Method and system for collaborative processing of aluminum electrolytic production full-process data

Abstract

The application provides an aluminum electrolysis full-process data collaborative processing method and system, and relates to the technical field of data processing.The method comprises the following steps: calculating the area of a region enclosed by a closed contour in a two-dimensional plane, taking the area equal to a preset standard area value as a target, and optimizing a time lag parameter of a cell voltage state trajectory relative to an anode rod current state trajectory to obtain a final time lag parameter.The application can solve the final time lag and complete time sequence compensation through phase space reconstruction, composite phase path diagram closed contour area iterative optimization, can resist industrial noise, and can eliminate inherent time sequence lag of aluminum electrolysis current and voltage signals without manual intervention, output a collaborative data set matched in time sequence mechanism, and improve cell condition diagnosis and anode effect prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for collaborative data processing throughout the entire aluminum electrolysis production process. Background Technology

[0002] In the aluminum electrolysis production process, anode rod current and cell voltage are two key process parameters reflecting the operating status of the electrolytic cell. Anode rod current directly characterizes the workload distribution of each anode, while cell voltage comprehensively reflects the overall electrochemical state of the electrolytic cell; there is a close physical correlation between the two. With the rapid development of the aluminum electrolysis industry towards intelligent manufacturing, collaborative analysis and fusion modeling based on multi-source data has become an important means to improve the control accuracy and operational stability of the electrolytic cell. Achieving precise time-series coordination of these two types of data is a fundamental prerequisite for the effective implementation of the aforementioned analysis.

[0003] However, in actual production, the anode rod current and the cell voltage are not strictly synchronized. Changes in alumina concentration within the electrolytic cell first trigger a response in the cell voltage, which is then gradually transmitted to the anode rod currents through changes in inter-electrode resistance. This transmission process involves multiple physical mechanisms, including electrochemical response, bubble dynamics, and aluminum melt fluctuations, resulting in a natural time lag between the cell voltage signal and the anode rod current signal. More complexly, this lag is not a fixed value but exhibits time-varying characteristics as the electrolytic cell operating conditions (such as the rate of change in alumina concentration, inter-electrode spacing fluctuations, and electrolyte temperature drift) dynamically change. Differences in the response characteristics of different sampling channels and sensors themselves also introduce additional time base offsets.

[0004] Regarding the aforementioned time lag issue, existing processing strategies still have noteworthy problems. Some solutions employ empirical offset compensation based on fixed time windows, but fixed parameters are difficult to adapt to lag changes caused by operating condition drift. Other solutions rely on manual subjective judgment and alignment based on curve shape, but differences in experience among different operators can easily lead to poor consistency of results, and the processing efficiency is low when dealing with large amounts of historical data. In addition, while some existing data fusion methods focus on the uniformity of timestamps at the acquisition level, they rarely consider the inherent response lags caused by differences in process mechanisms between different physical quantities. When combining two types of data for dynamic process analysis, certain biases may still be introduced due to time misalignment.

[0005] For example, in a 300kA series prebaked aluminum electrolytic cell during normal production, the alumina feeder dispenses alumina at fixed intervals of approximately 2 minutes. When the alumina concentration in the cell gradually decreases from approximately 2.5% to close to 1.8%, the cell voltage initially exhibits a continuous upward trend for approximately 8 to 12 seconds. Subsequently, as the inter-electrode resistance changes, the current in each anode rod sequentially and responsively decreases. Actual measurements show that the delay between the start of the cell voltage rise and the significant response of the anode rod current is not fixed, typically fluctuating between 40 and 120 seconds. This delay is closely related to changes in electrolyte temperature and fluctuations in the aluminum melt interface. If fixed time-delay compensation is used or translation alignment is performed solely based on manual experience, the timing deviation between current and voltage may exceed 30 seconds under certain operating conditions. When subsequent calculations of cell resistance change rate and predictions of anode effect are performed based on these two types of data, this deviation will, to some extent, affect the timeliness and reliability of the characteristic indicators. Summary of the Invention

[0006] This invention provides a method and system for collaborative data processing throughout the entire aluminum electrolysis production process, which can adaptively eliminate the time lag between current and voltage signals in aluminum electrolysis production.

[0007] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a method for collaborative data processing throughout the entire aluminum electrolysis production process, the method comprising: Step 1: Obtain the anode rod current fluctuation sequence of the same aluminum electrolysis cell as the first time series data, and the cell voltage fluctuation sequence as the second time series data; perform phase space reconstruction on the first time series data and the second time series data respectively to obtain the anode rod current state trajectory and the cell voltage state trajectory. Step 2: Project the current state trajectory of the anode conductor and the voltage state trajectory of the tank onto the same two-dimensional plane to obtain the composite phase path diagram; Step 3: In the composite phase path diagram, trace the envelope boundary closed by the difference in local motion direction between the anode rod current state trajectory and the tank voltage state trajectory along the time progression direction to obtain a closed profile. Step 4: Calculate the area of ​​the region enclosed by the closed profile in the two-dimensional plane. With the area equal to the preset standard area value as the target, optimize the time lag parameter of the tank voltage state trajectory relative to the anode rod current state trajectory to obtain the final time lag parameter. Step 5: Use the final time delay parameter to perform time base shift compensation on the tank voltage fluctuation sequence to obtain the compensated tank voltage fluctuation sequence; combine the compensated tank voltage fluctuation sequence with the anode conductor current fluctuation sequence as the full process data after collaborative processing.

[0008] Secondly, the data collaborative processing system for the entire aluminum electrolysis production process includes: The phase space reconstruction module is used to acquire the anode rod current fluctuation sequence of the same aluminum electrolysis cell as the first time series data and the cell voltage fluctuation sequence as the second time series data; the phase space is reconstructed on the first time series data and the second time series data respectively to obtain the anode rod current state trajectory and the cell voltage state trajectory. The projection module is used to project the current state trajectory of the anode conductor and the voltage state trajectory of the tank onto the same two-dimensional plane to obtain a composite phase path diagram. The contour tracing module is used to trace the envelope boundary closed by the difference in local motion direction between the anode rod current state trajectory and the tank voltage state trajectory in the composite phase path diagram along the time progression direction to obtain a closed contour. The time delay optimization module is used to calculate the area of ​​the region enclosed by the closed profile in the two-dimensional plane. With the area equal to the preset standard area value, it optimizes the time delay parameters of the tank voltage state trajectory relative to the anode rod current state trajectory to obtain the final time delay parameters. The compensation combination module is used to perform time-base shift compensation on the tank voltage fluctuation sequence using the final time delay parameter to obtain the compensated tank voltage fluctuation sequence; the compensated tank voltage fluctuation sequence is combined with the anode conductor current fluctuation sequence as the full-process data after collaborative processing.

[0009] Thirdly, a computing device includes: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0010] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0011] The above-described solution of the present invention has at least the following beneficial effects: This method reconstructs the phase space of current and voltage time-series data separately, transforming one-dimensional time-series signals into high-dimensional state trajectories that characterize the system's operating rules. It quantifies time lag based on local motion differences in the trajectory, overcoming the traditional one-dimensional matching method's reliance on waveform similarity and improving the accuracy of time alignment from a mechanistic perspective. An iterative optimization mechanism driven by the difference between the closed contour area and a preset standard area value is established. The time lag parameters are automatically updated in the direction of decreasing absolute value of the difference, converging to the optimal value after multiple iterations. This closed-loop optimization strategy allows each data processing to adaptively determine the corresponding time lag parameters based on the actual data characteristics of the current operating condition. It adapts to changes in lag caused by operating condition drift during production without human intervention, solving the problem of deteriorating compensation effects of fixed-parameter compensation schemes under varying operating conditions.

[0012] Using the closed contour area formed by the difference in the motion direction of the two trajectories as the quantitative basis for dynamic differences, it belongs to the global matching index at the geometric feature level. It is not sensitive to local waveform noise and small-amplitude signal distortion, and can effectively avoid noise interference in the industrial field, characterizing the true temporal coupling relationship of the two types of parameters. Through time base translation compensation, the temporal misalignment problem caused by sensor response differences and production mechanism lag is eliminated, resulting in a full-process collaborative dataset with high temporal matching and strong mechanism coupling. This effectively solves the problems of distortion in the calculation of the tank resistance change rate, low accuracy of anode effect prediction, and large deviation in tank condition diagnosis caused by traditional data preprocessing methods. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating the data collaborative processing method for the entire aluminum electrolysis production process provided in an embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of a data collaborative processing system for the entire aluminum electrolysis production process provided in an embodiment of the present invention. Detailed Implementation

[0015] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0016] like Figure 1 As shown, embodiments of the present invention propose a method for collaborative data processing throughout the entire aluminum electrolysis production process, the method comprising the following steps: Step 1: Obtain the anode rod current fluctuation sequence of the same aluminum electrolysis cell as the first time series data, and the cell voltage fluctuation sequence as the second time series data; perform phase space reconstruction on the first time series data and the second time series data respectively to obtain the anode rod current state trajectory and the cell voltage state trajectory. Step 2: Project the current state trajectory of the anode conductor and the voltage state trajectory of the tank onto the same two-dimensional plane to obtain the composite phase path diagram; Step 3: In the composite phase path diagram, trace the envelope boundary closed by the difference in local motion direction between the anode rod current state trajectory and the tank voltage state trajectory along the time progression direction to obtain a closed profile. Step 4: Calculate the area of ​​the region enclosed by the closed profile in the two-dimensional plane. With the area equal to the preset standard area value as the target, optimize the time lag parameter of the tank voltage state trajectory relative to the anode rod current state trajectory to obtain the final time lag parameter. Step 5: Use the final time delay parameter to perform time base shift compensation on the tank voltage fluctuation sequence to obtain the compensated tank voltage fluctuation sequence; combine the compensated tank voltage fluctuation sequence with the anode conductor current fluctuation sequence as the full process data after collaborative processing.

[0017] In this embodiment, the data acquisition and processing system includes the following components: The anode rod current acquisition unit includes a Hall current sensor installed on each anode rod. The Hall current sensor utilizes the Hall effect principle: when a current-carrying conductor is placed in a magnetic field perpendicular to the current direction, a Hall voltage proportional to the current magnitude is generated across the conductor. In other words, it converts the large DC current flowing through the rod into a collectable voltage signal. In this embodiment, a Hall current sensor with a range of 0 to 10 kA is selected. The electrolytic cell in this embodiment has 20 anode rods, with one Hall current sensor installed on each rod, resulting in a total of 20 current acquisition channels.

[0018] The cell voltage acquisition unit's core hardware is a high-precision isolated DC voltage transmitter. The device uses a parallel connection method, directly bridging the anode and cathode busbars of the aluminum electrolysis cell, to acquire the DC operating voltage between the two electrodes of the electrolysis cell around the clock. The actual operating voltage of the aluminum electrolysis cell is a low-voltage DC signal, with a normal stable operating range of 4.0V to 5.0V. This voltage signal cannot be accurately identified and acquired directly by a general-purpose data acquisition card; therefore, a voltage transmitter is needed to perform electrical isolation, proportional scaling, and signal standardization conversion. The voltage transmitter is a linear proportional signal conversion device. In this embodiment, a dedicated transmitter with a primary side range of 0 to 6V is selected to adapt to voltage fluctuations under all operating conditions of the electrolysis cell. The actual linear ratio relationship of this transmitter is: when the transmitter input terminal (primary side, connected to the electrolysis cell) receives a DC voltage of 0 to 6V, the corresponding output terminal (secondary side, connected to the data acquisition card) linearly outputs a standard DC voltage of 0 to 10V.

[0019] A data acquisition card, also known as a data acquisition card, is a hardware module that converts the analog voltage signal output by the sensor into a digital quantity via an analog-to-digital converter (ADC) and transmits it to a host computer. This embodiment uses a synchronous sampling data acquisition card with no fewer than 21 channels (e.g., a 24-channel or 32-channel card), of which 20 channels are used for anode conductor current signal acquisition and 1 channel is used for tank voltage signal acquisition. Synchronous sampling means that the data acquisition card, under the control of the same clock trigger signal, performs sampling and holding and analog-to-digital conversion on the analog input signals of all channels at the same time, ensuring that the timestamps of the data from each channel are consistent. The sampling frequency is set to 1Hz (i.e., one data point is acquired per second for each channel). The sampling frequency refers to the number of times the analog signal is converted to digital and the digital value is recorded per unit time.

[0020] An industrial control computer (host computer) is used in this embodiment. The data collaborative processing program refers to a computer program compiled according to steps 1 to 5, which is used to perform a series of timing collaborative processing operations on the collected anode rod current data and tank voltage data, such as phase space reconstruction, composite phase path diagram generation, closed contour tracking, time delay parameter optimization, and time base translation compensation.

[0021] In a preferred embodiment of the present invention, step 1 involves acquiring the anode rod current fluctuation sequence of the same aluminum electrolytic cell as first time-series data and the cell voltage fluctuation sequence as second time-series data; the execution process of reconstructing the phase space of the first time-series data and the second time-series data to obtain the anode rod current state trajectory and the cell voltage state trajectory is as follows: The data acquisition card synchronously acquires the output signals of Hall current sensors (e.g., 20) and one cell voltage transmitter installed on the anode guide rods of the aluminum electrolysis cell at a preset sampling frequency (e.g., 1Hz). Synchronous acquisition means that all channels are triggered simultaneously at the rising edge of the same sampling clock, and the sample-and-hold circuits of each channel capture the analog voltage value at the same instant. Subsequently, the analog-to-digital converter sequentially quantizes the voltage values ​​held by each channel.

[0022] For the j-th anode conductor (j=1,2,...,20), the digital current sequence after analog-to-digital conversion is obtained from the data acquisition card. Based on the sensor's calibration coefficient—the conversion ratio between the Hall current sensor's output voltage and the measured current on the primary side (in this embodiment, this coefficient is 0.4V / kA (meaning 1kA of primary side current corresponds to 0.4V sensor output voltage)—the digital value is converted to an actual current value in kiloamperes (kA). The conversion method is: divide the digital voltage read from the data acquisition card by 0.4; the quotient is the current value (kA). This yields the current fluctuation sequence for the j-th anode conductor. The current fluctuation sequence refers to a one-dimensional current value sequence arranged chronologically by sampling time. The k-th element in the sequence (k=1,2,...,N, where N is the total number of sampling points) represents the instantaneous current value at the k-th sampling moment.

[0023] In this embodiment, the current fluctuation sequence of the first anode guide rod (the one closest to the aluminum outlet end, whose current change is more sensitive to changes in the tank condition) is used as the first time series data. This sequence is denoted as I, where i k This represents the current value at the k-th sampling time, in kA.

[0024] Simultaneously, the secondary-side standard voltage digital sequence of the cell voltage transmitter after analog-to-digital conversion is read from the data acquisition card, and the actual primary-side cell voltage of the electrolytic cell is calculated by combining it with the transmitter's preset linear scaling relationship. In this embodiment, the voltage transmitter adopts a linear scaling rule of 0 to 6V (primary side) corresponding to 0 to 10V (secondary side). The specific conversion method is as follows: the secondary-side standard voltage value (range 0 to 10V) acquired by the data acquisition card is multiplied by a fixed conversion factor of 0.6 (derived from the maximum primary-side range of 6V ÷ the maximum secondary-side range of 10V), and the calculated result is the actual value of the primary-side cell voltage of the electrolytic cell (unit: V). All sampled data are processed point by point through this conversion method, and finally a complete cell voltage fluctuation sequence is obtained. This sequence is denoted as U, where u_k represents the actual cell voltage value corresponding to the k-th sampling time, in V.

[0025] In this embodiment, it is assumed that 600 sampling points are taken, corresponding to 10 minutes of continuous data collection. This data duration is sufficient to cover multiple typical cycles of alumina concentration changes during aluminum electrolysis (the alumina feeding interval is approximately 2 minutes, and 10 minutes contains approximately 5 feeding cycles).

[0026] For the aforementioned first time-series data I, phase space reconstruction is performed according to the following process. Phase space reconstruction refers to a technique that maps a one-dimensional time series to a higher-dimensional state space using delayed coordinate embedding. For a one-dimensional observation sequence generated by a deterministic dynamical system, by selecting an appropriate embedding dimension and delay time, a state trajectory equivalent to the original system's dynamics can be recovered in the reconstructed higher-dimensional state space. The anode conductor current state trajectory refers to a curve formed in the state space after phase space reconstruction, composed of discrete points connected sequentially in time. This curve reflects the state change pattern of the anode conductor current at different times.

[0027] Given a predetermined embedding dimension m1=3 and a delay time τ1=5, constructing a delay vector requires extracting m1 sample values ​​from the original sequence, with each sample value separated by τ1 sampling intervals. The minimum sampling index k that can construct the delay vector must satisfy k-(m1-1)×τ1≥1. Solving for k, we obtain the possible values. Assuming k is 11, the effective sampling index k ranges from k=11,12,13,...,600 effective sampling points. The first 10 sampling points (k=1 to 10) cannot participate in phase space reconstruction because they do not meet the construction conditions.

[0028] A first delay vector sequence is constructed point by point. The delay vector is a multi-dimensional vector composed of m1 sample values ​​extracted sequentially at equal intervals (τ1 sample points) from the original one-dimensional time series. Each vector corresponds to a point in the reconstructed state space. For each sample index k within the effective range, a three-dimensional delay vector is constructed according to the following rules: Extract the first coordinate component (or first-dimensional coordinate value) of the delay vector, which is the current value i at index k in the original sequence I. k That is, the current value at the current sampling moment; take the second coordinate component of this delay vector, whose value is the current value i at index k-5 in the original sequence I. k－5 That is, tracing back 5 sampling intervals (5 seconds ago); taking the 3rd coordinate component of this delay vector, its value is the current value i at index k-10 in the original sequence I. k－10 This involves tracing back 10 sampling intervals (10 seconds ago) to obtain the current value. From this, a three-dimensional delay vector is obtained at the k-th valid sampling time (the first coordinate = i). k Second coordinate = i k－5 The third coordinate = i k－10 Incrementing k from 11 to 600 sequentially forms the first delay vector sequence.

[0029] Each delay vector in the first delay vector sequence is considered as an independent state point in the three-dimensional state space. Following the time progression order of sampling index k from smallest to largest, adjacent state points are connected end-to-end with straight line segments, ultimately forming a continuously extending polygonal line trajectory in three-dimensional space. This trajectory is the anode conductor current state trajectory, denoted as T. I Each trajectory point is composed of three sets of current data from different time series, forming a three-dimensional coordinate system: the first dimension represents the current real-time current value, the second dimension represents the historical current value from 5 seconds ago, and the third dimension represents the historical current value from 10 seconds ago. Traditional one-dimensional current sequences can only display the instantaneous current magnitude at each moment, failing to reflect the current's changing trend, fluctuation inertia, and temporal correlation characteristics. However, through three-dimensional phase space reconstruction, a single point integrates multi-dimensional information from the current state, short-term historical state, and long-term historical state, fully revealing the dynamic changes and fluctuation characteristics of the electrolytic cell current hidden within the one-dimensional time series data and not readily observable. The final constructed T... I The trajectory can characterize the dynamic evolution path of the anode rod current over time.

[0030] For the second time-series data U mentioned above, the reconstruction method is exactly the same as that used for the current sequence, i.e., embedding parameters are selected, including a second embedding dimension m2=3 and a second delay time τ2=5; the effective sampling range is k=11,12,...,600. The second delay vector sequence is constructed point-by-point, i.e., for each effective sampling index k, a three-dimensional delay vector is constructed according to the following rules. Specifically, the first coordinate component takes the value u. k (Slot voltage value at the current sampling time); the second coordinate component is u. k－5 (Slot voltage value 5 seconds ago); the third coordinate component is u. k－10(Slot voltage value 10 seconds ago). The second delay vector sequence is constructed by assigning k values ​​from 11 to 600. Connecting the delay vectors in the second delay vector sequence with straight line segments according to their k values ​​from smallest to largest forms the slot voltage state trajectory, denoted as T. U The sampling time corresponding to each trajectory point and T I The trajectory points correspond one-to-one.

[0031] In a preferred embodiment of the present invention, step 2, which involves projecting the anode conductor current state trajectory and the tank voltage state trajectory onto the same two-dimensional plane to obtain a composite phase path diagram, is performed as follows: After obtaining the above two three-dimensional state trajectories T I and T U Then, the two are projected onto the same two-dimensional plane to obtain a composite phase path diagram. The composite phase path diagram is a graphical representation of the projection of the state trajectories of two different physical quantities onto the same two-dimensional coordinate system. It is used to visually compare the relative positional relationship, differences in motion direction, and spatial topology of the two trajectories at the same time index in the same plane.

[0032] The specific rule-making process for the projection operation in this embodiment is as follows: The anode conductor current state trajectory and the tank voltage state trajectory output in step 1 correspond one-to-one in terms of time index. In this embodiment, the first coordinate component (corresponding to the sampled value at the current moment) and the second coordinate component (corresponding to the sampled value 5 seconds ago) of each trajectory point are selected to form a two-dimensional projection. These two dimensions are selected because the first coordinate component carries the main information at the current moment, and the second coordinate component carries short-term historical information. The combination of the two can better reflect the local motion characteristics of the trajectory.

[0033] For the current state trajectory T I The trajectory point corresponding to the kth valid sampling time has three coordinate components i, i, and i respectively. k i k－5 i k－10 Extract the first coordinate component i k As the original value of the horizontal coordinate of the trajectory point on the two-dimensional plane, the second coordinate component i is extracted. k－5 Used as the original value for the ordinate. For the voltage state trajectory T U The trajectory point corresponding to the same k-th valid sampling time has three coordinate components as u, ... k u k－5 u k－10 Extract the first coordinate component u k As the original value of the horizontal axis, extract the second coordinate component u. k－5 As the original value of the ordinate.

[0034] Coordinate normalization refers to a data preprocessing method that maps original physical quantities with different dimensions and numerical ranges to a common dimensionless scale (the [0,1] interval in this embodiment). Since current (unit kA, typical range 2.0 to 5.0 kA) and voltage (unit V, typical range 4.0 to 5.0 V) have different dimensions and numerical ranges, they need to be normalized before they can be compared in the same coordinate system. The specific rules for normalization are as follows: Calculate the maximum value i of the sampled values ​​in the current sequence I respectively. max and minimum value i min Then calculate the maximum value u of all sampled values ​​in the voltage sequence U. max and minimum value u min A linear mapping is performed on the coordinates of each original projection point of the current trajectory, i.e., the normalized abscissa = (original abscissa i...) / ( ... k -i min ) / (i max -i min Normalized ordinate = (original ordinate i) k－5 -i min ) / (i max -i min Similarly, a linear mapping is performed on the coordinates of each original projection point of the voltage trajectory, i.e., the normalized abscissa = (u k -u min ) / (u max -u min The quotient obtained; the normalized ordinate = (u k－5 -u min ) / (u max -u min The quotient obtained.

[0035] After the above mapping, the coordinates of all projected points of the current trajectory and voltage trajectory fall within a unit square region of [0,1]×[0,1], making them comparable at the same scale. Connecting adjacent points in the normalized current projection point sequence with straight line segments according to the sampling index k in ascending order yields the normalized current state trajectory projection curve C. I Connect the normalized voltage projection point sequence in the same way to obtain the normalized voltage state trajectory projection curve C. U C I and C U Simultaneously plotted in the same two-dimensional coordinate system, a composite phase path diagram is obtained. By unifying the two-dimensional projection and normalizing the alignment, the two independent trajectories reflecting the kinetic state of the electrolyzer are mapped to the same evaluation scale.

[0036] In a preferred embodiment of the present invention, step 3, in the composite phase path diagram, involves tracing the envelope boundary closed by the local motion direction difference between the anode conductor current state trajectory and the tank voltage state trajectory along the time progression direction to obtain a closed profile. The execution process is as follows: The composite phase path diagram output in step 2 includes the normalized current state trajectory projection curve and the voltage state trajectory projection curve, both of which are formed by connecting the projection points in chronological order. Each projection point is represented by a two-dimensional coordinate (x-axis, y-axis). For each valid sampling time k (k=11,12,...,599, the last time k=600 cannot be calculated due to the lack of data for the next time step), the following two sets of extraction operations are performed simultaneously: Extract the first trajectory tangent point and the first instantaneous tangential vector, i.e., locate the first trajectory tangent point on the current trajectory side: find the normalized projection point P corresponding to the k-th sampling time on the current state trajectory projection curve. I(k) This point is the first tangent point of the anode conductor current state trajectory at the k-th sampling time. The tangent point refers to the specific geometric location of the state trajectory at a given sampling time; it is a discrete node on the trajectory curve. Calculate the tangential vector D at the first instant. I(k) That is, take the (k+1)th projection point P on the projection curve. I(k+1) and the k-th projection point P I(k) The coordinate difference is obtained from P I(k) Point to P I(k+1) The directed line segment. The instantaneous tangential vector refers to the directed line segment pointing from the trajectory point at the current moment to the trajectory point at the next moment. The direction of this vector represents the local motion direction of the trajectory at the current moment, and the length of this vector represents the motion speed of the trajectory at the current moment. The abscissa component of the first instantaneous tangential vector is calculated as follows: Take P I(k+1) Subtract P from the x-coordinate I(k) The difference in the horizontal coordinates; the calculation method for the vertical coordinate components is: take P I(k+1) ordinate minus P I(k) The difference in the ordinate.

[0037] Extract the second trajectory tangent point and the second instantaneous tangential vector, i.e., locate the second trajectory tangent point on the voltage trajectory side: find the normalized projection point P corresponding to the k-th sampling time on the voltage state trajectory projection curve. U(k) This point is the second tangent point of the slot voltage state trajectory at the k-th sampling time. Calculate the second instantaneous tangential vector D. U(k) Take the projection curve C U The (k+1)th projection point P U(k+1) and the k-th projection point P U(k) The coordinate difference is obtained. Its x-coordinate component is P. U(k+1) Subtract P from the x-coordinateU(k) The difference in the x-coordinates, with the y-coordinate component being P. U(k+1) ordinate minus P U(k) The difference in the ordinate.

[0038] For each valid sampling time k, a set of data is obtained, namely the first trajectory tangent and its first instantaneous tangential vector, and the second trajectory tangent point and its second instantaneous tangential vector. Each sampling time k corresponds to the first trajectory tangent point, the first instantaneous tangential vector, the second trajectory tangent point, and the second instantaneous tangential vector. For each sampling time k, two local reference lines are established. The local reference line is a straight line segment extending from the trajectory tangent point along the direction of the instantaneous tangential vector at that point. It linearly extrapolates the instantaneous motion trend of the trajectory at that point, serving as a local reference line that approximately describes the trajectory's direction near that point.

[0039] A first local reference line is established, which is a line segment starting from the tangent point of the first trajectory and extending along the direction of the first instantaneous tangential vector. The length of the line segment is taken as the length of the first instantaneous tangential vector. The length of this vector is calculated by taking the square of the abscissa component and the square of the ordinate component of the first instantaneous tangential vector, and then taking the square root of the sum. The resulting first local reference line approximately describes the local motion trend of the current state trajectory at time k.

[0040] A second local reference line is established, which is a line segment starting from the tangent point of the second trajectory and extending along the direction of the second instantaneous tangential vector. The length of the line segment is taken as the length of the second instantaneous tangential vector (calculated in the same way as above). The resulting second instantaneous tangential vector approximately describes the local motion trend of the voltage state trajectory at time k. Each sampling time k corresponds to a pair of local reference lines, namely the first local reference line and the second local reference line.

[0041] The trajectory tangent points at each sampling time k, i.e., the first and second trajectory tangent points, the instantaneous tangential vectors, i.e., the first and second instantaneous tangential vectors, the local reference lines, i.e., the first and second local reference lines, and the trajectory point coordinate data for the two times before and after the current time (a total of 5 consecutive sampling times k-2, k-1, k, k+1, k+2). For a few times at the beginning and end (k=11, 12 or k=598, 599) where the preceding and following data are missing, the available times are used as substitutes.

[0042] The rate of change of the relative positions of the two trajectories is calculated. This rate of change refers to how quickly the Euclidean distance between the current trajectory point and the voltage trajectory point on the two-dimensional projection plane changes with time near the current sampling moment. A large rate of change indicates that the two trajectories are accelerating apart or accelerating closer together (significant difference in motion), while a small rate of change indicates that the two trajectories maintain a relatively stable spatial relationship (their motion trends are nearly identical). The calculation process is as follows: For each of the five consecutive sampling times (current time k and four adjacent times), calculate the Euclidean distance d(t) between the current trajectory point and the voltage trajectory point. Calculate the change in distance between adjacent times, i.e., calculate three changes: Δd1 = d(k) - d(k-1); Δd2 = d(k+1) - d(k); Δd3 = d(k+2) - d(k+1); where d(k-1), d(k), d(k+1), and d(k+2) are the Euclidean distances between the current and voltage trajectory points corresponding to the times before k, the current time k, the time after k, and the two times after k, respectively; Δd1 is the forward distance change; Δd2 is the current instantaneous distance change; and Δd3 is the backward distance change. Calculate the relative position change rate r(k) at the current time k. r(k) is the arithmetic mean of the three changes. Using a three-neighborhood average can smooth out random fluctuations in single-step calculations.

[0043] The constant curvature value is determined by the rate of change. Curvature refers to a geometric quantity describing the degree of bending of a curve on a surface. Zero curvature represents a straight line, and a larger absolute value of curvature indicates a more severe bending. The constant curvature refers to the design requirement that the curvature at all points on the entire arc constructed at the current sampling time remains at the same constant value. For the k-th sampling time, the constant curvature value κ(k) is calculated by multiplying the absolute value of the rate of change |r(k)| by a preset curvature scaling factor α (α=1.0 in this embodiment). All trajectory data in this scheme are normalized in the interval [0,1]. The rate of change of trajectory spacing itself is a dimensionless small-scale value. Setting α to 1.0 can achieve a 1:1 mapping between curvature and the rate of change of motion. This avoids amplifying small fluctuations due to an excessively large scaling factor, causing excessive bending of the arc and introducing computational noise, and also avoids weakening the real motion differences and causing the loss of trajectory features due to an excessively small scaling factor. When |r(k)| is large (the spatial distance between the two trajectories is changing rapidly), κ(k) is also large, and the arc connecting the tangent points of the two trajectories is more curved; when the rate of change is small, κ(k) is small, and the arc is relatively straight. Thus, κ(k) becomes a geometric quantification of the degree of difference in the local motion direction of the two trajectories at the current moment.

[0044] To find the minimum arc length of a continuous circular arc connecting two tangent points, using the constant curvature value determined in the second step as a constraint, find a continuous circular arc that simultaneously satisfies the following three geometric constraints: Constraint 1 (Endpoint Condition): The starting point of the arc must be exactly the tangent point of the first trajectory, and the ending point must be exactly the tangent point of the second trajectory. Constraint 2 (Tangential Condition): The tangent direction at the starting point of the arc must be consistent with the direction of the tangent vector at the first instant, and the tangent direction at the ending point must be consistent with the direction of the tangent vector at the second instant, ensuring smooth connection between the arc and its respective trajectories at both ends. Constraint 3 (Curvature Condition): The curvature at any point on the arc must be always equal to a constant curvature value, and no change in curvature is allowed along the arc path. Under these three constraints, select the arc with the shortest arc length (the path length along the arc curve from the starting point to the ending point) from all possible candidate arcs as the continuous circular arc at the k-th sampling time. The minimum arc length is chosen because, given the endpoint and tangential conditions, the curve with the shortest arc length is geometrically the most compact and provides the most refined quantification of the local motion differences between the two trajectories. For each valid sampling time k, a corresponding continuous circular arc is obtained.

[0045] Take the current-side endpoints of all arcs (i.e., the sequence of tangent points of the first trajectory) {P I(11) ,P I(12) ,...,P I(599) These 589 points are arranged in a two-dimensional plane in the direction of time progression. Connecting adjacent points from k=11 to k=599 in chronological order with straight line segments yields a broken line composed of 588 small straight line segments connected end-to-end. This broken line forms the upper boundary of the envelope. The upper boundary traces the historical path of the current state trajectory's positional changes during time progression. Taking the voltage-side endpoints of all arcs (i.e., the sequence of tangent points of the second trajectory) {P} U(11) ,P U(12) ,...,P U(599) These 589 points are also arranged in the direction of time progression on a two-dimensional plane. Connecting adjacent points from k=11 to k=599 in chronological order with straight line segments yields a broken line composed of 588 small straight line segments. This broken line forms the lower boundary of the envelope. The lower boundary tracks the historical path of the voltage state trajectory's positional changes during the time progression.

[0046] At the beginning of the time series (corresponding to k=11), the starting point P of the upper boundary is set. I(11) The starting point P of the lower boundary U(11) Connect the points with a straight line segment; this line represents the chord of the arc at the 11th sampling time (the straight line segment between the two endpoints of the arc). At the end of the time series (corresponding to k=599), the endpoint P of the upper boundary is... I(599) The endpoint P of the lower boundary U(599)Connecting these lines with straight lines creates a chord corresponding to the arc ARC(599). Thus, the upper boundary, lower boundary, and the two closed lines at the beginning and end together enclose a completely closed planar region.

[0047] The four boundaries mentioned above together form a closed envelope, which defines a complete closed profile in a two-dimensional plane, denoted as profile Ω. Profile Ω represents the cumulative quantification of the dynamic motion deviation and timing lag difference between the anode conductor current excitation and the cell voltage response within the time window from sampling time k=11 to k=599. If the motion directions of the current and voltage trajectories are always completely consistent, it indicates that there is no transmission lag between the current excitation and voltage response of the electrolytic cell, and the timing is completely synchronized. The dynamic changes of the two types of data are highly coupled. In this case, the upper and lower boundaries of the envelope almost completely coincide, and the closed profile degenerates into a linear trajectory, with the enclosed area approaching zero. Conversely, when there is signal transmission lag and dynamic fluctuation deviation within the electrolytic cell, the motion trends of the two trajectories are misaligned, the difference in motion direction increases, and the more severe the timing lag problem, the greater the opening and closing degree of profile Ω, and the larger the enclosed area of ​​the plane. Traditional aluminum electrolysis data processing relies solely on single-point numerical comparisons, failing to quantify the dynamic timing misalignment of current and voltage, thus hindering precise alignment of these two core production data types. This method transforms the closed contour area of ​​the trajectory into a quantitative indicator of current-voltage timing lag deviation. The size of the contour area quantitatively characterizes the degree of data mismatch under dynamic operating conditions of the electrolytic cell. By minimizing the contour area as the optimization objective, the optimal compensation parameters to offset the inherent timing lag of the electrolytic cell can be iteratively solved, addressing the pain points of asynchronous and poorly coordinated current and voltage timing data in aluminum electrolysis.

[0048] In a preferred embodiment of the present invention, step 4, calculating the area of ​​the region enclosed by the closed contour in the two-dimensional plane, and optimizing the time lag parameter of the tank voltage state trajectory relative to the anode conductor current state trajectory with the area equal to a preset standard area value, to obtain the final time lag parameter, is as follows: The initial value of the time lag parameter is set to 0 seconds. This initial value means that there is no time lag deviation between the anode rod current signal and the cell voltage signal. This lag-free baseline state is used as the starting condition for iterative optimization, and the iterative search for the globally optimal time lag parameter is carried out. Based on the set current time lag parameter, a time-base translation operation is performed on the cell voltage state trajectory. The time-base translation refers to a timing calibration operation that shifts the complete cell voltage state trajectory forward or backward along the time axis by a fixed time. The specific translation rules and timing matching principles are as follows: When the time lag parameter value is greater than 0 seconds, it means that the cell voltage signal lags behind the anode rod current signal. After the current state inside the electrolytic cell changes, it takes a corresponding time of propagation delay before the cell voltage will generate a synchronous response. The voltage timing data lags behind the current timing data as a whole. To eliminate timing misalignment and achieve precise alignment of operating condition data, the entire tank voltage state trajectory needs to be shifted forward in the direction of decreasing time on the time axis. This advances the timing reference of the voltage data, offsets inherent signal lag differences, and ensures precise time matching between the lagging voltage response data and the current change data that triggers the response. When the time lag parameter is less than 0 seconds, it indicates that the tank voltage signal leads the anode conductor current signal, and voltage changes precede current changes. In this case, the entire tank voltage state trajectory needs to be shifted backward in the direction of increasing time on the time axis, delaying the timing reference of the voltage data and correcting the data deviation caused by the timing lead. When the time lag parameter is equal to 0 seconds, it indicates that the current signal and voltage signal have no timing lag and are completely synchronized. No shifting operation is required; the shift offset is zero, and the shifted tank voltage trajectory is completely consistent with the original trajectory. The specific calculation rules for the translation offset are as follows: The unified unit of measurement for the time lag parameter is seconds. The sampling frequency of this system's data acquisition is 1 Hz, meaning that full-channel synchronous sampling is completed once per second. Therefore, the fixed sampling time interval between two adjacent sampling points is 1 second. Divide the current time lag parameter value by the sampling time interval value to obtain the theoretical number of translation sampling points. Finally, round the calculation result to the nearest integer to obtain the integer number of sampling point offsets that can be directly used for trajectory translation.

[0049] Following the complete process of step 2 above, the unchanged anode rod current state trajectory and the tank voltage state trajectory after time-base translation are reprojected onto the same two-dimensional plane (using the same normalization method as the first time, using the i of the original I sequence and U sequence). max i min u max u min (Normalization is performed) to obtain the updated composite phase path diagram. Following the complete process of step 3 above, the trajectory tangent point and instantaneous tangential vector are extracted sequentially from the updated composite phase path diagram, local reference lines are established, constant curvature is determined and continuous circular arcs are solved, and the circular arcs are connected to form the envelope boundary, finally obtaining the updated closed contour.

[0050] Calculate the area of ​​the region enclosed by the closed contour in a two-dimensional plane. The enclosed region refers to the two-dimensional connected planar region completely bounded by the boundary of the closed contour. The area calculation uses a discretized form of the polygon area formula (shoelace formula): The updated closed contour's complete boundary curve is divided into several ordered vertices. The upper boundary is formed by connecting the aforementioned 589 trajectory points sequentially, and the lower boundary is also formed by connecting the aforementioned 589 trajectory points sequentially. The first and last closed lines are each line segments. All these vertices are arranged in a counter-clockwise direction around the contour to obtain an ordered vertex sequence {(x1,y1),(x2,y2),...,(xM,yM)}, where M is the total number of vertices, and the coordinates of the first vertex and the Mth vertex are the same (i.e., the contour is closed).

[0051] For each index value j from 1 to M-1, calculate the cross product between the j-th vertex and the (j+1)-th vertex. The calculation is as follows: multiply the x-coordinate of the j-th vertex by the y-coordinate of the (j+1)-th vertex to obtain the first product; multiply the y-coordinate of the j-th vertex by the x-coordinate of the (j+1)-th vertex to obtain the second product; subtract the second product from the first product, and the difference is the cross product term. Add all M-1 cross product terms together to obtain the sum; take the absolute value of the sum and divide it by 2; the quotient is the area of ​​the region enclosed by the closed contour in the two-dimensional plane.

[0052] The difference ΔA between the area of ​​the region enclosed by the closed contour in the two-dimensional plane and the preset standard area value is calculated, along with the absolute value of this difference, |ΔA|. If ΔA is positive (the current area is greater than the standard area), it indicates that the difference in the motion direction of the two trajectories under the current time delay parameters is greater than the ideal synchronization state, meaning that the timing misalignment has not yet been eliminated by the current time delay parameters. If ΔA is negative (the current area is less than the standard area), it also indicates that the current time delay parameters have deviated from the optimal value. Only when ΔA approaches zero does the difference in the motion direction between the two trajectories return to the level consistent with the ideal synchronization state.

[0053] The system checks whether the absolute value of the difference is less than the preset allowable range, i.e., the product of ε and the preset standard area value. In this embodiment, ε = 0.02, representing a 2% allowable steady-state error. The standard area value is set to 0.85, which is a statistical constant obtained by iteratively solving 20 sets of current-voltage data from the same electrolytic cell under historical stable operating conditions (i.e., periods without significant time lag deviation) using this method, and then taking the arithmetic mean of the converged area values ​​for each iteration. If the absolute value of the difference is less than the preset allowable range (i.e., the absolute value of the difference is less than 2% of the standard area value), it is determined that convergence has occurred. At this time, the iteration is terminated, and the current value of the current time lag parameter is used as the final time lag parameter, output, and proceed to step 5. If the absolute value of the difference is greater than or equal to the preset allowable range, it is determined that convergence has not occurred. The following parameter update steps are executed, and then a new round of iteration is restarted.

[0054] The parameter adjustment amount is calculated by multiplying the absolute value of the difference by the preset step size coefficient kp (kp=0.5). The preset step size coefficient kp avoids problems such as excessively large step sizes leading to excessively strong single parameter adjustments, iterative oscillations, and inaccurate convergence, while avoiding problems such as excessively small step sizes leading to slow iterative convergence, redundant iterations, and low computational efficiency. If ΔA>0 (current area is greater than standard area), it indicates that the current time lag parameter is too small (insufficient time base compensation), and the current value of the time lag parameter needs to be increased in the positive direction, with the direction sign set to +1. If ΔA<0 (current area is less than standard area), it indicates that the current time lag parameter is too large (excessive time base compensation), and the current value of the time lag parameter needs to be decreased in the negative direction, with the direction sign set to -1. The logical basis for the above direction judgment is that the time lag parameter and the closed contour area have a monotonic relationship within the geometric framework of this scheme. That is, near the optimal value, a smaller time lag parameter results in a larger area, and a larger time lag parameter results in a smaller area. The updated time lag parameter = the current value of the current time lag parameter + the product of the direction sign and the current parameter adjustment. The updated time lag parameter is then assigned to the current value of the time lag parameter as the new time lag parameter.

[0055] Based on the updated time lag parameters, the cell voltage state trajectory is re-shifted using the time base. Then, the entire process of regenerating the composite phase path diagram, extracting the closed profile, calculating the area and difference is executed sequentially. Finally, the convergence judgment of this step is executed again. This constitutes a closed-loop iteration. In each iteration, the time lag parameter is automatically adjusted in the direction of decreasing absolute value of the difference until |ΔA| drops below the allowable threshold. In addition, this embodiment sets the upper limit of the number of iterations to 50. If convergence is not achieved after 50 iterations (|ΔA| is still ≥ the preset allowable range), the time lag parameter corresponding to the round with the smallest absolute value of the difference is selected as the final time lag parameter from all the fully recorded iteration results. This iterative algorithm takes minimizing the deviation of the closed profile area as the core optimization objective. The smaller the absolute value of the difference, the smaller the difference in the movement of the current and voltage trajectories after the current time lag parameter correction, the higher the degree of synchronous coupling between the two time-series trajectories, and the closer it is to the real steady-state cooperative working condition of the aluminum electrolysis cell. It is the optimal approximate solution among all iteration results.

[0056] In a preferred embodiment of the present invention, step 5 involves using the final time delay parameter to perform time-base shift compensation on the tank voltage fluctuation sequence to obtain a compensated tank voltage fluctuation sequence. The process of combining the compensated tank voltage fluctuation sequence with the anode conductor current fluctuation sequence as the overall data after collaborative processing is as follows: Based on the final time delay parameters output in step 4, timing compensation calculations are performed using the sampling time interval of the slot voltage fluctuation sequence. The sampling frequency is once per second, and the fixed sampling time interval between two adjacent sampling points is 1 second. This sampling time interval is defined as the time span between two adjacent sampling moments, and its value is the reciprocal of the sampling frequency. The specific calculation method for the number of shifted sampling points is as follows: Take the absolute value of the final time delay parameter; divide the absolute value by the sampling time interval of 1 second to obtain the quotient; round the quotient according to the rounding rules (round up if the decimal part of the quotient is ≥0.5, discard if the decimal part is <0.5), to obtain the shifted sampling point offset. The sampling point offset refers to the number of sampling points that need to be shifted along the time axis for the recording time of all sampling points in the voltage sequence. Assuming the shifted sampling point offset is 68>0, it means that the tank voltage signal lags behind the anode rod current signal by 68 seconds. Therefore, the voltage sequence needs to be shifted 68 sampling points in the direction of decreasing time, that is, the effective recording time of the voltage data needs to be moved forward by 68 seconds, so that the voltage data originally recorded at the 69th second is physically aligned with the current data recorded at the 1st second.

[0057] Obtain the original slot voltage fluctuation sequence (the original one-dimensional voltage sequence obtained in step 1, before any translation processing), and the sampling point offset of 68. Based on the current value of the time lag parameter > 0 (voltage lag), the translation rules are as follows: Discard the first 68 sampled values ​​in the original voltage sequence, because there is no effective current data aligned with its physical mechanism within the time period corresponding to these sampled values ​​(the effect of current changes needs 68 seconds to be transmitted to the voltage signal and recorded; the voltage data of the first 68 seconds corresponds to the slot state 68 seconds before the start of the acquisition, and there is no causal correspondence with the current data acquired in the same time period). Use the remaining effective sampled values ​​in the original sequence, from the 69th sampled value to the 600th sampled value, as part of the compensated slot voltage fluctuation sequence. Renumber these values ​​sequentially, i.e., the first value of the compensated sequence equals the 69th sampled value of the original sequence, the second value equals the 70th sampled value of the original sequence, and so on. The time base of the compensated slot voltage fluctuation sequence has been compensated for time delay, that is, the effective physical time of each sample value is considered to be physically aligned with the recording time of the corresponding sequence number sample value in the anode rod current sequence.

[0058] Obtain the anode conductor current fluctuation sequence (original one-dimensional current sequence) obtained in step 1, and the compensated tank voltage fluctuation sequence. Since the length of the compensated voltage sequence is 532, while the length of the original current sequence is 600, their lengths are inconsistent. To maintain consistency in timing, a corresponding truncation operation is performed on the current sequence: discard the last 68 sample values ​​(i.e., values ​​533 to 600) because there is no effective voltage data aligned with their physical mechanism within the time period corresponding to these sample values. The current sequence retained after truncation also has a length of 532. Combine the truncated current sequence and the compensated voltage sequence one-to-one according to the same index, i.e., the first combined data pair is the first current sample value and the first compensated voltage sample value, the second combined data pair is the second current sample value and the second compensated voltage sample value, and so on. Each combined data pair contains the current value at the same physical moment and the voltage value aligned after time-delay compensation, ultimately obtaining the fully processed data sequence. The current and voltage values ​​at each time index in this data sequence have been aligned at the physical mechanism level. That is, the voltage value has been compensated for a time lag of 68 seconds, eliminating the timing misalignment caused by factors such as electrochemical response conduction, bubble dynamics, aluminum liquid fluctuations and differences in sensor response characteristics.

[0059] like Figure 2 As shown, embodiments of the present invention also provide a data collaborative processing system for the entire aluminum electrolysis production process, including: The phase space reconstruction module is used to acquire the anode rod current fluctuation sequence of the same aluminum electrolysis cell as the first time series data and the cell voltage fluctuation sequence as the second time series data; the phase space is reconstructed on the first time series data and the second time series data respectively to obtain the anode rod current state trajectory and the cell voltage state trajectory. The projection module is used to project the current state trajectory of the anode conductor and the voltage state trajectory of the tank onto the same two-dimensional plane to obtain a composite phase path diagram. The contour tracing module is used to trace the envelope boundary closed by the difference in local motion direction between the anode conductor current state trajectory and the tank voltage state trajectory in the composite phase path diagram along the time progression direction to obtain a closed contour. The time delay optimization module is used to calculate the area of ​​the region enclosed by the closed profile in the two-dimensional plane. With the area equal to the preset standard area value, it optimizes the time delay parameters of the tank voltage state trajectory relative to the anode rod current state trajectory to obtain the final time delay parameters. The compensation combination module is used to perform time-base shift compensation on the tank voltage fluctuation sequence using the final time delay parameter to obtain the compensated tank voltage fluctuation sequence; the compensated tank voltage fluctuation sequence is combined with the anode conductor current fluctuation sequence as the full-process data after collaborative processing.

[0060] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0061] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0062] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0063] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for collaborative data processing throughout the entire aluminum electrolysis production process, characterized in that: The method includes: Step 1: Obtain the anode rod current fluctuation sequence of the same aluminum electrolysis cell as the first time series data, and the cell voltage fluctuation sequence as the second time series data; perform phase space reconstruction on the first time series data and the second time series data respectively to obtain the anode rod current state trajectory and the cell voltage state trajectory. Step 2: Project the current state trajectory of the anode conductor and the voltage state trajectory of the tank onto the same two-dimensional plane to obtain the composite phase path diagram; Step 3: In the composite phase path diagram, trace the envelope boundary closed by the difference in local motion direction between the anode rod current state trajectory and the tank voltage state trajectory along the time progression direction to obtain a closed profile. Step 4: Calculate the area of ​​the region enclosed by the closed profile in the two-dimensional plane. With the area equal to the preset standard area value as the target, optimize the time lag parameter of the tank voltage state trajectory relative to the anode rod current state trajectory to obtain the final time lag parameter. Step 5: Use the final time delay parameter to perform time base shift compensation on the tank voltage fluctuation sequence to obtain the compensated tank voltage fluctuation sequence; combine the compensated tank voltage fluctuation sequence with the anode conductor current fluctuation sequence as the full process data after collaborative processing.

2. The method for collaborative data processing throughout the entire aluminum electrolysis production process according to claim 1, characterized in that, Phase space reconstruction is performed on the first and second time-series data respectively to obtain the anode conductor current state trajectory and the tank voltage state trajectory, including: For the anode rod current fluctuation sequence, a first embedding dimension and a first delay time are selected, and a first delay vector sequence is constructed according to the first embedding dimension and the first delay time. The first delay vector sequence is connected in time sequence in the phase space to form the anode rod current state trajectory. For the slot voltage fluctuation sequence, a second embedding dimension and a second delay time are selected. A second delay vector sequence is constructed according to the second embedding dimension and the second delay time. The second delay vector sequence is connected in time order in the phase space to form the slot voltage state trajectory.

3. The method for collaborative data processing throughout the entire aluminum electrolysis production process according to claim 2, characterized in that, Step 3 includes: Using the tangent points of the anode conductor current state trajectory and the tank voltage state trajectory at each moment as a reference, a continuous circular arc with constant curvature is constructed; the circular arcs are connected sequentially along the time progression direction to form the envelope boundary closed by the difference in the local motion direction of the two trajectories, thereby tracing the closed contour.

4. The method for collaborative data processing throughout the entire aluminum electrolysis production process according to claim 3, characterized in that, Using the tangent points of the anode conductor current trajectory and the tank voltage trajectory at various times as a reference, a continuous circular arc with constant curvature is constructed, including: For the current sampling moment, extract the first trajectory tangent point and the first instantaneous tangential vector of the anode conductor current state trajectory at the corresponding moment, and at the same time extract the second trajectory tangent point and the second instantaneous tangential vector of the tank voltage state trajectory at the same moment; A first local reference line is established starting from the first trajectory tangent point and along the direction of the first instantaneous tangential vector; a second local reference line is established starting from the second trajectory tangent point and along the direction of the second instantaneous tangential vector. Between the first local reference line and the second local reference line, a constant curvature value is determined based on the rate of change of the relative positions of the two trajectories at the current time and within several adjacent sampling times. With the constant curvature value as a constraint, a continuous circular arc with the minimum arc length connecting the tangent points of the first trajectory and the tangent points of the second trajectory is solved as a geometric quantitative expression of the difference in the local motion direction of the two trajectories at the corresponding time. The above construction process is repeated for each sampling moment along the time progression direction to obtain a series of continuous arcs corresponding to each moment.

5. The method for collaborative data processing throughout the entire aluminum electrolysis production process according to claim 4, characterized in that, The time lag parameters of the tank voltage state trajectory relative to the anode conductor current state trajectory are optimized to obtain the final time lag parameters, including: Set the current value of the time lag parameter to zero; use the current time lag parameter to perform time base shift on the slot voltage state trajectory to obtain the shifted slot voltage state trajectory; The composite phase path diagram is regenerated by combining the current state trajectory of the anode conductor with the translated tank voltage state trajectory, and the corresponding closed contour is extracted. Calculate the area of ​​the region enclosed by the re-extracted closed contour in the two-dimensional plane, and obtain the difference between the area and the standard area value; With the absolute value of the difference decreasing as the direction, update the current value of the time lag parameter, return to the step of shifting the time base to calculate the difference and iterate until the absolute value of the difference falls within the preset allowable range, and take the current time lag parameter at this time as the final time lag parameter.

6. The method for collaborative data processing throughout the entire aluminum electrolysis production process according to claim 5, characterized in that, With the absolute value of the difference decreasing as the direction, update the current value of the time lag parameter, return to the step of time base shifting to difference calculation and iterate until the absolute value of the difference falls within a preset allowable range. The current time lag parameter at this point is taken as the final time lag parameter, including: Multiply the absolute value of the difference by the preset step size coefficient to obtain the parameter adjustment amount for this time; determine the adjustment direction according to the positive or negative sign of the difference, and move the current time lag parameter by the parameter adjustment amount along the adjustment direction to form the updated time lag parameter; The updated time lag parameter is used as the new current time lag parameter. The process of time-base shifting the slot voltage state trajectory is then executed, followed by the sequential execution of regenerating the composite phase path diagram, extracting the closed profile, calculating the area and the difference. When the absolute value of the latest calculated difference is less than the preset allowable range threshold, the current time lag parameter corresponding to the condition is used as the final time lag parameter.

7. The method for collaborative data processing throughout the entire aluminum electrolysis production process according to claim 6, characterized in that, The time-base shift compensation of the slot voltage fluctuation sequence is performed using the final time delay parameter to obtain the compensated slot voltage fluctuation sequence, including: Based on the final time delay parameters and the sampling time interval of the slot voltage fluctuation sequence, determine the number of sampling points to be shifted and the offset. The entire sampling point in the slot voltage fluctuation sequence is shifted along the time axis by the number of sampling points offset to obtain the compensated slot voltage fluctuation sequence. The anode conductor current fluctuation sequence and the compensated tank voltage fluctuation sequence are combined in a time-corresponding manner to form a fully processed data sequence.

8. A data collaborative processing system for the entire aluminum electrolysis production process, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The phase space reconstruction module is used to acquire the anode conductor current fluctuation sequence of the same aluminum electrolysis cell as the first time series data, and the cell voltage fluctuation sequence as the second time series data. Phase space reconstruction was performed on the first and second time series data respectively to obtain the current state trajectory of the anode conductor and the voltage state trajectory of the tank. The projection module is used to project the current state trajectory of the anode conductor and the voltage state trajectory of the tank onto the same two-dimensional plane to obtain a composite phase path diagram. The contour tracing module is used to trace the envelope boundary closed by the difference in local motion direction between the anode conductor current state trajectory and the tank voltage state trajectory in the composite phase path diagram along the time progression direction to obtain a closed contour. The time delay optimization module is used to calculate the area of ​​the region enclosed by the closed profile in the two-dimensional plane. With the area equal to the preset standard area value, it optimizes the time delay parameters of the tank voltage state trajectory relative to the anode rod current state trajectory to obtain the final time delay parameters. The compensation combination module is used to perform time base shift compensation on the slot voltage fluctuation sequence using the final time delay parameter to obtain the compensated slot voltage fluctuation sequence. The compensated tank voltage fluctuation sequence and the anode conductor current fluctuation sequence are combined as the full-process data after collaborative processing.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images