Method for optimizing matching between press roller powder height and discharging speed of powder sowing machine

Abstract

The invention discloses a method for optimizing matching of the height of compression roller powder and the discharging speed of a powder spreader, and belongs to the technical field of powder discharging regulation and control. The method comprises the steps that compression roller powder height data and a current change signal of a compression roller motor are obtained; the change trend of the powder height is calculated according to the compression roller powder height data; if the absolute value of the change trend exceeds a preset height trend threshold value, determining a use speed fluctuation index of the powder based on a current change signal of a compression roller motor; a feedback control algorithm is adopted to adjust the using speed fluctuation index of the powder, and a discharging speed target value is obtained; and extracting speed deviation data according to the difference between the blanking speed target value and the actual blanking speed, and processing the speed deviation data by adopting a logical reasoning algorithm to obtain a speed control instruction. According to the method for optimizing the matching of the height of the press roller powder and the discharging speed of the powder sowing machine, the problems that existing powder discharging has remarkable defects in the aspects of dynamic regulation and control and real-time optimization, and complex production scenes are difficult to meet are solved.

Description

Technical Field

[0001] This invention relates to the field of powder feeding control technology, and in particular to a method for optimizing the matching of powder height on the pressure roller and feeding speed of the powder distributor. Background Technology

[0002] Powder processing plays a vital role in modern industrial production. In the field of electrode sheet pressing, powder distributors output powder onto pressure rollers for pressing and shaping. With the increasing level of industrial automation, the efficiency and precision of powder distributors, as core equipment in powder processing, have become a focus of industry attention.

[0003] Matching the feeding speed of the powder distributor with the powder height on the pressure roller is crucial to the equipment's performance. A proper match not only improves processing quality but also reduces resource waste. However, existing methods for matching the feeding speed and powder height often rely on fixed parameters or manual adjustment of the distributor's servo motor, making it difficult to reflect real-time changes and leading to powder accumulation or insufficient supply. Therefore, current technical solutions have significant shortcomings in dynamic control and real-time optimization, making it difficult to meet the needs of complex production scenarios. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the present invention provides a method for optimizing the matching of the powder height of the pressure roller and the feeding speed of the powder distributor, so as to solve the above-mentioned problems.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for optimizing the matching of powder height on the pressure roller and the feeding speed of the powder distributor, comprising the following steps: S1: Acquire data on the powder height of the pressure roller and the current change signal of the pressure roller motor; S2: Calculate the trend of powder height variation based on the powder height data of the pressure roller; S3: If the absolute value of the change trend exceeds the preset height trend threshold, the powder usage speed fluctuation index is determined based on the current change signal of the pressure roller motor. S4: The feedback control algorithm is used to adjust the fluctuation index of the powder usage speed to obtain the target value of the feeding speed; S5: Extract speed deviation data based on the difference between the target value of the feeding speed and the actual feeding speed, process the speed deviation data using a logical reasoning algorithm, and obtain speed control instructions.

[0006] Preferably, in step S2, a sequence of height values ​​at consecutive time points is obtained from the powder height data of the pressure roller; The height value sequence is divided into multiple fixed-length time interval sequences using a fixed window. Within each time interval sequence, a corresponding difference operation is performed. The difference operation involves subtracting the height values ​​at adjacent time points to obtain the difference. Then, the differences are arranged by time to obtain a local height change difference value sequence. The LinearRegression class of the sklearn library is used to fit a trend line to the local height change difference sequence. The difference in the local height change difference sequence is used as the y array, and the time point corresponding to the difference is used as the x array. The fit method is called to solve the linear model and output the fitted trend line equation y=mx+b, where m is the local trend slope value, x is the time point, y is the difference value, and b is the intercept value. The trend is obtained by averaging the slope values ​​of all local trends.

[0007] Optionally, in step S3, the step of analyzing the current change signal of the pressure roller motor includes: The deviation value is obtained by comparing the timing deviation between the current change signal of the pressure roller motor and the preset uniformity reference signal, and the deviation value is used as the indicator of the fluctuation of the powder usage speed at each time point.

[0008] Specifically, in step S3, if the absolute value of the change trend exceeds a preset height trend threshold, a current sequence for a continuous time period is obtained from the current change signal of the pressure roller motor. The frequency domain transformation of the current sequence was performed using the Fast Fourier Transform function of the NumPy library to obtain the spectral distribution characteristics; A high-frequency component sequence is formed by selecting amplitude values ​​higher than a preset frequency threshold from the spectral distribution characteristics; If the mean of the high-frequency component sequence exceeds the preset upper limit, the usage speed fluctuation index of the powder is determined based on the current change signal of the pressure roller motor.

[0009] It is worth noting that in step S4, the fluctuation index of the powder usage rate is arranged in time sequence to obtain a deviation value sequence. The feeding speed adjustment parameters are obtained by adjusting the deviation value sequence using a standard PID controller. The standard PID controller inputs the deviation value sequence, multiplies the current deviation value by the proportional term Kp, multiplies the integral value of the deviation value sequence by the integral term Ki, and multiplies the rate of change of the deviation value by the derivative term Kd. Then, the three are summed to obtain the feeding speed adjustment parameters, where Kp is the proportional gain constant, Ki is the integral gain constant, and Kd is the derivative gain constant. The target value for the feeding speed is obtained by multiplying the feeding speed adjustment parameter by the average value of the powder usage speed fluctuation index.

[0010] Preferably, in step S5, the difference between the target value of the feeding speed and the actual feeding speed is calculated to obtain speed deviation data; Speed ​​control commands are obtained by processing speed deviation data using Python conditional statements. The Python conditional statements take speed deviation data as input and output speed control commands based on a preset rule set.

[0011] Optionally, in step S1, a data processing method is used to filter the pressure roller powder height data to obtain smoothed pressure roller powder height data.

[0012] Specifically, in step S1, the step of obtaining the smoothed roller powder height data includes: arranging the roller powder height data by time to obtain the original sequence; The median filter of the NumPy library is used to select a neighborhood of size 5 from the original sequence. The median value is calculated for each point and replaced with the original value to obtain a preliminary sequence of powder height. The moving average filter from the NumPy library is used to select a neighborhood of size 10 from the initial sequence, calculate the average value for each point, and generate smoothed roller powder height data.

[0013] The beneficial effects of this invention are as follows: In the method for optimizing the matching of powder height on the pressure roller and the feeding speed of the powder distributor, the powder height on the pressure roller and the current change signal of the pressure roller motor are acquired in real time through a data acquisition device. The trend of powder height change is calculated. When its absolute value exceeds a preset threshold, the current change signal is analyzed to determine the fluctuation index of the powder feeding speed. A feedback control algorithm is used to adjust the degree of fluctuation, generate a target value for feeding speed, and a logical reasoning algorithm is used to process the deviation between the target value and the actual speed, outputting precise speed control commands. This solution achieves adaptive adjustment of the powder feeding speed through the fusion analysis of height and current data and the collaborative control of dual algorithms, significantly improving processing stability and product quality. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating a method for optimizing the matching of powder height on the pressure roller and the feeding speed of the powder distributor in one embodiment of the present invention; Figure 2 This is a flowchart of the step S1 in one embodiment of the present invention, which involves obtaining the smoothed roller powder height data. Figure 3 This is a flowchart illustrating step S3 in one embodiment of the present invention. Detailed Implementation

[0015] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that these descriptions are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0016] like Figure 1-3 As shown, a method for optimizing the matching of powder height on the pressure roller and the feeding speed of the powder distributor includes the following steps: S1: Acquire data on the powder height of the pressure roller and the current change signal of the pressure roller motor; S2: Calculate the trend of powder height variation based on the powder height data of the pressure roller; S3: If the absolute value of the change trend exceeds the preset height trend threshold, the powder usage speed fluctuation index is determined based on the current change signal of the pressure roller motor. S4: The feedback control algorithm is used to adjust the fluctuation index of the powder usage speed to obtain the target value of the feeding speed; S5: Extract speed deviation data based on the difference between the target value of the feeding speed and the actual feeding speed, process the speed deviation data using a logical reasoning algorithm, and obtain speed control instructions.

[0017] In the method for optimizing the matching of powder height on the pressure roller and the feeding speed of the powder spreader, the data acquisition device acquires real-time signals of powder height on the pressure roller and current changes in the pressure roller motor. The trend of powder height change is calculated, and when its absolute value exceeds a preset threshold, the current change signal is analyzed to determine the powder feeding speed fluctuation index. A feedback control algorithm is used to adjust the fluctuation level, generating a target feeding speed value. A logical reasoning algorithm processes the deviation between the target value and the actual speed, outputting precise speed control commands to adjust the servo motor of the powder spreader. This solution achieves adaptive adjustment of the powder feeding speed through the fusion analysis of height and current data and the collaborative control of dual algorithms, significantly improving processing stability and product quality.

[0018] Preferably, in step S2, a sequence of height values ​​at consecutive time points is obtained from the powder height data of the pressure roller; The height value sequence is divided into multiple fixed-length time interval sequences using a fixed window. Within each time interval sequence, a corresponding difference operation is performed. The difference operation involves subtracting the height values ​​at adjacent time points to obtain the difference. Then, the differences are arranged by time to obtain a local height change difference value sequence. The LinearRegression class of the sklearn library is used to fit a trend line to the local height change difference sequence. The difference in the local height change difference sequence is used as the y array, and the time point corresponding to the difference is used as the x array. The fit method is called to solve the linear model and output the fitted trend line equation y=mx+b, where m is the local trend slope value, x is the time point, y is the difference value, and b is the intercept value. The trend is obtained by averaging the slope values ​​of all local trends.

[0019] A fixed window divides the entire sequence into segments of equal length. For example, if the window length is set to 30 time points, a long sequence is divided into several segments of 30 points each. For instance, if the total sequence length is 300 points and the fixed window is 30 time points, it can be divided into 10 time-segment sequences. Each segment contains continuous height values. This division helps in the local analysis of short-term fluctuations in powder height, avoiding excessive complexity in the overall data. In one possible implementation, when performing the corresponding difference operation within each time-segment sequence, the specific process involves subtracting the height values ​​of adjacent time points. For example, in a time-segment sequence with height values ​​[20, 22, 21, 23], the differences are 22-20=2, 21-22=-1, and 23-21=2. These differences, arranged chronologically, form a local height change difference sequence [2, 1, 2]. This operation reveals the immediate increase or decrease in powder height. If the difference sequence shows mostly positive values, it indicates that powder tends to accumulate during that period; conversely, it may indicate accelerated consumption.

[0020] The `LinearRegression` class is a linear regression model that uses the least squares method to find the best-fit line, with the `x` array as the independent variable and the `y` array as the dependent variable. In practice, for a local difference sequence such as [2,-1,2,3] corresponding to the time point [1,2,3,4], the time points are set as the `x` array and the differences as the `y` array. After calling the `fit` method, the model parameters are obtained. The `fit` method is a method in Python's scikit-learn library, where `m` represents the slope, reflecting the overall trend direction of the differences, and `b` is the intercept, representing the initial offset. The fitted trend line equation `y=mx+b` can summarize the linear change of that segment. In one possible implementation, the specific solution process for the fitted trend line equation y=mx+b involves inputting data into the model and calculating coefficients. In this application, each local height change difference sequence is processed to form a corresponding trend line. If the fitting results of multiple local sequences show m values ​​of 0.5, 0.3, and 0.4 respectively, these slope values ​​quantify the rate of increase or decrease of the powder height difference over each time period, helping to identify uneven distribution in production.

[0021] The process of calculating the average of all local trend slope values ​​to obtain the trend is to sum all the m values ​​and then divide by the quantity. For example, if there are 5 local slopes [0.2, 0.4, 0.1, 0.3, 0.5], the average value is 0.3, which represents the overall trend. If this average value is positive, it indicates that the powder height tends to increase, possibly due to the efficient operation of the feeding system; conversely, if it is negative, it indicates that the powder height tends to decrease.

[0022] Optionally, in step S3, the step of analyzing the current change signal of the pressure roller motor includes: The deviation value is obtained by comparing the timing deviation between the current change signal of the pressure roller motor and the preset uniformity reference signal, and the deviation value is used as the indicator of the fluctuation of the powder usage speed at each time point.

[0023] In this embodiment, the preset uniformity reference signal is an ideal stable current curve, such as a reference sequence constant at 12A. The comparison process specifically calculates the difference between the two signals at the same time point. For example, when the actual current is 13A and the reference is 12A, the deviation is 1A. These deviation values ​​are then used as fluctuation indicators. During production, if indicators at multiple time points show accumulated positive deviations, it may indicate that the powder consumption rate on the pressure rollers is too fast. The powder usage rate represents the amount of powder extruded by the pressure rollers and conveyed to the next process per unit time. In this embodiment, since the deviation value is proportional to the fluctuation of the powder usage rate, the deviation value can be used to reflect the fluctuation index of the powder usage rate.

[0024] Specifically, in step S3, if the absolute value of the change trend exceeds a preset height trend threshold, a current sequence for a continuous time period is obtained from the current change signal of the pressure roller motor. The frequency domain transformation of the current sequence was performed using the Fast Fourier Transform function of the NumPy library to obtain the spectral distribution characteristics; A high-frequency component sequence is formed by selecting amplitude values ​​higher than a preset frequency threshold from the spectral distribution characteristics; If the mean of the high-frequency component sequence exceeds the preset upper limit, the usage speed fluctuation index of the powder is determined based on the current change signal of the pressure roller motor.

[0025] In one possible implementation, when the absolute value of the powder height change trend exceeds a preset height trend threshold, a continuous time-span current sequence needs to be extracted from the current change signal of the pressure roller motor. This current change signal refers to the current fluctuation data generated by the motor driving the pressure roller during the operation of the roller press. For example, in a production line, the powder spreader is responsible for evenly distributing the powder. If the height trend shows an abnormal increase or decrease, the system will automatically extract the current values ​​of the pressure roller motor collected over the past 5 minutes, forming a sequence containing multiple sampling points. These values ​​may fluctuate from an initial 10A to 15A, reflecting changes in the workload of the roller press, thus providing a real-time data basis for subsequent analysis. This acquisition process emphasizes continuity, ensuring the sequence covers the complete operating cycle and avoiding data gaps that could affect the accuracy of the judgment.

[0026] The Fast Fourier Transform (FFT) is an algorithm that converts a time-domain signal into a frequency-domain signal, revealing hidden periodic patterns by calculating the frequency components of the signal. The FFT function takes a current sequence as input and outputs the corresponding amplitude sequence in the frequency domain as its spectral distribution characteristics. Assuming the current sequence is an array such as [10,12,11,13,10], inputting it into NumPy's FFT function will output complex frequency-domain data. The amplitude portion is then taken as the spectral distribution characteristics. Here, the amplitude sequence represents the energy intensity at each frequency; for example, the low-frequency portion might correspond to a stable current baseline, while the high-frequency portion reflects sudden fluctuations. This conversion helps extract useful information from noise.

[0027] Specifically, the process of extracting high-frequency component sequences from spectral distribution features involves selecting the portion of the amplitude sequence that is higher than a preset frequency threshold. For example, if the preset frequency threshold is 10 Hz, then all amplitude values ​​exceeding this frequency will be collected into a new sequence.

[0028] This corresponds to capturing the current signal of the rapid vibration of the roller motor of the roller press. If the amplitude sequence is [5,8,12,15,3] and the high-frequency component sequence extracted after combining with the preset frequency threshold is [12,15], these values ​​quantify the short-term instability in powder conveying. After extraction, its statistical characteristics can be further analyzed to evaluate the overall uniformity, such as calculating the mean of the high-frequency component sequence as mentioned in this embodiment. By judging that the mean of the high-frequency component sequence exceeds the preset upper limit, the powder usage speed fluctuation index is calculated, which can reduce some unnecessary calculations. For example, when the mean of the high-frequency component sequence is lower than or equal to the preset upper limit, it indicates that the production is stable, and there is no need for additional intervention to adjust its operation.

[0029] It is worth noting that in step S4, the fluctuation index of the powder usage rate is arranged in time sequence to obtain a deviation value sequence. The feeding speed adjustment parameters are obtained by adjusting the deviation value sequence using a standard PID controller. The standard PID controller inputs the deviation value sequence, multiplies the current deviation value by the proportional term Kp, multiplies the integral value of the deviation value sequence by the integral term Ki, and multiplies the rate of change of the deviation value by the derivative term Kd. Then, the three are summed to obtain the feeding speed adjustment parameters, where Kp is the proportional gain constant, Ki is the integral gain constant, and Kd is the derivative gain constant. The target value of the feeding speed is obtained by multiplying the feeding speed adjustment parameter by the average value of the powder usage speed fluctuation index, where the target value of the feeding speed is used to adjust the feeding speed of the powder distributor.

[0030] Assuming the resulting deviation value sequence is [0.5, 0.8, -0.3, 1.2, 0.4, -0.6, 0.9, 1.1, -0.2, 0.7], positive values ​​represent positive deviations of the current signal, and negative values ​​represent negative deviations of the current signal. This arrangement ensures the continuity of the time sequence, making it easier for subsequent analysis to capture the dynamic trend of the powder usage rate.

[0031] A standard PID controller is a feedback control mechanism. Its core lies in correcting system errors using three terms: proportional, integral, and derivative. The proportional term Kp, multiplied by the current deviation value, is used for rapid response (in this embodiment, the current deviation value is the latest deviation value in the deviation value sequence at any given time). The integral term Ki, multiplied by the integral value of the deviation value sequence, eliminates steady-state error (in this embodiment, the integral value of the deviation value sequence is the sum of all deviation values ​​in the sequence multiplied by the acquisition period). The derivative term Kd, multiplied by the rate of change of the deviation value, predicts future trends and suppresses overshoot (in this embodiment, the rate of change of the deviation value is the difference between the current deviation value and the previous deviation value divided by the acquisition period). Finally, the result of multiplying the proportional term Kp by the current deviation value, the result of multiplying the integral term Ki by the integral value of the deviation value sequence, and the result of multiplying the derivative term Kd by the rate of change of the deviation value are added together to obtain the feeding speed adjustment parameters.

[0032] In one possible implementation, the mean value of the powder usage speed fluctuation index is obtained by extracting all deviation values ​​from the entire deviation value sequence and then averaging these deviation values. For example, if the mean value of the deviation value sequence [0.5, 0.8, -0.3, 1.2] is 0.55, and the feeding speed adjustment parameter is 3.5, then the target feeding speed value is 1.925, which is used to adjust the feeding speed of the powder distributor.

[0033] Preferably, in step S5, the difference between the target value of the feeding speed and the actual feeding speed is calculated to obtain speed deviation data. Through this calculation, dynamic deviations in the production process can be captured in a timely manner, providing a data basis for subsequent control, thereby enabling precise adjustment of powder distribution in business operations and avoiding quality problems caused by speed mismatch. Speed ​​control commands are obtained by processing speed deviation data using Python conditional statements. The Python conditional statements take speed deviation data as input and output speed control commands based on a preset rule set.

[0034] Python's conditional statements are a programming logic structure that allows different operation paths to be executed based on conditions of input data. Here, it takes speed deviation data as input and outputs corresponding control commands based on a preset rule set. In this embodiment, the preset rule set includes intervals of speed deviation data and corresponding speed control commands for these intervals. After determining the intervals using the input speed deviation data, the corresponding speed control commands can be found and output based on the intervals, thus achieving automated response. The speed control commands then drive the servo motor of the powder-spreading machine.

[0035] Specifically, after driving the servo motor of the powder distributor via speed control commands, the powder output per unit time of the pressure roller and the powder output per unit time of the powder distributor are obtained. The difference between the powder output per unit time of the powder distributor and the powder output per unit time of the pressure roller is then calculated as the powder reserve per unit time of the pressure roller. If the powder reserve per unit time of the pressure roller is not within the preset reserve range, steps S1 to S5 are repeated until the powder reserve per unit time of the pressure roller falls within the preset reserve range. The powder output per unit time of the pressure roller is obtained by multiplying the current powder height data of the pressure roller, the width of the powder on the pressure roller, the conveying speed of the pressure roller, and the density of the powder. The width of the powder on the pressure roller and the conveying speed of the pressure roller are fixed values ​​determined according to the settings of the pressure roller. The powder output per unit time of the powder distributor is obtained by multiplying the rotational speed of the servo motor of the powder distributor, the rated parameters of the powder distributor (such as the volume conveyed per revolution of the servo motor in the powder distributor), and the density of the powder.

[0036] Optionally, in step S1, a data processing method is used to filter the pressure roller powder height data to obtain smoothed pressure roller powder height data.

[0037] Preferably, in step S1, the step of obtaining the smoothed roller powder height data includes: arranging the roller powder height data by time to obtain the original sequence; The median filter of the NumPy library is used to select a neighborhood of size 5 from the original sequence. The median value is calculated for each point and replaced with the original value to obtain a preliminary sequence of powder height. The moving average filter from the NumPy library is used to select a neighborhood of size 10 from the initial sequence, calculate the average value for each point, and generate smoothed roller powder height data.

[0038] In one possible implementation, when acquiring powder height data from sensors on the pressure rollers, a laser sensor could be deployed on the production line. This laser sensor is mounted above the pressure rollers and monitors changes in powder accumulation height in real time. For example, the laser sensor collects data once per second, forming a time series. During production, the powder height might fluctuate from an initial 20 cm to 25 cm, then drop back to 18 cm; these data points constitute the raw sequence. In this way, the raw sequence captures the immediate fluctuations in powder distribution, including noise caused by uneven feeding, thus providing fundamental data support for subsequent processing.

[0039] In one possible implementation, the median filter in the NumPy library is a non-linear filtering method. It selects a sliding window in the sequence, sorts the data points within the window, and then replaces the current point with the median value. This effectively removes isolated noise points without overly smoothing the signal's edge features. Specifically, for a neighborhood with a window size of 5, assuming a segment in the original sequence is [22,24,50,23,21], where 50 might be a noise point, the sorted result is [21,22,23,24,50], with a median of 23. Therefore, the 5 numbers in the original sequence are replaced with 23. After this processing, abnormal peaks in the entire sequence are suppressed, resulting in a preliminary powder height sequence. This sequence more closely approximates the true powder height trend, avoiding erroneous readings caused by sensor jitter or dust interference.

[0040] The principle of moving average filtering is a linear method. It generates new points by calculating the arithmetic mean of the data within a window, thereby reducing random fluctuations and highlighting the overall trend. Specifically, for a neighborhood with a window size of 10, assuming a segment of the initial sequence of powder height is [20,21,22,21,20,19,20,21,22,23], the average value of the first window is (20+21+22+21+20+19+20+21+22+23) / 10=20.9. Then, the sliding window is used to calculate the next average value, and so on, to generate the smoothed roller powder height data.

[0041] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and these variations still fall within the protection scope of the present invention.

Claims

1. A method of optimizing the matching of the height of the press roll powder with the speed of the powder feeder, characterized in that, The method comprises the following steps: S1: obtaining the pressure roller powder height data and the current change signal of the pressure roller motor; S2: calculating the change trend of the powder height according to the pressure roller powder height data; S3: if the absolute value of the change trend exceeds the preset height trend threshold, determining the use speed fluctuation index of the powder based on the current change signal of the pressure roller motor; S4: adjusting the use speed fluctuation index of the powder by using a feedback control algorithm to obtain a target value of the discharging speed; S5: extracting speed deviation data according to the difference between the target value of the discharging speed and the actual discharging speed, and processing the speed deviation data by using a logical reasoning algorithm to obtain a speed control instruction.

2. The method of claim 1, wherein the method is characterized in that: In the step S2, the height value sequence at continuous time points is obtained from the pressure roller powder height data; The height value sequence is divided into a plurality of time period sequences with fixed lengths by using a fixed window, and a corresponding difference operation is performed in each time period sequence, wherein the difference operation is to subtract the height values at adjacent time points to obtain a difference value, and then the difference value is arranged according to time to obtain a local height change difference value sequence; The local height change difference value sequence is fitted with a trend line by using the LinearRegression class of the sklearn library, wherein the difference values in the local height change difference value sequence are taken as a y array, and the time points corresponding to the difference values are taken as an x array, a linear model is solved by calling the fit method, and a fitted trend line equation y = mx + b is output, wherein m is a local trend slope value, x is a time point, y is a difference value, and b is an intercept value; The average of all local trend slope values is calculated to obtain the change trend.

3. The method of claim 1, wherein the method further comprises: determining a target powder height for the powder roll; and adjusting the speed of the powder roll to maintain the target powder height. In the step S3, the step of analyzing based on the current change signal of the pressure roller motor comprises: The time sequence deviation of the current change signal of the pressure roller motor and a preset uniformity reference signal is compared to obtain a deviation value, and the deviation value is taken as the use speed fluctuation index of the powder at each time point.

4. The method of claim 3, wherein the method further comprises: determining the powder height of the powder roll; and adjusting the speed of the powder dispenser based on the determined powder height of the powder roll. In the step S3, if the absolute value of the change trend exceeds the preset height trend threshold, the current sequence of a continuous time period is obtained from the current change signal of the pressure roller motor; The current sequence is frequency domain converted by using the fast Fourier transform function of the NumPy library to obtain a frequency spectrum distribution feature; The amplitude values higher than a preset frequency threshold are selected from the frequency spectrum distribution feature to form a high frequency component sequence; If the average of the high frequency component sequence exceeds a preset upper limit value, the use speed fluctuation index of the powder is determined based on the current change signal of the pressure roller motor.

5. The method of claim 1, wherein the method further comprises: determining a powder height of the powder roll; and adjusting the speed of the powder dispenser based on the determined powder height of the powder roll. In the step S4, the use speed fluctuation index of the powder is arranged according to time to obtain a deviation value sequence; The deviation value sequence is adjusted by using a standard PID controller to obtain a discharging speed adjustment parameter, wherein the standard PID controller inputs the deviation value sequence, multiplies the current deviation value by a proportional term Kp, multiplies the integral value of the deviation value sequence by an integral term Ki, and multiplies the change rate of the deviation value by a differential term Kd, and then sums the three to obtain the discharging speed adjustment parameter, wherein Kp is a proportional gain constant, Ki is an integral gain constant, and Kd is a differential gain constant; The target value of the discharging speed is obtained by multiplying the discharging speed adjustment parameter by the average of the use speed fluctuation index of the powder.

6. The method of claim 1, wherein the method further comprises: determining a powder height of the powder roll; and adjusting the speed of the powder dispenser based on the determined powder height of the powder roll. In the step S5, a speed deviation data is calculated by calculating the difference between the target value of the discharging speed and the actual discharging speed; The speed deviation data is processed by a conditional statement of Python to obtain a speed control instruction, wherein the conditional statement of Python inputs the speed deviation data and outputs the speed control instruction based on a preset rule set.

7. The method of claim 1, wherein the method further comprises: determining a powder height of the powder roll; and adjusting the speed of the powder dispenser based on the determined powder height. In the step S1, the data processing method is used to filter the powder height data of the compression roller to obtain smoothed powder height data of the compression roller.

8. The method of claim 7, wherein the method further comprises: determining a target powder height for the powder roller; and adjusting the speed of the powder dispenser to maintain the target powder height. In the step S1, the step of obtaining the smoothed powder height data of the compression roller includes: arranging the powder height data of the compression roller according to time to obtain an original sequence; A median filter of a NumPy library is used to select a neighborhood with a window size of 5 from the original sequence, calculate a median value for each point to replace the original value, and obtain a preliminary sequence of the powder height; A moving average filter of the NumPy library is used to select a neighborhood with a window size of 10 from the preliminary sequence, calculate an average value for each point, and generate the smoothed powder height data of the compression roller.

Images