Method and apparatus for heating control of aluminum dross for aluminum refining

Abstract

The present application relates to the technical field of aluminum refining control, and particularly relates to an aluminum ash heating control method and device for refining aluminum. The present application periodically calibrates the heat conduction delay duration of aluminum ash; in each control cycle, the heating power variation before and after the superimposed random power disturbance of the heating module is obtained, the delay torque variation of the stirring motor is extracted after the control cycle is delayed by the heat conduction delay duration, and the power-torque variation ratio is calculated; the resistance evolution parameter and the resistance disorder parameter after each control cycle are obtained, so as to determine the steady-state heat absorption softening stage and determine the disorder safety benchmark, and then after the current control cycle ends, it is determined whether the aluminum ash has abnormal sintering to adjust the heating. The present application eliminates the aluminum ash thermal hysteresis misplacement by calibrating the heat conduction delay and the random power disturbance; in combination with the macroscopic resistance evolution and the microscopic law characteristics, the normal cold charge and the abnormal sintering are accurately distinguished to perform accurate intervention, and the aluminum ash heating refining control effect is improved.

Description

Technical Field

[0001] This invention relates to the field of aluminum refining control technology, and specifically to a method and apparatus for controlling the heating of aluminum ash used in aluminum refining. Background Technology

[0002] In the pyrometallurgical refining of industrial aluminum ash, external heating combined with internal drum stirring is typically used to liquefy metallic aluminum and separate it from solid residues. Simultaneously, internal blade stirring ensures uniform reaction. Precise heating control is crucial for ensuring aluminum refining efficiency, energy economy, and safe equipment operation. Traditional heating control methods usually rely on fixed power-time curves or passive protection based on torque upper limit thresholds.

[0003] However, bulk aluminum ash has poor thermal conductivity and extremely high thermal inertia. There is a significant lag in the transfer of heat energy from external heating to the interior of the material. This makes it difficult for changes in the stirring resistance (stirring motor torque), which reflects the rheological state of the material, to correspond with the heating process in real time. As a result, it is impossible to perceive the material's true thermal response to heating in real time for dynamic control, which can easily lead to underheating or overheating. At the same time, normal tumbling friction of unmelted material or abnormal sintering of aluminum ash during stirring can cause abnormal increases in stirring motor torque. It is difficult to distinguish between the two, which may lead to frequent false alarms that reduce refining efficiency, or cause the reducer to seize and be damaged due to alarm lag. Summary of the Invention

[0004] To address the problem of poor control over the heating and refining effect of aluminum ash in existing technologies, the present invention aims to provide a method and apparatus for controlling the heating of aluminum ash for aluminum refining. The specific technical solution adopted is as follows: A method for controlling the heating of aluminum ash used in aluminum refining, the method comprising: During the refining and heating process, the thermal conduction delay time of aluminum ash is calibrated periodically; within each control cycle, the change in heating power before and after the heating module is superimposed with random power disturbance is obtained, and the change in delay torque of the stirring motor is extracted after the control cycle is extended by the thermal conduction delay time. The power-torque change ratio is calculated based on the change in heating power and the change in delay torque. After each control cycle, the power-torque change ratio sequence is updated using a sliding window. The drag evolution parameters are obtained based on the macroscopic distribution characteristics of the power-torque change ratio sequence, and the drag disorder parameters are obtained based on the structural disorder characteristics of the power-torque change ratio sequence. Based on the resistance evolution parameters, the steady-state endothermic softening stage in the refining heating process is determined, and the disorder safety benchmark is determined based on the resistance disorder parameters within the steady-state endothermic softening stage. After the current control cycle ends, the resistance disorder parameters and the disorder safety benchmark are compared, and combined with the resistance evolution parameters, it is determined whether there is abnormal sintering of aluminum ash, and the heating power of the heating module is adjusted based on the determination results.

[0005] Furthermore, methods for periodically calibrating the thermal conductivity delay time of aluminum ash include: From the start of refining and heating, a calibration cycle with a preset dynamic duration is continuously determined. Within each calibration cycle, the planned heating power at the start time is superimposed with a preset step power and heating is controlled. At the same time, timing is started to monitor the torque of the stirring motor in real time within the calibration cycle. When the torque change relative to the starting torque exceeds the preset torque threshold, timing is stopped, and the timing duration is used as the heat conduction delay duration.

[0006] Furthermore, the method for obtaining the change in heating power includes: Within each control cycle, a limited random power increment is generated. The planned heating power at each moment is superimposed with the random power increment and heating is controlled. The heating power of the heating module at the beginning and end of the control cycle is collected and the difference is calculated to obtain the change in heating power.

[0007] Furthermore, the method for obtaining the change in delayed torque includes: For each control cycle, determine the heat conduction delay time calibrated within its calibration cycle, and extend the heat conduction delay time backward along the time axis at the first and last two moments respectively to obtain the corresponding extension period of the control cycle. Collect the torque of the stirring motor at the first and last moments of the extension period and calculate the difference to obtain the change in delayed torque.

[0008] Furthermore, the method for obtaining the drag evolution parameters includes: The average value of the power-torque change ratio in the power-torque change ratio sequence is used as the drag evolution parameter.

[0009] Furthermore, the method for obtaining the resistance disorder parameter includes: In the power-torque change ratio sequence, all basic sub-vectors with a length of a preset embedding dimension and all upgraded sub-vectors with a length of a preset embedding dimension plus 1 are extracted respectively, and the similarity tolerance is determined based on the sequence fluctuation characteristics. Based on the differences between different basic sub-vectors and in conjunction with the similarity tolerance, the basic matching probability under the preset embedding dimension is obtained; based on the differences between different upgraded sub-vectors and in conjunction with the similarity tolerance, the upgraded matching probability under the preset embedding dimension plus 1 is obtained; the resistance disorder parameter is determined based on the basic matching probability and the upgraded matching probability.

[0010] Furthermore, the method for obtaining the steady-state endothermic softening stage includes: Since the start of refining and heating, the resistance evolution parameter obtained after each control cycle is continuously monitored. When the number of times the resistance evolution parameter is continuously less than the preset evolution threshold first reaches the preset number threshold, the corresponding time period of all control cycles corresponding to the resistance evolution parameter that has continuously reached the preset number threshold is extracted as the steady-state endothermic softening stage.

[0011] Furthermore, the method for obtaining the disordered security benchmark includes: Extract the resistance disorder parameters obtained after all control cycles during the steady-state endothermic softening stage. Determine the initial disorder safety benchmark based on the distribution characteristics of all resistance disorder parameters. Use preset margin weights to weight the initial disorder safety benchmark to obtain the disorder safety benchmark.

[0012] Furthermore, methods for determining whether abnormal sintering exists in aluminum ash include: If, after the current control cycle ends, the resistance disorder parameter is greater than the disorder safety benchmark and the resistance evolution parameter is greater than the preset evolution threshold, it is determined that the aluminum ash has abnormal sintering.

[0013] A heating control device for aluminum ash refining includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the heating control method for aluminum ash refining.

[0014] The present invention has the following beneficial effects: In the refining and heating process, this invention periodically calibrates the thermal conduction delay time of aluminum ash, providing a basis for subsequent alignment and matching of heating power and torque to analyze material phase changes. Within each control cycle, the change in heating power before and after a random power disturbance in the heating module is acquired, and the control cycle is extended by the thermal conduction delay time to match causal alignment, thereby extracting the change in delayed torque of the stirring motor. This allows for the calculation of the power-torque change ratio, characterizing the sensitivity of material rheological resistance changes to heat. Then, after each control cycle, a sliding window is used to update and construct the power-torque change ratio sequence, and a macro-analysis based on the power-torque change ratio sequence is performed. This invention obtains resistance evolution parameters characterizing macroscopic rheological changes in materials by observing distribution characteristics, and obtains resistance disorder parameters characterizing the presence of abnormal agglomeration impacts based on the structural disorder characteristics of the power-torque change ratio sequence. Then, based on the resistance evolution parameters, a steady-state endothermic softening stage is determined during the refining heating process, and a disorder safety benchmark is adaptively determined based on the resistance disorder parameters within this stage. After the current control cycle ends, the resistance disorder parameters are compared with the disorder safety benchmark, and combined with the resistance evolution parameters, it is determined whether abnormal sintering exists in the aluminum ash. Based on the determination result, the heating power of the heating module is adjusted. This invention eliminates thermal hysteresis misalignment of aluminum ash by calibrating thermal conduction delay and random power disturbances; by combining macroscopic resistance evolution and microscopic regularity characteristics, it accurately distinguishes between normal cold materials and abnormal sintering, enabling precise intervention in anomalies and improving the control effect of aluminum ash heating and refining. Attached Figure Description

[0015] Figure 1 This is a flowchart of a heating control method for aluminum ash used in aluminum refining, provided as an embodiment of the present invention. Detailed Implementation

[0016] The following description, in conjunction with the accompanying drawings, details a specific scheme for a heating control method and apparatus for aluminum ash refining provided by the present invention.

[0017] It should be noted that, in the embodiments of the present invention, the device for refining aluminum heats according to a set program (such as planned heating power), and the heating is adjusted accordingly.

[0018] Please see Figure 1 The diagram illustrates a flowchart of a heating control method for aluminum ash used in aluminum refining, provided by an embodiment of the present invention, specifically including: Step S1: During the refining and heating process, the thermal conduction delay time of aluminum ash is calibrated periodically; within each control cycle, the change in heating power before and after the heating module is superimposed with random power disturbance is obtained, and the change in delayed torque of the stirring motor is extracted after the control cycle is extended by the thermal conduction delay time. The power-torque change ratio is calculated based on the change in heating power and the change in delayed torque.

[0019] Since aluminum ash is a very poor thermal conductor with severe heat conduction lag, calibrating the thermal conduction delay time of aluminum ash can bind the current power disturbance with the lag torque response, thereby eliminating data misalignment caused by thermal inertia. Furthermore, since the density, viscosity, and thermal conductivity of aluminum ash change continuously with the melting progress during refining, and the aluminum content, moisture content, and ambient temperature difference (such as the initial cold furnace and the hot furnace after operation) of different batches of aluminum ash have a great impact on the heat conduction speed, the thermal conduction delay time may be different at different stages. Based on this, the embodiments of the present invention will periodically update the thermal conduction delay time of the calibrated aluminum ash during the refining and heating process. The thermal conduction delay time refers to the time difference between the change in heating power and the actual change in stirring torque caused by the heating of aluminum ash, providing a basis for subsequent alignment and matching of heating power and torque to analyze material phase change.

[0020] Preferably, in one embodiment of the present invention, the calibration period during the refining and heating process can be dynamically divided for periodic calibration. Considering that within each calibration period, a step power is superimposed on the planned heating power in the program to generate a strong excitation signal, the strong heat input will change the viscous state of the aluminum ash, thereby changing the torque of the stirring motor, by setting a preset torque threshold that characterizes the conduction of external heat energy to the interior of the aluminum ash and changes its rheological state, it is helpful to evaluate whether the heating power induces a torque response (rheological state of the aluminum ash), and thus determine the thermal conduction delay time of the aluminum ash; the method for periodically calibrating the thermal conduction delay time of the aluminum ash includes: From the start of refining and heating, a calibration cycle with a preset dynamic duration is continuously determined. Within each calibration cycle, the planned heating power at the start time is superimposed with a preset step power and heating is controlled. At the same time, timing is started to monitor the torque of the stirring motor in real time within the calibration cycle. When the torque change relative to the starting torque exceeds the preset torque threshold, timing is stopped, and the timing duration is used as the heat conduction delay duration.

[0021] Specifically, the preset dynamic duration is set to 15-30 minutes, and implementers can adjust it according to actual application. In this example, a calibration cycle is set every 15 minutes from the start of refining and heating. Alternatively, the first calibration cycle can be set to 15 minutes, and the calibration cycle can be smoothly extended to 30 minutes as the aluminum ash gradually melts in the middle and later stages. Taking any calibration cycle as an example, firstly extract the planned heating power in the setting program; at the beginning of the calibration cycle, read the starting torque of the stirring motor, start the timer, and add a preset step power to the planned heating power so that the heating module heats according to the added heating power, and then monitor the torque of the stirring motor in real time during the calibration cycle. A preset torque threshold is set. When the absolute value of the difference between the monitored torque and the initial torque (i.e., the amount of torque change) exceeds the preset torque threshold for the first time, it indicates that the step heat has been conducted into the aluminum ash and substantially changed the rheological resistance of the aluminum ash. At this time, the timer is stopped immediately, and the timing duration is fixed as the heat conduction delay duration. If the amount of torque change fails to exceed the preset torque threshold within the calibration period, the timer will not be able to stop. In this case, the corresponding duration of the calibration period or the heat conduction delay duration calibrated in the previous calibration period is directly used as the heat conduction delay duration of the current calibration period. The preset step power is set to 10-15% of the full scale of the heating module to provide instantaneous step heat sufficient to penetrate poor thermal conductors (aluminum ash); in this example, it is set to 10% of the full scale power. The preset torque threshold is set to 1-2% of the full scale of the torque sensor on the stirring motor to filter out mechanical background noise generated by gear meshing and normal tumbling, while helping to assess changes in rheological resistance; in this example, it is set to 1% of the full scale torque.

[0022] It should be noted that the preset step power is superimposed until the timer stops in order to maintain a sufficient thermal potential difference to overcome the thermal resistance of the aluminum ash, that is, to cover the entire process from heat source emission to heat penetration causing changes in rheological resistance; after the timer stops, it is immediately canceled, and the subsequent control heating module heats according to the planned heating power.

[0023] Since refining and heating is a long-term continuous reaction process, this embodiment of the invention further discretizes the entire refining and heating process into several single event units to prepare for subsequent analysis.

[0024] Specifically, the duration of the control cycle is set to 15-30 seconds, which can be adjusted by the implementer according to the actual application. In this example, a control cycle is set every 30 seconds starting from the refining heating. The lower limit of the control cycle duration can ensure that subsequent random small power disturbances can be responded to and accurately measured by the underlying hardware (such as contactors, thyristors and sensors) in a steady state, avoiding response dead zones. The upper limit of the control cycle duration can limit the disturbance dwell time, prevent small temperature fluctuations from accumulating excessive interference with the normal temperature control, and ensure that the control system has a sufficient refresh frequency to capture sudden aluminothermic reaction agglomeration.

[0025] Since the planned heating power changes little within a short control period, in order to improve the sensitivity to changes in the rheological resistance of aluminum ash, a positive power disturbance is superimposed to help assess torque changes, thereby indirectly assessing changes in rheological resistance; while by superimposing a random power disturbance, mechanical resonance of the equipment induced by periodic fluctuations can be avoided. Furthermore, due to the lag in heat conduction, the heat from the heating power command issued within the control cycle cannot instantly penetrate the aluminum ash to reach the core of the material and change the rheological resistance. However, by extending the control cycle by the heat conduction delay time, the effects of physical conduction lag can be eliminated, thereby accurately matching (causal alignment) the power disturbance and torque response in the time dimension. Based on this, the embodiments of the present invention further obtain the change in heating power of the heating module before and after the superimposed random power disturbance in each control cycle, and extract the change in the delayed torque of the stirring motor after extending the control cycle by the heat conduction delay time. The change in heating power can characterize the actual injected thermal excitation (planned heating power superimposed with random power) within the control cycle. The change in delayed torque can help assess the actual torque change caused by the injected thermal excitation (indirectly characterizing the change in rheological resistance) through delay alignment, thus preparing for the subsequent evaluation of the power-torque ratio, which characterizes the sensitivity of material rheological resistance changes to heat.

[0026] Preferably, in one embodiment of the present invention, the method for obtaining the change in heating power includes: Within each control cycle, a limited random power increment is generated. The planned heating power at each moment is superimposed with the random power increment and heating is controlled. The heating power of the heating module at the beginning and end of the control cycle is collected and the difference is calculated to obtain the change in heating power.

[0027] Specifically, taking any control cycle as an example, a small random power increment (with a positive sign) is generated using a random number algorithm. The random power increment is limited to 2-5% of the full-scale power to avoid affecting the heating process. The actual heating power of the heating module is read in real time (which may deviate from the planned heating power). Then, the planned heating power at each moment is superimposed with the random power increment, and the heating module is heated in real time according to the superimposed heating power. The actual heating power of the heating module at the first and last moments of the control cycle is collected, and the difference between the actual heating power at the last moment and the actual heating power at the first moment is calculated. This difference is used as the change in heating power.

[0028] In another embodiment of the present invention, the actual heating power in each control cycle can be directly collected and the average value can be calculated. Then, the difference between the average actual heating power of each control cycle and the average actual heating power of the previous control cycle can be used as the change in heating power.

[0029] Preferably, in one embodiment of the present invention, the method for obtaining the change in delayed torque includes: For each control cycle, determine the heat conduction delay time calibrated within its calibration cycle, and extend the heat conduction delay time backward along the time axis at the first and last two moments respectively to obtain the corresponding extension period of the control cycle. Collect the torque of the stirring motor at the first and last moments of the extension period and calculate the difference to obtain the change in delayed torque.

[0030] Specifically, taking any control cycle as an example, firstly, the calibration cycle to which the control cycle belongs is determined, thereby extracting the heat conduction delay time calibrated within the calibration cycle. In order to prevent the extreme case of a control cycle spanning two calibration cycles, it is also necessary to perform a determination: for example, based on the proportion of the control cycle's duration within the calibration cycle, it is determined that the control cycle belongs to the calibration cycle with a larger proportion of duration. The heat conduction delay time of the corresponding calibration cycle is extended sequentially along the time sequence of the first and last two moments of the control cycle, thereby obtaining the extension period corresponding to the control cycle; the actual torque of the stirring motor at the first and last moments of the control cycle is collected, and the difference between the actual torque at the last moment and the actual torque at the first moment is calculated, and this difference is used as the change in delay torque.

[0031] The power-torque ratio is further calculated based on the change in heating power and the change in delayed torque.

[0032] Specifically, the change in delayed torque within each control cycle is directly used as the numerator, and the change in heating power is used as the denominator. To avoid division by zero due to a denominator of 0, an effective threshold for power change needs to be set first. In this example, it is 0.5% of the full-scale power. When the absolute value of the change in heating power is greater than or equal to the effective threshold for power change, the denominator is the change in heating power, and the sign of the change in heating power is retained (representing the direction of physical evolution). When the absolute value of the change in heating power is less than the effective threshold for power change, the denominator is the effective threshold for power change, to avoid the evaluation failure caused by an extremely small denominator when the denominator is signed.

[0033] It should be noted that the dimension of torque is N·m (Newton-meter), and the dimension of heating power is W (watt, i.e., J / s or N·m / s). Therefore, the dimension of the power-torque change ratio is s (second), which is used to quantify the sensitivity of material rheological resistance changes to heat within the control cycle.

[0034] It should be noted that during the calibration of the thermal conduction delay duration in each calibration cycle, the superposition of random power disturbances and the calculation of the power-torque change ratio within the current control cycle are paused; after the thermal conduction delay duration calibration is completed and the preset step power is canceled, the monitoring and calculation steps of the control cycle are resumed.

[0035] Step S2: After each control cycle, the power-torque change ratio sequence is updated using a sliding window. The drag evolution parameters are obtained based on the macroscopic distribution characteristics of the power-torque change ratio sequence, and the drag disorder parameters are obtained based on the structural disorder characteristics of the power-torque change ratio sequence.

[0036] Because the power-torque change ratio calculated within a single control cycle is highly susceptible to accidental material drops, gear backlash, or electromagnetic interference, it contains significant background noise. By using a sliding window, discrete and isolated power-torque change ratios can be converted into a continuous time series, ensuring that the power-torque change ratio series has both the ability to smooth historical noise and the ability to dynamically track and keenly reflect the latest resistance evolution state of aluminum ash after each control cycle.

[0037] Based on this, in this embodiment of the invention, after each control cycle, a power-torque change ratio sequence is updated using a sliding window. That is, an updated power-torque change ratio sequence can be obtained after each control cycle. The sliding window follows the first-in-first-out principle, and the length N of the sliding window is preset to an integer greater than the minimum sample size required for sample entropy calculation, with a value range of 100-500. In this example, N is 100.

[0038] Specifically, after each control cycle ends, the total number K of all historical control cycles since the start of refining heating is counted. If K < N, the power-torque change ratio of the control cycle is pushed into the sliding window in chronological order (this is the pre-filling stage, and the subsequent power-torque change ratio sequence is not constructed). Until K≥N, the sliding window is initialized and filled, obtaining the power-torque change ratio after sorting all the timing sequences within the sliding window, thus constructing a power-torque change ratio sequence; among them, the power-torque change ratio of the latest completed control cycle is pushed to the tail of the sliding window queue; as the control cycle is updated, the sliding window is also continuously updated to construct the power-torque change ratio sequence, and the length of the power-torque change ratio sequence is N.

[0039] During the heating and stirring process of aluminum ash, the impact characteristics of aluminum ash blocks with different physical properties (including unmelted blocks or abnormally sintered blocks) differ during stirring and tumbling, resulting in different torque performance of the stirring motor. Unmelted blocks gradually melt as they are heated and acquire rheological properties. During stirring and tumbling, the torque (i.e., rheological resistance) exhibits slow, continuous, and relatively stable fluctuations, with a high degree of concentration in the fluctuation trend. It also maintains a certain regularity and self-similarity in terms of time sequence. In contrast, abnormally sintered blocks are rigid and brittle solids formed by sintering. During stirring and tumbling, the torque will exhibit random, high-frequency, and sharp fluctuations corresponding to disordered collisions, and will show low self-similarity and high complexity in terms of time sequence. Furthermore, since the power-torque change ratio quantifies the increase in rheological resistance of aluminum ash caused by unit thermal energy disturbance, it reflects the response characteristics of aluminum ash material to thermal energy. That is, under random disturbance of heating power, whether the stirring resistance decreases gradually or is random and disordered at high frequency, it can be unaffected by the amount of material and help to assess whether the material is essentially melting and softening or sintering and hardening after absorbing heat. Based on this, embodiments of the present invention will obtain resistance evolution parameters based on the macroscopic distribution characteristics of the power-torque change ratio sequence, and obtain resistance disorder parameters based on the structural disorder characteristics of the power-torque change ratio sequence. Among them, the resistance evolution parameters characterize the macroscopic evolution trend of the rheological resistance of aluminum ash as it changes with heating, which can help assess whether the aluminum ash material as a whole is softening (rheological resistance decreases) or hardening (rheological resistance increases). The resistance disorder parameters reflect whether random, high-frequency, and irregular rigid collisions occur in the aluminum ash during stirring, which helps to accurately distinguish abnormal agglomeration collisions from a microscopic perspective. The two resistance-related parameters evaluate the changes in the material state of aluminum ash as it is heated from both macroscopic and microscopic perspectives, providing an evaluation basis for subsequent differentiation between abnormal sintering and normal unmelted material to adaptively control heating.

[0040] Preferably, in one embodiment of the present invention, considering that the arithmetic mean can characterize the dominant statistical direction, and the mean sign directly characterizes the macroscopic evolution direction; and that the mean can suppress instantaneous random disturbances and extract slow evolution characteristics; then the method for obtaining the resistance evolution parameter includes: The mean of the power-torque change ratio in the power-torque change ratio sequence is used as the drag evolution parameter.

[0041] When the resistance evolution parameter is less than 0, it indicates that the material in the sliding window mainly exhibits endothermic softening, and the rheological resistance evolution of aluminum ash shows a downward trend; when the resistance evolution parameter is greater than 0, it indicates that the material in the sliding window shows a hardening trend with increasing resistance after being heated.

[0042] Since sample entropy (SampEn) can help assess the complexity and irregularity of a time series by statistically analyzing the probability of similar patterns repeating, and aluminum ash rheology or cold material tumbling has inertial continuity in physics, and anomalous agglomeration impact is a memoryless discrete event in physics, the idea of ​​sample entropy can be used to assess the disorder characteristics of the power-torque change ratio sequence and quantify the drag disorder parameter. By setting the embedding dimension to divide the power-torque ratio sequence into fixed-length basic sub-vectors, it is possible to capture similar change patterns of consecutive equal-length sub-vectors. Further dividing into relatively longer, higher-dimensional sub-vectors helps to verify the preservation of similar change patterns. At the same time, based on the fluctuation characteristics of the sequence itself, a similarity tolerance is determined to ensure the strictness of similarity, thereby adapting to the current operating noise of the equipment and shielding conventional mechanical vibrations. Furthermore, the loss of regularity in the power-torque ratio sequence can be evaluated by comprehensively considering similar change patterns under different dimensions, so as to quantify the drag disorder parameter. Based on this, in a preferred embodiment of the present invention, the method for obtaining the resistance disorder parameter includes: In the power-torque change ratio sequence, all basic sub-vectors with a length of a preset embedding dimension and all upgraded sub-vectors with a length of a preset embedding dimension plus 1 are extracted respectively, and the similarity tolerance is determined based on the sequence fluctuation characteristics. Based on the differences between different basic sub-vectors and combined with similarity tolerance, the basic matching probability under the preset embedding dimension is obtained; based on the differences between different upgraded sub-vectors and combined with similarity tolerance, the upgraded matching probability under the preset embedding dimension plus 1 is obtained; the resistance disorder parameter is determined based on the basic matching probability and the upgraded matching probability.

[0043] Specifically, the preset embedding dimension is set to 2, and all continuous sequence elements of length 2 are sequentially extracted from the power-torque change ratio sequence as the basic sub-vector; similarly, all continuous sequence elements of length 3 are sequentially extracted from the power-torque change ratio sequence as the up-dimensional sub-vector; then the standard deviation of the power-torque change ratio in the power-torque change ratio sequence is calculated, and the standard deviation is multiplied by a preset weight such as 0.1-0.25, which is 0.2 in this example, and the product is used as the similarity tolerance; When the product is less than the minimum tolerance threshold, the similarity tolerance is forcibly assigned to the minimum tolerance threshold, such as 0.01-0.05, based on the background noise range of the stirring motor torque sensor under no-load steady state, or 1 to 2 times the minimum resolution of ADC sampling; this prevents the standard deviation from being 0 when the material state is stagnant or the sensor accuracy is truncated, resulting in the sequence being completely identical, which would lead to the similarity tolerance being 0 and causing the sample entropy calculation to fail and crash. For any basic subvector A, combine the remaining basic subvectors to construct a set XA; use the Chebyshev distance between A and each basic subvector in set XA as the vector difference; count the basic subvectors in XA whose vector difference with A is less than the similarity tolerance, and calculate their proportion in set XA; use the proportion as the matching probability of basic subvector A; calculate the matching probability of all basic subvectors and take the mean; use the mean as the basic matching probability under the preset embedding dimension. Similarly, the dimensionality-upgrading matching probability under the preset embedding dimension plus 1 can be calculated; to prevent the calculation from crashing due to the matching probability being 0, a minimal positive parameter is introduced to prevent zero; the basic matching probability is added to the minimal positive parameter such as 0.00001 and used as the denominator, and the dimensionality-upgrading matching probability is added to the minimal positive parameter such as 0.00001 and used as the numerator. The ratio is then taken as the negative natural logarithm, which is mapped to the -ln() function to obtain the sample entropy, which is the resistance disorder parameter; Among them, as the embedding dimension increases, the matching conditions between sub-vectors become more stringent, so the probability of dimensionality-upgrading matching is less than or equal to the basic matching probability. When aluminum ash absorbs heat normally or cold material tumbles, the drag change is constrained by thermal inertia and physical inertia, exhibiting strong self-similarity, i.e., the ratio approaches 1, and the drag disorder parameter is close to 0. When the oxidation reaction causes the hard block to sinter and impact the blade, the power-torque change ratio sequence may contain more random high-frequency disorder features, the probability of dimensionality-upgrading matching decreases relatively, and the ratio drops significantly. After negative logarithmic operation, it is mapped to a very large positive value, resulting in an increase in the drag disorder parameter.

[0044] Step S3: Determine the steady-state endothermic softening stage in the refining heating process based on the resistance evolution parameters, and determine the disorder safety benchmark based on the resistance disorder parameters within the steady-state endothermic softening stage; after the current control cycle ends, compare the resistance disorder parameters with the disorder safety benchmark, and in conjunction with the resistance evolution parameters, determine whether there is abnormal sintering of aluminum ash, and adjust the heating power of the heating module based on the determination result.

[0045] Because the composition and initial temperature of different batches of aluminum ash vary greatly, the basic mechanical noise during stirring is also different. Using a fixed resistance evolution parameter threshold can easily lead to false alarms or missed alarms of abnormal sintering. Therefore, the embodiments of the present invention first determine the steady-state endothermic softening stage in the refining heating process based on the resistance evolution parameter, so as to adaptively extract an adaptive alarm benchmark that is compatible with the current batch of materials.

[0046] Preferably, in one embodiment of the present invention, considering that a resistance evolution parameter less than 0 characterizes the material's endothermic softening, a preset evolution threshold can be set to 0. However, a single instance of a resistance evolution parameter less than 0 does not necessarily indicate overall phase change melting of the material. Only when the resistance evolution parameter remains consistently less than zero can it be considered that the material is macroscopically in a stable state of continuous endothermic softening and melting. Therefore, the method for obtaining the steady-state endothermic softening stage includes: Since the start of refining and heating, the resistance evolution parameters obtained after each control cycle are continuously monitored. When the number of times the resistance evolution parameters are continuously less than the preset evolution threshold reaches the preset number threshold for the first time, the corresponding time period of all control cycles corresponding to the resistance evolution parameters that have continuously reached the preset number threshold is extracted as the steady-state endothermic softening stage.

[0047] Specifically, the preset evolution threshold is set to 0, and the preset number threshold is set to 5-N. In this example, 5 times is used, but the implementer can also adjust it. From the start of refining and heating, when the resistance evolution parameter obtained after each control cycle is continuously negative and reaches 5 times for the first time, it is determined that the system has entered a stable and heat-absorbing melting safe stage. The time period corresponding to these 5 consecutive control cycles is taken as the steady-state heat-absorbing softening stage.

[0048] Considering that even in the steady-state endothermic softening stage, due to factors such as material non-uniformity, mechanical vibration of stirring blades, and sensor noise, the resistance disorder parameter will not be absolutely zero, but rather there is an inherent, non-zero background benchmark. This background benchmark is determined by the current refining heating process state and environment, and cannot be uniformly applied with a fixed threshold. Once the steady-state endothermic softening stage is determined, the material is in a relatively safe endothermic melting state, namely macroscopic softening, no abnormal sintering, and stable thermal response. The resistance disorder parameters in this stage can help quantify the normal tumbling characteristics of aluminum ash, thereby helping to assess the background benchmark, namely the disorder safety benchmark, in the current refining heating process, and thus use it as a safety benchmark for risk warning. Based on this, a disordered safety benchmark is further determined based on the resistance disorder parameters during the steady-state endothermic softening stage.

[0049] Preferably, in one embodiment of the present invention, considering that the distribution characteristics of the disordered resistance parameters obtained after all control cycles within the steady-state endothermic softening stage can help smooth out the random errors of a single measurement; setting a preset margin weight for weighting can accommodate normal process fluctuations on the basis of an absolute safety baseline; then the method for obtaining the disordered safety benchmark includes: Extract the resistance disorder parameters obtained after all control cycles during the steady-state endothermic softening stage. Determine the initial disorder safety benchmark based on the distribution characteristics of all resistance disorder parameters. Use preset margin weights to weight the initial disorder safety benchmark to obtain the disorder safety benchmark.

[0050] Specifically, the resistance disorder parameters obtained after all control cycles during the steady-state endothermic softening stage are extracted, and the mean of all resistance disorder parameters is used as the initial disorder safety benchmark. The preset margin weight is set to 1.25-1.5, and 1.5 is used in this example. The initial disorder safety benchmark is multiplied by the preset margin weight, and the product is used as the disorder safety benchmark.

[0051] It should be noted that if the aluminum ash loaded at the start of the refining process is extremely refractory or the ambient temperature is extremely low, the aluminum ash material may remain in a state of slight hardening or resistance stagnation for a long time in the initial stage, making it impossible to determine the steady-state endothermic softening stage for an extended period. To prevent process deadlock caused by extreme abnormal operating conditions, a timeout check must be performed when extracting the steady-state endothermic softening stage: a timer is used to determine whether the preset maximum time (45 minutes in this example) has been exceeded since the start of refining heating. If the steady-state endothermic softening stage cannot be determined even after exceeding the preset maximum time, the disorder safety benchmark is directly set to an empirical value. This empirical value is based on the long-term historical healthy operation data statistics of this model of refining equipment, which will not be elaborated further.

[0052] Once the disorder safety benchmark is determined, the resistance disorder parameters can be compared with the disorder safety benchmark after the current (latest) control cycle ends. Combined with the resistance evolution parameters, it can be determined in real time whether there is abnormal sintering of aluminum ash, and the heating power of the heating module can be adjusted based on the determination results.

[0053] Preferably, in one embodiment of the present invention, considering that the physical essence of abnormal sintering is that the rigid solid formed by sintering undergoes random, high-frequency, and irregular rigid collisions during stirring and tumbling, i.e., the aluminum ash material hardens (resistance evolution parameter is greater than 0) and the resistance disorder parameter significantly increases and exceeds the disorder safety benchmark, which can help in the determination; the method for determining whether abnormal sintering exists in the aluminum ash includes: If, after the current control cycle ends, the resistance disorder parameter is greater than the disorder safety benchmark and the resistance evolution parameter is greater than the preset evolution threshold, it is determined that the aluminum ash has abnormal sintering.

[0054] It should be noted that the preset evolution threshold is 0. Through dual-constraint cross-verification, only when the resistance evolution parameter is greater than 0 (confirming that the material has left the softening trajectory and the overall macro resistance tends to harden), and at the same time the resistance disorder parameter is greater than the safety benchmark (confirming that there is a high-frequency random rigid impact that is different from the conventional friction), is the aluminum ash abnormal sintering determined; otherwise, the aluminum ash is determined to be normally melted or cold material is under heavy load (i.e., the normal rolling friction of the unmelted material briefly causes abnormal torque).

[0055] When abnormal sintering of aluminum ash is detected, the abnormal intervention mechanism is immediately activated after the current control cycle ends: a command to reduce the heating power to zero is sent to the heating module to prevent the external heat from continuing to accumulate, thus avoiding accelerated aluminum ash caking or causing equipment safety accidents; at the same time, the stirring motor is controlled to apply alternating mechanical shear force to physically break up the clumps. Specifically, the upper limit of the output frequency of the inverter (controlling the stirring motor) is set to 50Hz, and the lower limit is set to 10Hz; the preset rated acceleration and deceleration time parameters of the inverter are called (e.g., 3 seconds), and the safe torque limiting function of the inverter is activated. Then, an alternating frequency control command is sent to the inverter: the inverter drives the stirring motor to smoothly increase to 50Hz within 3 seconds according to the set acceleration slope and maintain the set high frequency time window (e.g., 2 seconds); then, according to the set deceleration slope, it smoothly decreases to 10Hz within 3 seconds and maintains the set low frequency time window (e.g., 2 seconds); this cycle is repeated. This alternating speed change operation continuously generates alternating mechanical shear stress in the drum to shear and crush the agglomerates to a rheological state.

[0056] It should be noted that during the alternating shearing process, the real-time resistance disorder parameters are continuously monitored until it is determined that there is no abnormal sintering of aluminum ash.

[0057] When it is determined that there is no abnormal sintering of aluminum ash (i.e., normal melting of aluminum ash or heavy loading of cold material), the abnormal intervention mechanism will not be activated. The device used for aluminum refining will continue to heat according to the set program and maintain the planned frequency of the frequency converter to control the operation of the stirring motor.

[0058] Based on the same inventive concept, the present invention also proposes an aluminum ash heating control device for aluminum refining. The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aluminum ash heating control method for aluminum refining described in steps S1-S3.

[0059] In summary, this invention periodically calibrates the thermal conduction delay time of aluminum ash; within each control cycle, it acquires the change in heating power before and after the heating module experiences a random power disturbance, and extracts the change in the delayed torque of the stirring motor by extending the control cycle by the thermal conduction delay time, calculating the power-torque change ratio; it acquires the resistance evolution parameters and resistance disorder parameters after each control cycle, thereby determining the steady-state endothermic softening stage and establishing a disorder safety benchmark. Furthermore, at the end of the current control cycle, it determines whether abnormal sintering of the aluminum ash exists and adjusts the heating accordingly. This invention eliminates thermal hysteresis misalignment of aluminum ash by calibrating the thermal conduction delay and random power disturbance; combining macroscopic resistance evolution and microscopic regularity characteristics, it accurately distinguishes between normal cold material and abnormal sintering, enabling precise intervention in anomalies and improving the control effect of aluminum ash heating and refining.

Claims

1. A method for controlling the heating of aluminum ash used in aluminum refining, characterized in that, The method includes: During the refining and heating process, the thermal conduction delay time of aluminum ash is calibrated periodically; within each control cycle, the change in heating power before and after the heating module is superimposed with random power disturbance is obtained, and the change in delay torque of the stirring motor is extracted after the control cycle is extended by the thermal conduction delay time. The power-torque change ratio is calculated based on the change in heating power and the change in delay torque. After each control cycle, the power-torque change ratio sequence is updated using a sliding window. The drag evolution parameters are obtained based on the macroscopic distribution characteristics of the power-torque change ratio sequence, and the drag disorder parameters are obtained based on the structural disorder characteristics of the power-torque change ratio sequence. Based on the resistance evolution parameters, the steady-state endothermic softening stage in the refining heating process is determined, and the disorder safety benchmark is determined based on the resistance disorder parameters within the steady-state endothermic softening stage. After the current control cycle ends, the resistance disorder parameters and the disorder safety benchmark are compared, and combined with the resistance evolution parameters, it is determined whether there is abnormal sintering of aluminum ash, and the heating power of the heating module is adjusted based on the determination results.

2. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, Methods for periodically calibrating the thermal conduction delay time of aluminum ash include: From the start of refining and heating, a calibration cycle with a preset dynamic duration is continuously determined. Within each calibration cycle, the planned heating power at the start time is superimposed with a preset step power and heating is controlled. At the same time, timing is started to monitor the torque of the stirring motor in real time within the calibration cycle. When the torque change relative to the starting torque exceeds the preset torque threshold, timing is stopped, and the timing duration is used as the heat conduction delay duration.

3. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, The method for obtaining the change in heating power includes: Within each control cycle, a limited random power increment is generated. The planned heating power at each moment is superimposed with the random power increment and heating is controlled. The heating power of the heating module at the beginning and end of the control cycle is collected and the difference is calculated to obtain the change in heating power.

4. The method for controlling the heating of aluminum ash for aluminum refining according to claim 2, characterized in that, The method for obtaining the change in delayed torque includes: For each control cycle, determine the heat conduction delay time calibrated within its calibration cycle, and extend the heat conduction delay time backward along the time axis at the first and last two moments respectively to obtain the corresponding extension period of the control cycle. Collect the torque of the stirring motor at the first and last moments of the extension period and calculate the difference to obtain the change in delayed torque.

5. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, The methods for obtaining the resistance evolution parameters include: The average value of the power-torque change ratio in the power-torque change ratio sequence is used as the drag evolution parameter.

6. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, The method for obtaining the resistance disorder parameter includes: In the power-torque change ratio sequence, all basic sub-vectors with a length of a preset embedding dimension and all upgraded sub-vectors with a length of a preset embedding dimension plus 1 are extracted respectively, and the similarity tolerance is determined based on the sequence fluctuation characteristics. Based on the differences between different basic sub-vectors and in conjunction with the similarity tolerance, the basic matching probability under the preset embedding dimension is obtained; based on the differences between different upgraded sub-vectors and in conjunction with the similarity tolerance, the upgraded matching probability under the preset embedding dimension plus 1 is obtained; the resistance disorder parameter is determined based on the basic matching probability and the upgraded matching probability.

7. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, The method for obtaining the steady-state endothermic softening stage includes: Since the start of refining and heating, the resistance evolution parameter obtained after each control cycle is continuously monitored. When the number of times the resistance evolution parameter is continuously less than the preset evolution threshold first reaches the preset number threshold, the time period corresponding to all control cycles that have continuously reached the preset number threshold is extracted as the steady-state endothermic softening stage.

8. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, The method for obtaining the disordered security benchmark includes: Extract the resistance disorder parameters obtained after all control cycles during the steady-state endothermic softening stage. Determine the initial disorder safety benchmark based on the distribution characteristics of all resistance disorder parameters. Use preset margin weights to weight the initial disorder safety benchmark to obtain the disorder safety benchmark.

9. The method for controlling the heating of aluminum ash for aluminum refining according to claim 1, characterized in that, Methods for determining whether aluminum ash has abnormal sintering include: If, after the current control cycle ends, the resistance disorder parameter is greater than the disorder safety benchmark and the resistance evolution parameter is greater than the preset evolution threshold, it is determined that the aluminum ash has abnormal sintering.

10. A heating control device for aluminum ash refining, the device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the aluminum ash heating control method for aluminum refining as described in any one of claims 1 to 9.

Images