing sensor mounting base of the high-precision dynamic weighing and flow sensing unit to capture stress wave signals generated by microcracks in the mechanical structure. The infrared thermal imager is aimed at the motor housing and reducer surface of the actuator drive unit, with a sampling period of 5 seconds, and the risk of equipment overheating is determined by monitoring the temperature rise gradient.

5. A high-precision method for quantitative feeding of raw materials in aluminum smelting, characterized in that, The method includes using the high-precision aluminum smelting raw material quantitative feeding device as described in claim 4, specifically comprising the following steps: Step S110: System initialization and parameter self-tuning, loading the physical property parameters of each raw material, the target feed formula and the initial values ​​of the control parameters, and executing the equipment no-load operation to calibrate the sensor zero position; Step S120: Multimodal data synchronous acquisition and preprocessing. Raw data from weighing sensors, Coriolis mass flow meters, vibration acceleration sensors and infrared thermal imagers are acquired in parallel. The data are then filtered by moving average and outliers are removed. Step S130: The feeding rate is adaptively controlled. Based on the deviation between the real-time flow feedback value and the target set value, the control output is calculated by the adaptive predictive controller, and independent speed control commands for each channel are generated by the multivariable decoupling compensator. Step S140: Real-time system status assessment and fault diagnosis, comprehensively analyze vibration spectrum characteristics, acoustic emission signal energy and temperature distribution, assess the health status of mechanical components and identify potential fault modes; Step S150: Periodic online calibration and parameter update. After every 4 hours of continuous operation, the static calibration program is automatically executed to update the sensor calibration coefficients and correct the control model parameters. Step S160: Multi-channel coordinated proportioning control, dynamically coordinating the feeding rate of each channel according to the preset raw material proportioning matrix.

6. The high-precision quantitative feeding method for aluminum smelting raw materials according to claim 5, characterized in that, The data preprocessing in step S120 specifically includes: using a 4th-order Butterworth low-pass filter with a cutoff frequency of 50 Hz to reduce noise in the weighing sensor signal; using the 3σ criterion to remove instantaneous abnormal values ​​caused by bubble or particle impact in the density reading output by the Coriolis mass flow meter; and performing a 1024-point fast Fourier transform on the vibration signal after windowing through a Hanning window to extract the amplitude values ​​of the 1st, 2nd, and 3rd harmonics as feature quantities.

7. The high-precision quantitative feeding method for aluminum smelting raw materials according to claim 5, characterized in that: The adaptive predictive controller parameter tuning process in step S130 is as follows: the initial value of the forgetting factor λ is set to 0.98, and the control weight coefficient ρ is set between 0.1 and 1.0 according to the system response speed requirements; the prediction time domain N covers the main dynamic response time of the system and is set to 15 to 25 sampling periods; the control time domain M is set to 1 / 3 to 1 / 2 of the prediction time domain; and the controller output limit is set according to the maximum safe feeding rate of the twin-helix feeder.

8. The high-precision quantitative feeding method for aluminum smelting raw materials according to claim 5, characterized in that: Step S150 includes two stages: static weight calibration and dynamic material calibration. Static weight calibration is performed under no-load conditions on the twin-screw feeder. Five standard weights are applied sequentially, covering 20%, 40%, 60%, 80%, and 100% of the weighing range. The sensor output is recorded and a linear regression equation is fitted. Dynamic material calibration uses standard materials of known mass and feeds them at three different feeding rates. The cumulative feeding mass is compared with the weighing sensor readings to correct the system's dynamic error coefficient.

Images