Powder and particle material transport semitrailer unloading control method and system

Abstract

This invention relates to the field of industrial vision technology, specifically to a method and system for controlling the unloading of a semi-trailer transporting powdery materials. The method includes the following steps: acquiring grayscale image sequences to construct a visualized fluid feature dataset; calculating the peak angle drift rate and comparing it with a safety benchmark to generate a safety judgment record; combining the safety record with frequency domain features to obtain flow energy distribution data; outputting purging control data based on the degree of deviation of the center of gravity coefficient from the energy efficiency target; and calculating the conveying impedance sensitivity in conjunction with air intake disturbance to search for the efficiency limit point. In this invention, the fluid movement trend is quantified through angle drift rate calculation, a safety benchmark is established to identify flow continuity, the flow center of gravity is analyzed using a frequency domain energy spectrum, the purging valve is automatically adjusted according to the energy efficiency target range to eliminate proportional imbalance, and the limit point approximation search is performed in conjunction with impedance sensitivity analysis to achieve fully automatic closed-loop optimization control, thereby improving the stability of unloading operations and energy utilization efficiency.

Description

Technical Field

[0001] This invention relates to the field of industrial vision technology, and in particular to a method and system for controlling the unloading of a semi-trailer used for transporting powdery or granular materials. Background Technology

[0002] Industrial vision technology encompasses the engineering field of acquiring image information of physical objects using optical imaging sensors and combining it with computer vision algorithms for analysis. Its core aspects include the design of light source illumination systems, the selection and matching of optical lenses, the digital transmission of image acquisition cards, and specific processing steps such as grayscale transformation, edge extraction, and feature matching of the image pixel matrix. This field captures the surface morphology data of target objects by constructing imaging units composed of industrial cameras, lenses, and light sources. Based on preset geometric parameters or defect sample libraries, it performs dimensional measurement, appearance defect identification, and real-time spatial positioning of products. This includes traditional powder and granular material transportation... Trailer unloading control method refers to the process adjustment means for tank trucks loaded with bulk powder materials during unloading operations. This method mainly relies on on-site operators to obtain the air pressure value inside the tank by observing the pointer position of the mechanical pressure gauge installed on the outside of the tank, and manually rotating and adjusting the opening of the main air inlet valve connected to the on-board air compressor to change the compressed air flow into the fluidized bed at the bottom of the tank based on manual experience. At the same time, the operator judges the material conveying status by visually observing the vibration amplitude of the discharge hose or listening to the friction sound of the material flowing in the pipeline, and then manually turns the handle of the discharge butterfly valve to control the opening angle of the discharge port and manually adjusts the ball valve of the secondary auxiliary blowing pipeline to maintain the conveying of gas-solid mixed flow.

[0003] Traditional unloading operations rely on manual experience to observe the swing amplitude of the mechanical pressure gauge pointer and the vibration frequency of the discharge hose. Subjective judgment lacks precise quantitative standards, resulting in a serious lag in flow perception. It is difficult to capture the transient changes in the gas-solid two-phase flow in real time. Relying on auditory identification of friction sounds is easily affected by on-site noise, leading to misjudgments. Manual adjustment of the air inlet valve and the blowing pipeline cannot accurately match the fluidization requirements of the material. Coarse control is prone to pipeline blockage or air wastage. Frequent misoperation results in low unloading efficiency and significant energy loss. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a method for controlling the unloading of a semi-trailer transporting powdery or granular materials, comprising the following steps:

[0005] S1: Collect a sequence of grayscale images of gas-solid two-phase flow, crop the central region of the sequence to obtain pixel columns, and stitch the pixel columns together to form a two-dimensional pixel matrix to construct a visualized fluid feature dataset containing spatiotemporal slice image data.

[0006] S2: Call the visualized fluid feature dataset, perform projection transformation on the spatiotemporal slice image data, calculate the extreme point angle differential result to obtain the peak angle drift rate, and compare it with the safety inertial benchmark to generate a transport safety judgment record;

[0007] S3: Call the transport safety determination record to determine continuity, combine the visualized fluid feature dataset, calculate the gray mean of the gas-solid two-phase flow grayscale image sequence to construct a time series signal, obtain the frequency domain energy distribution spectrum through frequency domain conversion, calculate the full-band energy ratio of low-frequency components, and obtain flow energy distribution data;

[0008] S4: Extract the flow energy centroid coefficient from the flow energy distribution data, compare it with the preset optimal energy efficiency target range, obtain the blowing valve adjustment configuration according to the degree of deviation, and output the blowing flow regulation data.

[0009] S5: Call the blowing flow state control data to determine the execution status, combine the visualized fluid feature dataset, and calculate the material ratio and opening change ratio of the gas-solid two-phase flow grayscale image sequence with the inlet valve disturbance signal to obtain the conveying impedance sensitivity, generate the optimal configuration and search for the efficiency limit point, and output the inlet limit search data.

[0010] As a further aspect of the present invention, the visualized fluid feature dataset specifically comprises an original grayscale image sequence, a centrally cropped pixel column, and a two-dimensional spatiotemporal slice matrix; the conveying safety judgment record includes a parameter space projection matrix, a peak angle drift rate value, a rigid blockage cutoff marker, and a flow continuity confirmation marker; the flow energy distribution data specifically comprises a regional grayscale mean time series signal, a frequency domain energy distribution spectrum, a total energy integral in the low-frequency range, and a weighted proportion of energy integral across the entire frequency band; the blowing flow regulation data includes the deviation value of the optimal energy efficiency target range, the step adjustment configuration parameters of the blowing valve, the gas-solid ratio compensation command, and the fluidization state stability control signal; and the inlet limit search data specifically refers to a periodic opening disturbance signal, a material pixel ratio change rate, a conveying impedance sensitivity value, and inlet valve optimization control configuration parameters.

[0011] As a further aspect of the present invention, the steps for obtaining the visualized fluid feature dataset are specifically as follows:

[0012] S101: Monitor the real-time flow status of gas-solid two-phase flow in the semi-trailer discharge sight glass area, collect a series of grayscale images of gas-solid two-phase flow in a continuous time period, traverse each frame of the gas-solid two-phase flow grayscale image sequence, calculate the geometric center coordinates of the image plane, and use them as a reference to delineate a central vertical region with a width of one pixel, perform a cropping operation and extract the grayscale values ​​of all pixels in the region to generate a central vertical cropping pixel column.

[0013] S102: Call the central vertical cropping pixel column, read the acquisition time timestamp corresponding to each pixel column, and perform splicing processing on all pixel columns in the horizontal direction according to the increasing order of timestamp values. Establish a two-dimensional coordinate system with time dimension as the horizontal axis and spatial position as the vertical axis. Map and fill the spliced ​​pixel gray values ​​into the coordinate system to form a two-dimensional pixel matrix and generate a two-dimensional spatiotemporal slice matrix.

[0014] S103: Call the two-dimensional spatiotemporal slice matrix, standardize and format the matrix data, construct spatiotemporal slice image data containing fluid motion texture features, establish the correlation mapping relationship between the data and the current semi-trailer discharge condition parameters, integrate the spatiotemporal slice image data and the correlation mapping relationship, and generate a visualized fluid feature dataset.

[0015] As a further aspect of the present invention, the step of obtaining the transport safety determination record specifically includes:

[0016] S201: Based on the visualized fluid feature dataset, extract spatiotemporal slice image data, set the projection angle scanning sequence, perform linear integral projection transformation on the spatiotemporal slice image data, accumulate pixel gray values ​​along the projection direction to calculate path energy intensity, map the energy intensity to the corresponding coordinate points in the parameter transformation domain, and generate a parameter space projection matrix.

[0017] S202: Extract the parameter space projection matrix, retrieve the coordinates of the extreme point with the highest energy accumulation in the matrix, extract the angle parameter components corresponding to the extreme point coordinates, combine the angle parameter components of the previous acquisition cycle, calculate the differential change rate of the current angle parameter components relative to the acquisition time, and obtain the peak angle drift rate.

[0018] S203: Using the peak angle drift rate, read the preset safety inertia benchmark, perform a numerical comparison between the peak angle drift rate and the safety inertia benchmark, establish rigid blockage cutoff markers and flow continuity judgment markers based on the comparison results, encapsulate the established marker data, and generate a transport safety judgment record.

[0019] As a further aspect of the present invention, the process of obtaining the preset safety inertial reference is specifically as follows:

[0020] The historical conveying operation log is retrieved, and historical periods marked as stable discharge states are filtered out. The gas-solid two-phase flow grayscale image sequence within the corresponding period is extracted as a benchmark verification sample. Following the calculation path of the peak angle drift rate, the historical angle drift rate set of the benchmark verification sample is obtained. The arithmetic mean and standard deviation of the set are calculated. The sum of the arithmetic mean and three times the standard deviation is selected as the basic fluctuation limit. The bulk density data and particle repose angle data of the conveyed material are collected in real time. Normalization weighted operation is performed on the bulk density data and particle repose angle data to obtain the working condition correction factor for the current material characteristics. The basic fluctuation limit and the working condition correction factor are multiplied to obtain the threshold value and set as the preset safety inertia benchmark.

[0021] As a further aspect of the present invention, the step of acquiring the flow energy distribution data specifically includes:

[0022] S301: Obtain the transport safety judgment record and parse the flow continuity judgment mark, call the visualized fluid feature dataset to extract the gas-solid two-phase flow grayscale image sequence, calculate the pixel grayscale arithmetic mean of each frame image in the sequence, arrange all arithmetic means according to the acquisition timestamp order to construct a one-dimensional numerical sequence, and generate a grayscale time series signal.

[0023] S302: Call the grayscale time-series signal, perform a discrete orthogonal transformation operation from the time domain to the frequency domain, analyze each frequency basis function component contained in the signal, calculate the square value of the amplitude modulus of each component as the energy intensity of the corresponding frequency, establish a mapping relationship table between frequency values ​​and energy intensity values, and generate a frequency domain energy distribution spectrum.

[0024] S303: Analyze the frequency domain energy distribution spectrum, extract the frequency components within the preset low frequency range, perform integral summation on the energy intensity of the extracted components to obtain the total low frequency energy value, calculate the integral sum of the energy intensity of the frequency components across the entire frequency band, calculate the weight of the total low frequency energy value in the integral sum, integrate the weight with the energy statistics results, and generate flow energy distribution data.

[0025] As a further aspect of the present invention, the process of obtaining the preset low-frequency range specifically includes:

[0026] During the fluidization calibration stage, a sequence of grayscale images of the gas-solid two-phase flow under stable delivery conditions is acquired. A discrete orthogonal transform is performed to generate a standard frequency domain energy distribution spectrum. Gaussian smoothing filtering is applied to the standard frequency domain energy distribution spectrum. The frequency coordinates corresponding to the maximum energy intensity are retrieved as the center frequency of the main peak. The energy distribution curve to the right of the center frequency of the main peak is scanned along the frequency axis in the positive direction. The frequency position where the energy intensity decays to 50% of the main peak energy intensity is identified and marked as the half-power frequency point. The difference between the half-power frequency point and the center frequency of the main peak is calculated to obtain the single-side bandwidth of the main peak. The sum of the center frequency of the main peak and 4 times the single-side bandwidth of the main peak is calculated as the cutoff frequency. A closed interval is constructed with zero Hz as the starting point and the cutoff frequency as the ending point. The target closed interval is set as the preset low-frequency range.

[0027] As a further aspect of the present invention, the step of acquiring the blowing flow regime control data specifically includes:

[0028] S401: Using the fluid energy distribution data, analyze the frequency domain energy integral ratio weight, calculate the concentration trend of fluid kinetic energy frequency domain distribution, obtain the fluid energy centroid coefficient, read the preset optimal energy efficiency target interval, perform the difference comparison operation between the coefficient value and the interval boundary, calculate the deviation difference, quantify the degree and direction of deviation, and generate an energy efficiency target deviation vector.

[0029] S402: Based on the energy efficiency target deviation vector, determine the adjustment polarity of the blowing air path according to the deviation direction, match the valve control parameters based on the degree of deviation, determine the valve opening step value and adjustment frequency, construct a compensation relationship for the gas-solid ratio imbalance state, establish a control configuration including valve action sequence and flow regulation parameters, and generate the blowing valve step adjustment configuration.

[0030] S403: Invoke the step adjustment configuration of the blowing valve, convert it into an electrical drive signal and apply it to the blowing valve actuator to dynamically adjust the on / off state and flow path of the blowing air path, collect the status feedback data during the execution process and encapsulate it with the control command to generate blowing flow state control data.

[0031] As a further aspect of the present invention, the process of setting the optimal energy efficiency target range is specifically as follows:

[0032] The historical operation log of the gas-solid two-phase flow conveying system is retrieved, and synchronous correlation records containing the flow energy centroid coefficient, inlet flow rate, and outlet flow rate are extracted. For each synchronous correlation record, a division operation is performed between the outlet flow rate and the inlet flow rate to obtain the single-point gas-solid ratio energy efficiency value. A mapping set between the flow energy centroid coefficient and the single-point gas-solid ratio energy efficiency value is constructed. Polynomial curve fitting is performed on the mapping set to generate a trend curve reflecting the change of energy efficiency with the centroid coefficient. The vertex position with the highest energy efficiency value in the trend curve is retrieved, and the ordinate corresponding to the vertex position is extracted as the global optimal energy efficiency value. The value 0.9 is selected as the effective coefficient, and a multiplication operation is performed between the global optimal energy efficiency value and the effective coefficient to obtain the energy efficiency boundary threshold. Continuous segments in the trend curve with energy efficiency values ​​higher than the energy efficiency boundary threshold are identified, and the distribution range of the flow energy centroid coefficient corresponding to the target continuous segment is extracted. The minimum and maximum boundary values ​​of the distribution range are set as the lower and upper limits of the target interval, respectively, to generate the optimal energy efficiency target interval.

[0033] As a further aspect of the present invention, the step of obtaining the intake limit search data specifically includes:

[0034] S501: Call the blowing flow state control data to verify the control command execution status, generate a periodic opening disturbance signal and apply it to the main intake valve, combine the visualized fluid feature dataset to extract the gas-solid two-phase flow grayscale image sequence within the disturbance time window, segment and calculate the area ratio of the material pixel region frame by frame, construct the time-series correlation pair between valve opening and material ratio, and generate a dynamic disturbance response sequence.

[0035] S502: Based on the dynamic disturbance response sequence, extract the peak-to-trough difference of the periodic opening disturbance signal as the valve opening change amplitude, simultaneously calculate the extreme value difference of the material proportion data in the corresponding time period as the material pixel proportion change amplitude, perform the ratio calculation of the material pixel proportion change amplitude and the valve opening change amplitude, analyze the flow response characteristics, and generate the conveying impedance sensitivity.

[0036] S503: Utilizing the aforementioned transmission impedance sensitivity, detect the gradient polarity of the sensitivity value, locate the relative position of the optimal operating point, analyze the response intensity, adapt the corresponding gradient descent step size, generate an intake valve optimization control configuration containing dynamic adjustment logic, drive the intake valve to perform approximation search and steady-state maintenance of the physical limit operating point, and generate intake limit search data.

[0037] A semi-trailer unloading control system for transporting powdery materials includes:

[0038] The image preprocessing module acquires a sequence of grayscale images of gas-solid two-phase flow, crops the central region of the sequence to obtain pixel columns, splices the pixel columns to form a two-dimensional pixel matrix, and constructs a visualized fluid feature dataset containing spatiotemporal slice image data.

[0039] The safety analysis module calls the visualized fluid feature dataset, performs projection transformation on the spatiotemporal slice image data, calculates the extreme point angle differential result to obtain the peak angle drift rate, compares it with the safety inertial benchmark, and generates a transport safety judgment record.

[0040] The frequency domain analysis module calls the transport safety judgment record to determine continuity, combines the visualized fluid feature dataset, calculates the gray mean of the gas-solid two-phase flow grayscale image sequence to construct a time series signal, obtains the frequency domain energy distribution spectrum through frequency domain transformation, calculates the full-band energy ratio of low-frequency components, and obtains flow energy distribution data.

[0041] The control and configuration module calls the flow energy distribution data to extract the flow energy centroid coefficient, compares it with the preset optimal energy efficiency target range, obtains the blowing valve adjustment configuration according to the degree of deviation, and outputs the blowing flow control data.

[0042] The optimization search module calls the blowing flow state control data to determine the execution status, combines the visualized fluid feature dataset, and calculates the material ratio and opening change ratio of the gas-solid two-phase flow grayscale image sequence with the inlet valve disturbance signal to obtain the conveying impedance sensitivity, generates the optimization configuration and searches for the efficiency limit point, and outputs the inlet limit search data.

[0043] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0044] In this invention, a visual feature dataset is constructed by collecting gas-solid two-phase flow images to replace human sensory experience. The fluid motion trend is accurately quantified by using projection transformation and angle drift rate calculation. A safe inertial benchmark is established to identify the flow continuity in real time and warn of blockage risks. The flow energy centroid coefficient is analyzed by combining frequency domain energy distribution spectrum. Based on the energy efficiency target range, the blowing valve adjustment strategy is automatically generated to eliminate gas-solid ratio imbalance. The conveying impedance sensitivity is calculated in conjunction with the inlet disturbance signal and the limit point approximation search is performed. Under the premise of ensuring conveying safety, fully automatic closed-loop optimization control is achieved, improving the stability of unloading operations and energy utilization efficiency. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of the steps of the present invention;

[0047] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0048] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0049] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0050] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0051] Figure 6 This is a detailed schematic diagram of S5 of the present invention;

[0052] Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0053] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0054] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0055] Please see Figure 1 This invention provides a method for controlling the unloading of a semi-trailer used for transporting powdery materials, comprising the following steps:

[0056] S1: Collect a sequence of grayscale images of gas-solid two-phase flow, crop the central region of the sequence to obtain pixel columns, and stitch the pixel columns together to form a two-dimensional pixel matrix to construct a visualized fluid feature dataset containing spatiotemporal slice image data.

[0057] S2: Call the visualized fluid feature dataset, perform projection transformation on the spatiotemporal slice image data, calculate the extreme point angle differential result to obtain the peak angle drift rate, compare it with the safety inertial benchmark, and generate a transport safety judgment record;

[0058] S3: Call the transport safety judgment record to judge continuity, combine with the visualized fluid feature dataset, calculate the gray mean of the gas-solid two-phase flow grayscale image sequence to construct a time series signal, obtain the frequency domain energy distribution spectrum through frequency domain transformation, calculate the full-band energy ratio of low frequency components, and obtain the flow energy distribution data;

[0059] S4: Extract the centroid coefficient of flow energy from the flow energy distribution data, compare it with the preset optimal energy efficiency target range, obtain the blowing valve adjustment configuration according to the degree of deviation, and output the blowing flow regulation data.

[0060] S5: Call the bleed flow regulation data to determine the execution status, combine the visualized fluid feature dataset, and calculate the material ratio and opening change ratio of the gas-solid two-phase flow grayscale image sequence with the inlet valve disturbance signal to obtain the conveying impedance sensitivity, generate the optimal configuration and search for the efficiency limit point, and output the inlet limit search data.

[0061] The visualized fluid feature dataset specifically includes the original grayscale image sequence, the central vertically cropped pixel column, and the two-dimensional spatiotemporal slice matrix. The conveying safety judgment record includes the parameter space projection matrix, the peak angle drift rate value, the rigid blockage cutoff mark, and the flow continuity confirmation mark. The flow energy distribution data specifically includes the regional grayscale mean time series signal, the frequency domain energy distribution spectrum, the total energy integral in the low-frequency range, and the weight of the energy integral ratio across the entire frequency band. The blowing flow regulation data includes the deviation value of the optimal energy efficiency target range, the step adjustment configuration parameters of the blowing valve, the gas-solid ratio compensation command, and the fluidization state stability control signal. The inlet limit search data specifically refers to the periodic opening disturbance signal, the material pixel ratio change rate, the conveying impedance sensitivity value, and the inlet valve optimization control configuration parameters.

[0062] Please see Figure 2 The specific steps for obtaining the visualized fluid feature dataset are as follows:

[0063] S101: Monitor the real-time flow status of gas-solid two-phase flow in the semi-trailer discharge sight glass area, collect a series of grayscale images of gas-solid two-phase flow in a continuous time period, traverse each frame of the gas-solid two-phase flow grayscale image sequence, calculate the geometric center coordinates of the image plane, and use them as a reference to delineate a central vertical region with a width of one pixel, perform a cropping operation and extract the grayscale values ​​of all pixels in the region to generate a central vertical cropping pixel column.

[0064] The process of monitoring the real-time flow state of the gas-solid two-phase flow within the sight glass area of ​​a semi-trailer first utilizes an industrial high-speed camera to continuously convert photoelectric signals through the physical window of the sight glass. This process, sampling at 60 frames per second, dynamically converts the gas-solid mixture within the sight glass into a digital grayscale signal sequence. The acquisition duration is set to 5 seconds, resulting in 300 frames of raw grayscale images. For each of these 300 frames of gas-solid two-phase flow grayscale image data (1024 x 1024 pixels), a geometric center positioning operation is performed. By reading the row and column numbers of the image matrix and dividing each row and column number by 2, the geometric center coordinates of the image plane are precisely located as x-coordinate 512 and y-coordinate 512. Using this central coordinate as a reference, the clipping logic extends 0 pixels to the left and right along the horizontal axis, retaining only the pixel data in the column with a horizontal coordinate of 512. Vertically, it retains the full length of pixels from a vertical coordinate of 1 to 1024, thus defining a central vertical region with a width of 1 pixel and a height of 1024 pixels. Subsequently, the extraction logic reads the grayscale values ​​of the pixels within this region, obtaining the brightness value of each pixel in the column. These values ​​are distributed within the grayscale range of 0 to 255; for example, the grayscale value of the pixel in row 100 is 128, and the grayscale value of the pixel in row 500 is 200, generating a central vertical clipped pixel column consisting of 1024 grayscale values. This clipping operation effectively eliminates noise interference caused by uneven illumination or pipe wall reflections at the edge of the sight glass, focusing only on the flow characteristics of the fluid core region, providing high signal-to-noise ratio foundational data for subsequent spatiotemporal reconstruction.

[0065] S102: Call the center vertical cropping pixel column, read the acquisition time timestamp corresponding to each pixel column, and stitch all pixel columns in the horizontal direction according to the increasing order of timestamp values. Establish a two-dimensional coordinate system with time dimension as the horizontal axis and spatial position as the vertical axis. Map and fill the stitched pixel gray values ​​into the coordinate system to form a two-dimensional pixel matrix and generate a two-dimensional spatiotemporal slice matrix.

[0066] After acquiring 300 sets of center-cropped pixel columns, this pixel column data is retrieved, and the hardware system timestamp corresponding to the acquisition time of each pixel column is read synchronously. For example, the timestamp of the first frame is 1000 milliseconds, the second frame is 1016 milliseconds, and so on up to the 300th frame. Based on the ascending order of the timestamp values, the stitching logic performs tight processing on all 300 pixel columns in the horizontal direction, constructing a new two-dimensional data structure. In this process, a two-dimensional coordinate system is established with time as the horizontal axis and spatial position as the vertical axis. The horizontal axis length is set to 300 units, corresponding to the 300 acquisition times, and the vertical axis length is set to 1024 units, corresponding to the vertical spatial height of the viewport window. The filling logic maps the 1024 grayscale values ​​in the vertically cropped pixel column corresponding to the first frame sequentially to a column vector with an x-coordinate of 1 in a two-dimensional coordinate system; the pixel column corresponding to the second frame is filled to a column vector with an x-coordinate of 2, and so on, until the data of the 300th frame is filled to a column vector with an x-coordinate of 300. Through this filling process, the originally discrete time-series image is transformed into a continuous two-dimensional spatiotemporal slice matrix. The value of each point (x, y) in this matrix represents the fluid grayscale value at the y-th position at the vertical height of the viewing mirror at the x-th sampling time. For example, the value of 150 at coordinates (50, 512) means that at the 50th sampling time, the gas-solid fluid at the center height of the viewing mirror exhibits an optical feature with a grayscale value of 150. Through dimensionality reduction and reconstruction, the three-dimensional video stream is compressed into a two-dimensional image containing the flow texture of the fluid throughout its entire life cycle, intuitively revealing the trajectory characteristics of particle aggregates flowing over time.

[0067] S103: Call the two-dimensional spatiotemporal slice matrix, standardize and format the matrix data, construct spatiotemporal slice image data containing fluid motion texture features, establish the correlation mapping relationship between the data and the current semi-trailer discharge condition parameters, integrate the spatiotemporal slice image data and the correlation mapping relationship, and generate a visualized fluid feature dataset.

[0068] The generated 2D spatiotemporal slice matrix with dimensions of 1024 rows by 300 columns is invoked. First, the original grayscale data within the matrix is ​​standardized and formatted to eliminate the impact of light intensity fluctuations on feature analysis. Specifically, each grayscale value in the matrix is ​​iterated, and the arithmetic mean and standard deviation of all 307,200 pixels are calculated (e.g., the arithmetic mean is 110 and the standard deviation is 35). Then, for each pixel in the matrix, its original grayscale value is subtracted from the arithmetic mean of 110, and the difference is divided by the standard deviation of 35. This converts the original integer grayscale data (0 to 255) into floating-point values ​​following a standard normal distribution, completing the Z-score standardization process and constructing spatiotemporal slice image data containing fluid motion texture features. Simultaneously, the data interface reads real-time parameters of the current semi-trailer discharge condition via an industrial fieldbus, including discharge port pressure, bleed valve opening, and tank tilt angle. For example, the current outlet pressure is 0.25 MPa, the bleeder valve opening is 45 degrees, and the tank tilt angle is 5 degrees. The correlation logic establishes a one-to-one mapping relationship between these operating parameters and standardized spatiotemporal slice image data. This means that the operating parameters are appended as metadata tags to the image data file header, integrating them to form a visualized fluid feature dataset. By integrating the spatiotemporal slice image data with the correlation mapping relationship, not only is the visual texture of fluid motion included, but also the physical boundary conditions driving this motion state are incorporated, providing complete data support for subsequent multimodal analysis.

[0069] Please see Figure 3 The specific steps for obtaining the safety assessment record are as follows:

[0070] S201: Based on the visualized fluid feature dataset, extract spatiotemporal slice image data, set the projection angle scanning sequence, perform linear integral projection transformation on the spatiotemporal slice image data, accumulate pixel gray values ​​along the projection direction to calculate path energy intensity, map the energy intensity to the corresponding coordinate points in the parameter transformation domain, and generate a parameter space projection matrix.

[0071] Based on the visualized fluid feature dataset, standardized spatiotemporal slice image data is extracted, and a projection angle scanning sequence is established. This scanning sequence covers an angle range from 0 degrees to 179 degrees, with a step size of 1 degree, totaling 180 projection directions. The computational logic performs a linear integral projection transformation (Radon transform) on the spatiotemporal slice image data. For each set projection angle, a set of virtual rays parallel to that angle is generated on the image plane, and the sum of the standardized grayscale values ​​of all pixels along each ray is calculated. This sum represents the energy intensity along that path. For example, when the scanning angle is 45 degrees, the image is integrated along the 45-degree direction to obtain the energy distribution curve at that angle; when the scanning angle is 90 degrees, integration is performed along the vertical direction. Subsequently, the mapping logic maps all the calculated path energy intensity values ​​at the 180 angles to a parametric transformation domain coordinate system with angle as the horizontal axis and distance as the vertical axis, generating a parameter space projection matrix. By mapping to the parameter space, the value of each coordinate point (θ, ρ) in the matrix reflects the texture intensity of the original image along the line defined by angle θ and distance ρ. If there are obvious striped flow characteristics (such as particle bands) in the gas-solid two-phase flow, then bright spots with energy accumulation will appear at the corresponding angular positions in the parameter space projection matrix.

[0072] S202: Extract the parameter space projection matrix, retrieve the coordinates of the extreme point with the highest energy accumulation in the matrix, extract the angle parameter components corresponding to the extreme point coordinates, combine the angle parameter components of the previous acquisition cycle, calculate the differential change rate of the current angle parameter components relative to the acquisition time, and obtain the peak angle drift rate.

[0073] The generated parameter space projection matrix is ​​extracted, and an extreme value search operation is performed across the entire matrix. The energy values ​​of each cell in the matrix are compared to pinpoint the coordinates of the extreme point with the highest accumulated energy. Assuming the retrieved extreme point coordinates are (35 degrees, 120), where 35 degrees represents the angular parameter component corresponding to the extreme point, indicating the dominant direction of the current fluid texture. Combining this with the angular parameter component recorded in the previous acquisition cycle (e.g., 0.5 seconds ago), assuming the angle at the previous moment was 33 degrees, the differential calculation logic performs a difference operation. Specifically, the calculation process is as follows: subtract the angle value of 33 degrees from the previous moment's angle value of 35 degrees to obtain an angle change of 2 degrees; the time interval between the two acquisition moments is read, assuming it is 0.5 seconds; the angle change of 2 degrees is divided by the time interval of 0.5 seconds to calculate the peak angle drift rate of 4 degrees per second. By obtaining the peak angle drift rate, the drastic dynamic change in the fluid flow direction over time is quantified. A larger drift rate indicates a more unstable flow state or the presence of turbulent disturbances, achieving a rapid quantitative characterization of the dynamic stability of the flow field.

[0074] S203: Using the peak angle drift rate, read the preset safety inertia reference, perform a comparison between the peak angle drift rate and the safety inertia reference, establish rigid blockage cutoff mark and flow continuity judgment mark according to the comparison result, encapsulate the established mark data, and generate a conveying safety judgment record.

[0075] Using the calculated peak angle drift rate (e.g., 4 degrees per second), a preset safety inertial benchmark is read for numerical comparison. The process of obtaining the preset safety inertial benchmark is as follows: Historical conveying operation logs in memory are retrieved, and historical periods manually marked as "stable discharge" are selected. A 1-hour-long grayscale image sequence of gas-solid two-phase flow within this period is extracted as a benchmark verification sample. The historical angle drift rate of this sample at each moment is calculated according to the aforementioned steps, forming a set containing 3600 data points. The arithmetic mean of this set is calculated to be 1.5 degrees per second, and the standard deviation is 0.5 degrees per second. The sum of the arithmetic mean 1.5 and three times the standard deviation (i.e., 1.5), i.e., 3.0 degrees per second, is selected as the basic fluctuation limit. Simultaneously, the sensor collects the physical properties of the currently conveyed material in real time, measuring a bulk density of 800 kg / m³ and a particle repose angle of 35 degrees. The calculation unit performs a normalized weighted operation on these two parameters: the bulk density of 800 is divided by the standard density of 1000 to obtain 0.8, and the angle of repose of 35° is divided by the standard angle of 45° to obtain 0.77. Each parameter is assigned a weight of 0.5, and the operating condition correction factor is calculated as (0.8 x 0.5) plus (0.77 x 0.5), which equals 0.785. Finally, the basic fluctuation limit of 3.0 is multiplied by the operating condition correction factor of 0.785, resulting in a value of 2.355, which is set as the preset safe inertia benchmark. Returning to the current judgment process, the logic unit compares the real-time calculated peak angle drift rate of 4 degrees per second with the safe inertia benchmark of 2.355 degrees per second. Since 4 is greater than 2.355, the judgment result is "unstable". Based on the comparison result, if the drift rate is extremely low (e.g., close to 0), a rigid blockage cutoff marker is established; if the drift rate is too high, a flow continuity anomaly marker is established. By encapsulating these marker data, a transport safety judgment record is generated.

[0076] Table 1 Historical Operation Data and Safety Benchmark Calculation Table

[0077]

[0078] As shown in Table 1, by introducing material characteristics to correct the historical statistical limits, it is ensured that the safety benchmark can dynamically adapt to the flow characteristics of different batches of materials. The current drift rate of 4.0 exceeds the benchmark of 2.355, which accurately identifies the abnormal flow pattern.

[0079] Please see Figure 4 The specific steps for obtaining fluid energy distribution data are as follows:

[0080] S301: Obtain the safety judgment record of the transport and parse the flow continuity judgment mark, call the visual fluid feature dataset to extract the grayscale image sequence of gas-solid two-phase flow, calculate the arithmetic mean of pixel grayscale of each frame image in the sequence, arrange all arithmetic means in the order of collection timestamp to construct a one-dimensional numerical sequence, and generate grayscale time series signal.

[0081] The system acquires the safety assessment record for the conveying process. The parsing logic first analyzes the flow continuity assessment markers to confirm the presence of rigid blockage (i.e., complete fluid stillness). After confirming a non-blockage state, it calls the visualized fluid feature dataset to extract the original gas-solid two-phase flow grayscale image sequence. For each frame in the sequence, the grayscale values ​​of all pixels are accumulated and divided by the total number of pixels to calculate the arithmetic mean of the grayscale values ​​for that frame. For example, the average grayscale value for the first frame is 120.5, for the second frame it is 121.2, and for the third frame it is 119.8. These average grayscale values ​​are arranged sequentially according to the image acquisition timestamps to construct a one-dimensional numerical sequence that changes over time, thus generating a grayscale time-series signal. By constructing this grayscale time-series signal, the overall fluctuation of material concentration within the viewing area is visually reflected. Higher grayscale values ​​generally represent lower material concentration (higher light transmittance), and vice versa. This compresses the high-dimensional image sequence into a one-dimensional time-series waveform, laying the data foundation for subsequent frequency domain feature analysis.

[0082] S302: Call the grayscale time sequence signal, perform discrete orthogonal transformation operation from time domain to frequency domain, analyze the basis function components of each frequency contained in the signal, calculate the square value of the amplitude modulus of each component as the energy intensity of the corresponding frequency, establish a mapping relationship table between frequency values ​​and energy intensity values, and generate a frequency domain energy distribution spectrum.

[0083] The grayscale time-series signal is called and subjected to a discrete orthogonal transform operation (such as Fast Fourier Transform, FFT) from the time domain to the frequency domain. This operation decomposes the complex time waveform into a series of sinusoidal basis function components of different frequencies. The analytical logic analyzes each basis function component, obtains its corresponding amplitude value, and squares the amplitude value to obtain the squared amplitude modulus of that frequency component, which physically represents the energy intensity of that frequency component. For example, at a frequency of 5 Hz, the calculated amplitude is 10, and its corresponding energy intensity is 100; at a frequency of 50 Hz, the amplitude is 2, and the energy intensity is 4. The traversal logic traverses all frequency points contained in the signal in this way, establishing a mapping table between frequency values ​​(horizontal axis) and energy intensity values ​​(vertical axis), thereby generating a frequency domain energy distribution spectrum. By generating the frequency domain energy distribution spectrum, the frequency structure of fluid pulsation is revealed. For example, the position of the main peak indicates the main pulsation period of the gas-solid two-phase flow, while high-frequency noise corresponds to subtle random disturbances.

[0084] S303: Analyze the frequency domain energy distribution spectrum, extract the frequency components within the preset low frequency range, perform integral summation on the energy intensity of the extracted components to obtain the total low frequency energy value, calculate the integral sum of the energy intensity of the frequency components across the entire frequency band, calculate the weight of the low frequency energy value in the integral sum, integrate the weight with the energy statistics results, and generate flow energy distribution data.

[0085] Analyzing the generated frequency domain energy distribution spectrum, the filtering logic first needs to determine the frequency range for energy extraction, i.e., the preset low-frequency range. The process for obtaining this range is as follows: During the fluidization calibration stage, image sequences under stable transport conditions are acquired and a standard frequency domain energy distribution spectrum is generated. Gaussian smoothing filtering is applied to this spectrum to remove glitches and noise. The highest point of the filtered curve is retrieved, and the frequency coordinate corresponding to the maximum energy intensity is locked, assumed to be 2 Hz, and marked as the center frequency of the main peak. Subsequently, the energy distribution curve is scanned to the right (positive direction) along the frequency axis to find the frequency position where the energy intensity decays to 50% of the maximum value of the main peak (i.e., the half-power point). Assuming that the energy decays by half at 2.5 Hz, 2.5 Hz is marked as the half-power frequency point. The difference between 2.5 Hz and the center frequency of the main peak (2 Hz) is calculated, yielding a single-sideband bandwidth of 0.5 Hz for the main peak. The center frequency of the main peak (2 Hz) is added to four times the single-sideband bandwidth of the main peak (4 multiplied by 0.5 equals 2.0 Hz), resulting in a cutoff frequency of 4.0 Hz. Based on this, a closed interval [0, 4.0] is constructed, starting at 0 Hz and ending at 4.0 Hz, and set as the preset low-frequency range. Returning to the real-time analysis workflow, all frequency components within this range (0 to 4.0 Hz) are extracted, and their energy intensities are integrated and accumulated. Assume the total energy integral value within 0 to 4.0 Hz is 500 joules (normalized units). Simultaneously, the sum of the energy intensities of all frequency components across the entire frequency band (e.g., 0 to 30 Hz) is calculated, assumed to be 600 joules. Finally, a division operation is performed, dividing the total low-frequency energy value of 500 by the total integral sum of 600 across the entire frequency band, yielding a weighting of 0.833. By acquiring flow energy distribution data, a higher value indicates that fluid pulsations are more concentrated in a stable fluidization mode dominated by low frequencies, while a lower value means that high-frequency turbulence or chaotic pulsations account for an excessive proportion.

[0086] Please see Figure 5 The specific steps for obtaining the blowing flow regime control data are as follows:

[0087] S401: Utilize fluid energy distribution data to analyze the frequency domain energy integral weight, calculate the concentration trend of fluid kinetic energy frequency domain distribution, obtain the fluid energy centroid coefficient, read the preset optimal energy efficiency target range, perform a difference comparison operation between the coefficient value and the range boundary, calculate the deviation difference, quantify the degree and direction of deviation, and generate an energy efficiency target deviation vector.

[0088] Using fluid energy distribution data, the frequency domain energy integral weight (such as the aforementioned 0.833) is analyzed, and the central tendency of the fluid kinetic energy frequency domain distribution, i.e., the fluid energy centroid coefficient, is further calculated. Assuming the frequency domain centroid frequency calculated by weighted averaging is 1.8 Hz, the normalized fluid energy centroid coefficient is 0.45. At this point, the read logic reads the preset optimal energy efficiency target range. The setting process for this range is as follows: The historical operation log of the gas-solid conveying system is called, and 1000 synchronous records containing the centroid coefficient, inlet flow rate (e.g., 10 cubic meters per minute), and outlet flow rate (e.g., 500 kilograms per minute) are extracted. For each record, the outlet flow rate of 500 is divided by the inlet flow rate of 10, resulting in a single-point gas-solid ratio energy efficiency value of 50. A mapping set between the centroid coefficient and energy efficiency value is constructed, and a cubic polynomial curve is fitted to it to generate a trend curve. Searching the curve, we found that the highest energy efficiency value peaked at a centroid coefficient of 0.55, corresponding to a global optimal energy efficiency value of 60. Selecting 0.9 as the efficiency coefficient, we calculated 60 multiplied by 0.9 to obtain 54, which was used as the energy efficiency boundary threshold. We found a continuous segment in the trend curve where the energy efficiency value was greater than 54, corresponding to a centroid coefficient range of 0.48 to 0.62. Setting 0.48 as the lower limit and 0.62 as the upper limit, we generated the optimal energy efficiency target range [0.48, 0.62]. Returning to real-time control, the comparison logic performed a difference comparison between the current centroid coefficient of 0.45 and the target range [0.48, 0.62]. Since 0.45 is less than the lower limit of 0.48, the calculated deviation difference was -0.03, the quantized deviation degree was 0.03, and the deviation direction was "too low". Based on this, we generated an energy efficiency target deviation vector, indicating that the current flow state is too biased towards low frequencies, potentially indicating a deposition tendency, requiring enhanced perturbation.

[0089] Table 2 Example of Calculating the Optimal Energy Efficiency Target Range

[0090]

[0091] As shown in Table 2, by fitting and optimizing historical data, the center of gravity coefficient window for high-efficiency operation was established. The current value of 0.45 falls into the inefficient zone on the left, triggering a clear adjustment requirement.

[0092] S402: Based on the energy efficiency target deviation vector, determine the adjustment polarity of the blowing air path according to the deviation direction, match the valve control parameters based on the degree of deviation, determine the valve opening step value and adjustment frequency, construct a compensation relationship for the gas-solid ratio imbalance state, establish a control configuration including valve action sequence and flow regulation parameters, and generate the blowing valve step adjustment configuration.

[0093] Based on the energy efficiency target deviation vector (deviation difference -0.03, direction is low), the adjustment polarity of the blowing air path is first determined. Since a low center of gravity coefficient indicates insufficient fluidization, it is determined that the blowing flow rate needs to be "increased" to improve fluid activity, i.e., the adjustment polarity is positive. Next, based on the deviation of 0.03, the valve control parameters are matched. The control unit has a preset proportional gain coefficient, assumed to be 100. A multiplication operation is performed: 0.03 multiplied by 100 equals 3, and the valve opening step value is determined to be 3%. At the same time, the adjustment frequency is determined based on the deviation. The deviation is small, and the adjustment frequency is set to once per second. A compensation relationship for the gas-solid ratio imbalance is constructed, i.e., the opening is increased by 3% based on the current opening. By establishing a control configuration that includes valve action timing (immediate execution) and flow regulation parameters (target increment + 3%), the final step adjustment configuration of the blowing valve is generated. Through this configuration instruction, it is clear that the blowing valve actuator needs to increase the valve core rotation angle by the corresponding step size to introduce more secondary air volume and break the current under-fluidization state.

[0094] S403: Calls the step adjustment configuration of the blowing valve, converts it into an electrical drive signal and applies it to the blowing valve actuator to dynamically adjust the on / off state and flow path of the blowing air path, collects the status feedback data during the execution process and encapsulates it with the control command to generate blowing flow state control data;

[0095] The step adjustment configuration of the booster valve is invoked and converted into a standard 4-20mA electrical drive signal or PWM pulse width modulation signal. Assuming the current signal is 12mA, corresponding to a 50% opening, the calculation logic increases the signal strength according to the step command, calculating the target signal value as 12 plus (16mA range multiplied by 3%), equaling 12.48mA. This current signal is applied to the electromagnetic coil or positioner of the booster valve actuator, driving the valve stem to produce physical displacement, changing the on / off state and flow path of the booster airway, thus expanding the actual ventilation cross-sectional area. During execution, the feedback acquisition unit collects real-time status feedback data (e.g., the actual opening has reached 53%) through the valve position feedback sensor, encapsulates the feedback data with the issued control command, confirms the action is executed correctly, and generates booster flow state control data. By generating booster flow state control data, the closed-loop system confirms that the physical actuator has responded to the algorithm-level optimization requirements, and the fluidized bed begins to receive stronger airflow disturbances, which is expected to push the center of gravity coefficient back to the target range [0.48, 0.62].

[0096] Please see Figure 6 The specific steps for obtaining intake limit search data are as follows:

[0097] S501: Call the assisted blowing flow state control data to verify the execution status of the control command, generate a periodic opening disturbance signal and apply it to the main intake valve, combine the visualized fluid feature dataset to extract the gas-solid two-phase flow grayscale image sequence within the disturbance time window, segment and calculate the area ratio of the material pixel region frame by frame, construct the time-series correlation pair between valve opening and material ratio, and generate a dynamic disturbance response sequence.

[0098] After confirming the execution of the control command by calling the assisted flow regime control data, a periodic opening disturbance signal is first generated. For example, a sine wave signal with an amplitude of 5% and a period of 10 seconds is set and superimposed on the current opening command of the main intake valve. When the main intake valve produces a slight fluctuation under the drive of this disturbance signal, the gas-solid two-phase flow grayscale image sequence within the disturbance time window (i.e., these 10 seconds) is extracted by combining the visualized fluid feature dataset. For each frame in the sequence, the material pixel region is identified using a threshold segmentation method, and the proportion of material pixels to the total number of pixels in the entire frame is calculated. For example, at the peak of the disturbance period (when the valve opening is at its maximum), the proportion of material pixels is 40%; at the trough (when the valve opening is at its minimum), the proportion of material pixels is 45%. The valve opening value at each moment is paired with the corresponding material proportion value to construct a time-series correlation pair and generate a dynamic disturbance response sequence. The dynamic disturbance response sequence is obtained through the sensitivity of the active detection system to input changes, i.e., the "excitation-response" test.

[0099] S502: Based on the dynamic disturbance response sequence, extract the peak-to-trough difference of the periodic opening disturbance signal as the valve opening change amplitude, simultaneously calculate the extreme value difference of the material proportion data in the corresponding time period as the material pixel proportion change amplitude, perform the ratio calculation of the material pixel proportion change amplitude and the valve opening change amplitude, analyze the flow response characteristics, and generate the conveying impedance sensitivity.

[0100] Based on the dynamic disturbance response sequence, the peak-to-trough difference of the periodic opening disturbance signal is extracted, i.e., 5% (positive amplitude) minus -5% (negative amplitude) equals 10%, which is taken as the valve opening change amplitude. Simultaneously, the extreme value difference of the material proportion data within the corresponding time period is calculated, i.e., 45% minus 40% equals 5%, which is taken as the material pixel proportion change amplitude. A ratio calculation is performed: the material proportion change amplitude of 5% is divided by the valve opening change amplitude of 10%, resulting in a ratio of 0.5. This value of 0.5 represents the conveying impedance sensitivity, quantifying the impact of flow rate changes on material concentration. If this value is too large, it indicates that the gas-solid flow is extremely unstable, and even small airflow changes can cause drastic fluctuations in material concentration; if the value is too small (close to 0), it indicates that airflow changes cannot effectively drive the material, possibly indicating blockage or an extremely dilute phase state. Through active disturbance testing, the conveying impedance sensitivity was obtained, thus revealing the current dynamic gain characteristics.

[0101] S503: Utilizes the transmission impedance sensitivity to detect the gradient polarity of the sensitivity value, locates the relative position of the optimal operating point, analyzes the response intensity, adapts the corresponding gradient descent step size, generates an intake valve optimization control configuration containing dynamic adjustment logic, drives the intake valve to perform the physical limit operating point approximation search and steady-state maintenance, and generates intake limit search data.

[0102] Using a transmission impedance sensitivity value of 0.5, the gradient polarity is first detected. By comparing the sensitivity changes over two consecutive cycles (assuming the previous cycle's sensitivity was 0.4 and the current is 0.5, showing an upward trend), and combining this with the direction of energy efficiency change, the relative position of the optimal operating point is determined. If increasing the air volume causes the sensitivity to tend towards the optimal response range (e.g., the preset optimal sensitivity is 0.6), it is determined that the current state is still in the efficiency ramp-up zone. The response intensity is analyzed, and the corresponding gradient descent step size is adapted. Since the current sensitivity of 0.5 is close to the target of 0.6, a small step size is calculated, such as increasing the basic intake opening by 1%. An intake valve optimization control configuration containing dynamic adjustment logic is generated, driving the intake valve to move the equilibrium position towards a 1% increase while maintaining a sinusoidal disturbance, performing a search to approximate the physical limit operating point. Simultaneously, sensitivity changes are continuously monitored. Once the sensitivity begins to decrease or exceeds the threshold, the search is immediately stopped and a steady state is maintained, thus outputting the intake limit search data. By automatically finding and locking the physical energy efficiency limit point of the gas-solid conveying system, it is ensured that the unloading process is always in a critically efficient state of "highest gas-solid ratio and no blockage".

[0103] Please see Figure 7 A semi-trailer unloading control system for transporting powdery materials, comprising:

[0104] The image preprocessing module acquires a sequence of grayscale images of gas-solid two-phase flow, crops the central region of the sequence to obtain pixel columns, splices the pixel columns to form a two-dimensional pixel matrix, and constructs a visualized fluid feature dataset containing spatiotemporal slice image data.

[0105] The safety analysis module calls the visualized fluid feature dataset, performs projection transformation on the spatiotemporal slice image data, calculates the extreme point angle differential result to obtain the peak angle drift rate, compares it with the safety inertial benchmark, and generates a transport safety judgment record.

[0106] The frequency domain analysis module calls the transport safety judgment record to determine continuity, combines the visualized fluid feature dataset, calculates the gray mean of the gas-solid two-phase flow grayscale image sequence to construct a time series signal, obtains the frequency domain energy distribution spectrum through frequency domain transformation, calculates the full-band energy ratio of low-frequency components, and obtains the flow energy distribution data.

[0107] The control and configuration module calls the flow energy distribution data to extract the flow energy centroid coefficient, compares it with the preset optimal energy efficiency target range, obtains the blowing valve adjustment configuration according to the degree of deviation, and outputs the blowing flow control data.

[0108] The optimization search module calls the blowing flow state control data to determine the execution status, combines the visualized fluid feature dataset, and calculates the material ratio and opening change ratio of the gas-solid two-phase flow grayscale image sequence with the inlet valve disturbance signal to obtain the conveying impedance sensitivity, generates the optimization configuration and searches for the efficiency limit point, and outputs the inlet limit search data.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for controlling the unloading of a semi-trailer transporting powdery or granular materials, characterized in that, Includes the following steps: S1: Collect a sequence of grayscale images of gas-solid two-phase flow, crop the central region of the sequence to obtain pixel columns, and stitch the pixel columns together to form a two-dimensional pixel matrix to construct a visualized fluid feature dataset containing spatiotemporal slice image data. S2: Call the visualized fluid feature dataset, perform projection transformation on the spatiotemporal slice image data, calculate the extreme point angle differential result to obtain the peak angle drift rate, and compare it with the safety inertial benchmark to generate a transport safety judgment record; S3: Call the transport safety determination record to determine continuity, combine the visualized fluid feature dataset, calculate the gray mean of the gas-solid two-phase flow grayscale image sequence to construct a time series signal, obtain the frequency domain energy distribution spectrum through frequency domain conversion, calculate the full-band energy ratio of low-frequency components, and obtain flow energy distribution data; S4: Extract the flow energy centroid coefficient from the flow energy distribution data, compare it with the preset optimal energy efficiency target range, obtain the blowing valve adjustment configuration according to the degree of deviation, and output the blowing flow regulation data. S5: Call the blowing flow state control data to determine the execution status, combine the visualized fluid feature dataset, and calculate the material ratio and opening change ratio of the gas-solid two-phase flow grayscale image sequence with the inlet valve disturbance signal to obtain the conveying impedance sensitivity, generate the optimal configuration and search for the efficiency limit point, and output the inlet limit search data.

2. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 1, characterized in that, The visualized fluid feature dataset specifically comprises the original grayscale image sequence, the central vertically cropped pixel column, and the two-dimensional spatiotemporal slice matrix. The conveying safety judgment record includes the parameter space projection matrix, the peak angle drift rate value, the rigid blockage cutoff mark, and the flow continuity confirmation mark. The flow energy distribution data specifically comprises the regional grayscale mean time series signal, the frequency domain energy distribution spectrum, the total energy integral in the low-frequency range, and the weight of the energy integral ratio across the entire frequency band. The blowing flow regulation data includes the deviation value of the optimal energy efficiency target range, the step adjustment configuration parameters of the blowing valve, the gas-solid ratio compensation command, and the fluidization state stability control signal. The inlet limit search data specifically refers to the periodic opening disturbance signal, the material pixel ratio change rate, the conveying impedance sensitivity value, and the inlet valve optimization control configuration parameters.

3. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 1, characterized in that, The specific steps for obtaining the visualized fluid feature dataset are as follows: S101: Monitor the real-time flow status of gas-solid two-phase flow in the semi-trailer discharge sight glass area, collect a series of grayscale images of gas-solid two-phase flow in a continuous time period, traverse each frame of the gas-solid two-phase flow grayscale image sequence, calculate the geometric center coordinates of the image plane, and use them as a reference to delineate a central vertical region with a width of one pixel, perform a cropping operation and extract the grayscale values ​​of all pixels in the region to generate a central vertical cropping pixel column. S102: Call the central vertical cropping pixel column, read the acquisition time timestamp corresponding to each pixel column, and perform splicing processing on all pixel columns in the horizontal direction according to the increasing order of timestamp values. Establish a two-dimensional coordinate system with time dimension as the horizontal axis and spatial position as the vertical axis. Map and fill the spliced ​​pixel gray values ​​into the coordinate system to form a two-dimensional pixel matrix and generate a two-dimensional spatiotemporal slice matrix. S103: Call the two-dimensional spatiotemporal slice matrix, standardize and format the matrix data, construct spatiotemporal slice image data containing fluid motion texture features, establish the correlation mapping relationship between the data and the current semi-trailer discharge condition parameters, integrate the spatiotemporal slice image data and the correlation mapping relationship, and generate a visualized fluid feature dataset.

4. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 3, characterized in that, The specific steps for obtaining the transport safety determination record are as follows: S201: Based on the visualized fluid feature dataset, extract spatiotemporal slice image data, set the projection angle scanning sequence, perform linear integral projection transformation on the spatiotemporal slice image data, accumulate pixel gray values ​​along the projection direction to calculate path energy intensity, map the energy intensity to the corresponding coordinate points in the parameter transformation domain, and generate a parameter space projection matrix. S202: Extract the parameter space projection matrix, retrieve the coordinates of the extreme point with the highest energy accumulation in the matrix, extract the angle parameter components corresponding to the extreme point coordinates, combine the angle parameter components of the previous acquisition cycle, calculate the differential change rate of the current angle parameter components relative to the acquisition time, and obtain the peak angle drift rate. S203: Using the peak angle drift rate, read the preset safety inertia benchmark, perform a numerical comparison between the peak angle drift rate and the safety inertia benchmark, establish rigid blockage cutoff markers and flow continuity judgment markers based on the comparison results, encapsulate the established marker data, and generate a transport safety judgment record.

5. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 4, characterized in that, The specific process for obtaining the preset safety inertial reference is as follows: The historical conveying operation log is retrieved, and historical periods marked as stable discharge states are filtered out. The gas-solid two-phase flow grayscale image sequence within the corresponding period is extracted as a benchmark verification sample. Following the calculation path of the peak angle drift rate, the historical angle drift rate set of the benchmark verification sample is obtained. The arithmetic mean and standard deviation of the set are calculated. The sum of the arithmetic mean and three times the standard deviation is selected as the basic fluctuation limit. The bulk density data and particle repose angle data of the conveyed material are collected in real time. Normalization weighted operation is performed on the bulk density data and particle repose angle data to obtain the working condition correction factor for the current material characteristics. The basic fluctuation limit and the working condition correction factor are multiplied to obtain the threshold value and set as the preset safety inertia benchmark.

6. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 4, characterized in that, The specific steps for obtaining the fluid energy distribution data are as follows: S301: Obtain the transport safety judgment record and parse the flow continuity judgment mark, call the visualized fluid feature dataset to extract the gas-solid two-phase flow grayscale image sequence, calculate the pixel grayscale arithmetic mean of each frame image in the sequence, arrange all arithmetic means according to the acquisition timestamp order to construct a one-dimensional numerical sequence, and generate a grayscale time series signal. S302: Call the grayscale time-series signal, perform a discrete orthogonal transformation operation from the time domain to the frequency domain, analyze each frequency basis function component contained in the signal, calculate the square value of the amplitude modulus of each component as the energy intensity of the corresponding frequency, establish a mapping relationship table between frequency values ​​and energy intensity values, and generate a frequency domain energy distribution spectrum. S303: Analyze the frequency domain energy distribution spectrum, extract the frequency components within the preset low frequency range, perform integral summation on the energy intensity of the extracted components to obtain the total low frequency energy value, calculate the integral sum of the energy intensity of the frequency components across the entire frequency band, calculate the weight of the total low frequency energy value in the integral sum, integrate the weight with the energy statistics results, and generate flow energy distribution data.

7. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 6, characterized in that, The process of obtaining the preset low-frequency range is as follows: During the fluidization calibration stage, a sequence of grayscale images of the gas-solid two-phase flow under stable delivery conditions is acquired. A discrete orthogonal transform is performed to generate a standard frequency domain energy distribution spectrum. Gaussian smoothing filtering is applied to the standard frequency domain energy distribution spectrum. The frequency coordinates corresponding to the maximum energy intensity are retrieved as the center frequency of the main peak. The energy distribution curve to the right of the center frequency of the main peak is scanned along the frequency axis in the positive direction. The frequency position where the energy intensity decays to 50% of the main peak energy intensity is identified and marked as the half-power frequency point. The difference between the half-power frequency point and the center frequency of the main peak is calculated to obtain the single-side bandwidth of the main peak. The sum of the center frequency of the main peak and 4 times the single-side bandwidth of the main peak is calculated as the cutoff frequency. A closed interval is constructed with zero Hz as the starting point and the cutoff frequency as the ending point. The target closed interval is set as the preset low-frequency range.

8. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 6, characterized in that, The specific steps for obtaining the blowing flow regime control data are as follows: S401: Using the fluid energy distribution data, analyze the frequency domain energy integral ratio weight, calculate the concentration trend of fluid kinetic energy frequency domain distribution, obtain the fluid energy centroid coefficient, read the preset optimal energy efficiency target interval, perform the difference comparison operation between the coefficient value and the interval boundary, calculate the deviation difference, quantify the degree and direction of deviation, and generate an energy efficiency target deviation vector. S402: Based on the energy efficiency target deviation vector, determine the adjustment polarity of the blowing air path according to the deviation direction, match the valve control parameters based on the degree of deviation, determine the valve opening step value and adjustment frequency, construct a compensation relationship for the gas-solid ratio imbalance state, establish a control configuration including valve action sequence and flow regulation parameters, and generate the blowing valve step adjustment configuration. S403: Invoke the step adjustment configuration of the blowing valve, convert it into an electrical drive signal and apply it to the blowing valve actuator to dynamically adjust the on / off state and flow path of the blowing air path, collect the status feedback data during the execution process and encapsulate it with the control command to generate blowing flow state control data.

9. The method for controlling the unloading of a semi-trailer transporting powdery materials according to claim 8, characterized in that, The specific steps for obtaining the intake limit search data are as follows: S501: Call the blowing flow state control data to verify the control command execution status, generate a periodic opening disturbance signal and apply it to the main intake valve, combine the visualized fluid feature dataset to extract the gas-solid two-phase flow grayscale image sequence within the disturbance time window, segment and calculate the area ratio of the material pixel region frame by frame, construct the time-series correlation pair between valve opening and material ratio, and generate a dynamic disturbance response sequence. S502: Based on the dynamic disturbance response sequence, extract the peak-to-trough difference of the periodic opening disturbance signal as the valve opening change amplitude, simultaneously calculate the extreme value difference of the material proportion data in the corresponding time period as the material pixel proportion change amplitude, perform the ratio calculation of the material pixel proportion change amplitude and the valve opening change amplitude, analyze the flow response characteristics, and generate the conveying impedance sensitivity. S503: Utilizing the aforementioned transmission impedance sensitivity, detect the gradient polarity of the sensitivity value, locate the relative position of the optimal operating point, analyze the response intensity, adapt the corresponding gradient descent step size, generate an intake valve optimization control configuration containing dynamic adjustment logic, drive the intake valve to perform approximation search and steady-state maintenance of the physical limit operating point, and generate intake limit search data.

10. A control system for unloading powdery material transport semi-trailers, characterized in that, The system is used to implement the unloading control method for a semi-trailer transporting powdery materials as described in any one of claims 1-9, and the system includes: The image preprocessing module acquires a sequence of grayscale images of gas-solid two-phase flow, crops the central region of the sequence to obtain pixel columns, splices the pixel columns to form a two-dimensional pixel matrix, and constructs a visualized fluid feature dataset containing spatiotemporal slice image data. The safety analysis module calls the visualized fluid feature dataset, performs projection transformation on the spatiotemporal slice image data, calculates the extreme point angle differential result to obtain the peak angle drift rate, compares it with the safety inertial benchmark, and generates a transport safety judgment record. The frequency domain analysis module calls the transport safety judgment record to determine continuity, combines the visualized fluid feature dataset, calculates the gray mean of the gas-solid two-phase flow grayscale image sequence to construct a time series signal, obtains the frequency domain energy distribution spectrum through frequency domain transformation, calculates the full-band energy ratio of low-frequency components, and obtains flow energy distribution data. The control and configuration module calls the flow energy distribution data to extract the flow energy centroid coefficient, compares it with the preset optimal energy efficiency target range, obtains the blowing valve adjustment configuration according to the degree of deviation, and outputs the blowing flow control data. The optimization search module calls the blowing flow state control data to determine the execution status, combines the visualized fluid feature dataset, and calculates the material ratio and opening change ratio of the gas-solid two-phase flow grayscale image sequence with the inlet valve disturbance signal to obtain the conveying impedance sensitivity, generates the optimization configuration and searches for the efficiency limit point, and outputs the inlet limit search data.

Images