Hollow fiber membrane universal cleaning head track intelligent planning method and system

Abstract

The present application relates to the technical field of water treatment equipment cleaning, and discloses a hollow fiber membrane universal cleaning head track intelligent planning method and system. The method comprises the following steps: obtaining membrane cluster installation parameter information to generate three-dimensional to-be-cleaned region position data; calculating the effective cleaning range of a cleaning nozzle to determine the lap degree and track spacing; dividing the surface of the membrane cluster into multiple sub-regions and sorting them according to the pollutant distribution density; calculating the cleaning path coordinate points based on the effective cleaning range, the lap degree and the track spacing parameters to determine the cleaning track; and converting the cleaning track into motion control instructions. The present application realizes accurate and uniform cleaning of the membrane surface, and improves the cleaning efficiency and quality.

Description

Technical Field

[0001] This invention relates to the field of water treatment equipment cleaning technology, and in particular to a method and system for intelligent trajectory planning of a hollow fiber membrane universal cleaning head. Background Technology

[0002] Hollow fiber membranes are a commonly used filtration material in water treatment projects. Due to their advantages such as large specific surface area and high flux, they are widely used in drinking water purification, wastewater treatment, and seawater desalination. During long-term operation, contaminants inevitably accumulate on the surface of hollow fiber membranes, leading to a decrease in membrane flux and filtration efficiency. Therefore, regular cleaning and maintenance are necessary.

[0003] Currently, there are two main methods for cleaning hollow fiber membranes: chemical cleaning and physical cleaning. While chemical cleaning is effective, it can shorten the membrane's lifespan and cause secondary pollution. Physical cleaning, on the other hand, is more environmentally friendly, with the use of a high-pressure water jet omnidirectional cleaning head being a particularly effective method.

[0004] However, existing hollow fiber membrane universal cleaning head technology has the following defects and shortcomings: Existing cleaning technologies lack the ability to accurately model the three-dimensional structure of hollow fiber membrane bundles, and cannot adaptively adjust the cleaning strategy according to the installation parameters of different membrane components, leading to cleaning dead zones and uneven cleaning, thus affecting the cleaning effect. Existing cleaning technologies typically use a fixed trajectory pattern for cleaning, failing to consider the effective cleaning range of the cleaning nozzles and the reasonable overlap between trajectories, resulting in insufficient cleaning or repeated cleaning in some areas, causing low cleaning efficiency and energy waste. Existing cleaning solutions lack the ability to perceive and respond to the uneven distribution of contaminants on the membrane surface, and cannot prioritize the cleaning sequence according to the degree of contamination in different areas, resulting in unreasonable allocation of cleaning time, difficulty in achieving efficient and precise directional cleaning, and reduced overall cleaning effect and equipment lifespan. Summary of the Invention

[0005] The present invention provides a method and system for intelligent trajectory planning of a hollow fiber membrane universal cleaning head, which can at least solve some of the problems existing in the prior art.

[0006] A first aspect of the present invention provides a method for intelligent trajectory planning of a hollow fiber membrane universal cleaning head, comprising:

[0007] Obtain the installation parameter information of the hollow fiber membrane bundle, and generate three-dimensional cleaning area location data of the hollow fiber membrane bundle based on the installation parameter information;

[0008] Real-time operating parameters of the hollow fiber membrane universal cleaning head are collected, the effective cleaning range of the cleaning nozzle is calculated based on the real-time operating parameters, the overlap degree of adjacent cleaning trajectories is determined based on the effective cleaning range, and the trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface are generated based on the overlap degree.

[0009] Based on the three-dimensional location data of the area to be cleaned, the surface of the hollow fiber membrane bundle is divided into multiple cleaning sub-regions. The cleaning sub-regions are prioritized according to the pollutant distribution density of each cleaning sub-region to determine the cleaning order.

[0010] Based on the effective cleaning range, the overlap degree, and the trajectory spacing parameters, the cleaning path coordinate points of the multiple cleaning sub-regions are calculated according to the cleaning sequence, and the cleaning trajectory curve is determined based on the cleaning path coordinate points.

[0011] The cleaning trajectory curve is converted into motion control commands for the hollow fiber membrane universal cleaning head, and the motion control commands are sent to the cleaning execution device.

[0012] Obtain the installation parameter information of the hollow fiber membrane bundle, and generate three-dimensional cleaning area location data of the hollow fiber membrane bundle based on the installation parameter information, including:

[0013] A strain sensing array is constructed on the surface of a hollow fiber membrane bundle support, and the strain sensing array collects support deformation data of the hollow fiber membrane bundle during the installation process based on the installation parameter information. The installation stress distribution of the hollow fiber membrane bundle is calculated based on the support deformation data.

[0014] Based on the stress distribution of the hollow fiber membrane bundle installation, a capacitive sensing array is constructed on the surface of the hollow fiber membrane bundle. The dielectric constant change data between the hollow fiber membrane filaments and the capacitive sensing array is collected through the capacitive sensing array, and the three-dimensional spatial coordinates of the hollow fiber membrane filaments are determined based on the dielectric constant change data.

[0015] Data on the surface charge distribution of hollow fiber membrane filaments are collected. Based on the charge distribution data and the stress distribution of the hollow fiber membrane bundle installation, the stress deformation area of ​​the hollow fiber membrane filament installation is determined. The stress magnitude and stress direction of the stress deformation area are determined. Based on the stress magnitude and stress direction, three-dimensional morphology data of the hollow fiber membrane bundle surface is generated.

[0016] Spatial registration is performed between the three-dimensional spatial coordinates of the hollow fiber membrane filaments and the three-dimensional morphology data of the hollow fiber membrane bundle surface to generate three-dimensional cleaning area location data of the hollow fiber membrane bundle.

[0017] Real-time operating parameters of the hollow fiber membrane universal cleaning head are collected. The effective cleaning range of the cleaning nozzle is calculated based on these parameters. The overlap between adjacent cleaning trajectories is determined based on the effective cleaning range. The trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface are generated based on the overlap. This includes:

[0018] A jet pressure sensor array is arranged around the cleaning nozzle, and a sensitivity calibration coefficient of the jet pressure sensor array is generated based on the real-time operating parameters.

[0019] The sensitivity calibration coefficient is used to enable the jet pressure sensor array to collect pressure field distribution data of the cleaning jet, calculate the jet pressure field gradient threshold based on the pressure field distribution data, and determine the effective cleaning range of the cleaning nozzle based on the jet pressure field gradient threshold.

[0020] The cleaning nozzle is moved along a preset trajectory, and the pressure field data of the adjacent trajectory jet collected by the jet pressure sensor array is calibrated based on the sensitivity calibration coefficient. The spatial change rate of the pressure field data of the adjacent trajectory jet is calculated.

[0021] The pressure superposition intensity matrix is ​​determined based on the spatial change rate and the jet pressure field gradient threshold. The pressure superposition intensity matrix is ​​spatially mapped to the boundary of the jet pressure isosurface to identify the distribution location of the cleaning dead zone and the over-cleaning zone. The overlap of adjacent cleaning trajectories is adjusted according to the distribution location.

[0022] The overlap degree of the adjacent cleaning trajectories and the effective cleaning range are correlated and calculated to generate trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface.

[0023] The pressure superposition intensity matrix is ​​determined based on the spatial rate of change and the jet pressure field gradient threshold. The pressure superposition intensity matrix is ​​then spatially mapped to the boundary of the jet pressure isosurface to identify the distribution locations of cleaning dead zones and over-cleaning zones. The overlap between adjacent cleaning trajectories is adjusted based on these distribution locations, including:

[0024] A three-dimensional tensor matrix is ​​generated based on the spatial rate of change, and the three-dimensional tensor matrix is ​​projected onto a two-dimensional plane to form a spatial rate of change distribution matrix.

[0025] The spatial rate of change distribution matrix is ​​compared with the jet pressure field gradient threshold to determine the coordinates of the boundary points where the spatial rate of change exceeds the jet pressure field gradient threshold, and a pressure superposition intensity matrix is ​​generated based on the boundary point coordinates.

[0026] The boundary of the jet pressure isosurface is fitted to generate a continuous and smooth jet pressure distribution surface. The pressure superposition intensity matrix is ​​then registered with the jet pressure distribution surface in spatial coordinates to generate a three-dimensional distribution map of the cleaning pressure field.

[0027] Based on the three-dimensional distribution map of the cleaning pressure field, calculate the spatial coordinates of the zero-value region and the peak region of the pressure field, and map the spatial coordinates onto the surface of the hollow fiber membrane to form a distribution map of the cleaning dead zone and the over-cleaning region.

[0028] The ratio of the area of ​​the dead cleaning zone to the area of ​​the over-cleaning zone is calculated based on the distribution map of the dead cleaning zone and the over-cleaning zone, and the overlap of adjacent cleaning trajectories is adjusted based on the ratio.

[0029] Based on the three-dimensional location data of the area to be cleaned, the surface of the hollow fiber membrane bundle is divided into multiple cleaning sub-regions. The cleaning sub-regions are then prioritized according to the contaminant distribution density of each sub-region to determine the cleaning order, including:

[0030] The surface feature point coordinates of the hollow fiber membrane bundle are extracted from the three-dimensional location data of the area to be cleaned, and the surface of the hollow fiber membrane bundle is divided into multiple cleaning sub-regions based on the surface feature point coordinates.

[0031] Calculate the boundary point set for each of the cleaning sub-regions, and determine the area and perimeter of each of the cleaning sub-regions based on the boundary point set;

[0032] Collect images of pollutant distribution on the surface of hollow fiber membrane bundles, extract gray values ​​from the pollutant distribution images, map the gray values ​​to the boundary point set corresponding to the cleaning sub-region, calculate the pollutant distribution density of each cleaning sub-region based on the boundary point set, and generate pollution load indexes for each cleaning sub-region based on the pollutant distribution density and the area.

[0033] Based on the boundary point set, the common boundary coordinates of adjacent cleaning sub-regions are identified, the connectivity length of the common boundary coordinates is calculated, and the ratio of the connectivity length to the perimeter is determined as the location association strength.

[0034] The pollutant distribution density gradient between adjacent cleaning sub-regions is calculated based on the location correlation strength. The cleaning sub-regions are then prioritized based on the pollutant distribution density gradient and the pollution load index to determine the cleaning order.

[0035] Based on the effective cleaning range, the overlap degree, and the trajectory spacing parameters, the cleaning path coordinates of the multiple cleaning sub-regions are calculated according to the cleaning sequence, and the cleaning trajectory curve is determined based on the cleaning path coordinates, including:

[0036] Based on the effective cleaning range data, the spatial curvature distribution of the hollow fiber membrane bundle surface is extracted, the effective cleaning range data is projected onto the spatial curvature distribution, the projection deformation coefficient is calculated, the effective cleaning range data is corrected based on the projection deformation coefficient, and the actual coverage area parameter is generated.

[0037] The local trajectory offset is calculated based on the actual coverage area parameter, the overlap degree and the trajectory spacing parameter, and the cleaning path coordinate points are generated based on the local trajectory offset and the projection deformation coefficient.

[0038] The angle between the cleaning nozzle jet direction and the normal vector of the membrane fiber surface is calculated based on the coordinate points of the cleaning path. The effective action distance of the local jet is calculated based on the angle. The area to be optimized is determined based on the effective action distance of the local jet and the projection deformation coefficient.

[0039] Within the defined area, coordinate point optimization criteria are established. Based on these criteria, the density and position of the cleaning path coordinate points are adjusted and optimized to generate a cleaning trajectory curve that satisfies both the requirements of local jet action and surface curvature constraints.

[0040] Within the defined area, coordinate point optimization criteria are established. Based on these criteria, the density and position of the cleaning path coordinate points are adjusted and optimized to generate a cleaning trajectory curve that satisfies both local jet action requirements and surface curvature constraints. This includes:

[0041] Calculate the topological connection features between the coordinate points of the cleaning path within the area, compare the topological connection features with the minimum turning radius of the trajectory, and determine the cleaning sub-regions that need to be optimized for coordinate point optimization.

[0042] Extract the local surface geometric features in the sub-region to be optimized for cleaning, calculate the principal curvature and Gaussian curvature of the local surface, determine the coordinate point density weight coefficient based on the numerical distribution of the principal curvature and Gaussian curvature, and combine the density weight coefficient with the coordinate points of the cleaning path to generate the coordinate point optimization criterion.

[0043] Based on the coordinate point optimization criteria, position information of newly added coordinate points and coordinate points to be deleted is generated in the cleaning sub-region to be optimized. The position information is input into the coordinate point self-organizing optimization equation based on the fluid diffusion principle to calculate the spatial position correction of each cleaning path coordinate point in the cleaning sub-region to be optimized.

[0044] The coordinate points of the cleaning path are reconstructed based on the spatial position correction amount, the directional features of the reconstructed coordinate point sequence are extracted, and a cleaning trajectory curve that satisfies the requirements of local jet action and surface curvature constraint is generated based on the directional features.

[0045] A second aspect of the present invention provides an intelligent trajectory planning system for a hollow fiber membrane universal cleaning head, comprising:

[0046] The first unit is used to acquire the installation parameter information of the hollow fiber membrane bundle and generate three-dimensional cleaning area location data of the hollow fiber membrane bundle based on the installation parameter information.

[0047] The second unit is used to collect real-time operating parameters of the hollow fiber membrane universal cleaning head, calculate the effective cleaning range of the cleaning nozzle based on the real-time operating parameters, determine the overlap degree of adjacent cleaning trajectories based on the effective cleaning range, and generate trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface based on the overlap degree.

[0048] The third unit is used to divide the hollow fiber membrane bundle surface into multiple cleaning sub-regions according to the three-dimensional location data of the area to be cleaned, and to prioritize the cleaning sub-regions according to the pollutant distribution density of each cleaning sub-region to determine the cleaning order.

[0049] The fourth unit is used to calculate the cleaning path coordinates of the multiple cleaning sub-regions according to the effective cleaning range, the overlap degree, and the trajectory spacing parameters, and to determine the cleaning trajectory curve based on the cleaning path coordinates.

[0050] The fifth unit is used to convert the cleaning trajectory curve into motion control commands for the hollow fiber membrane universal cleaning head, and to send the motion control commands to the cleaning actuator.

[0051] A third aspect of the present invention provides an electronic device, comprising:

[0052] processor;

[0053] Memory used to store processor-executable instructions;

[0054] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0055] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0056] This invention achieves precise positioning and uniform cleaning of the hollow fiber membrane bundle surface by acquiring installation parameter information of hollow fiber membrane bundles and generating three-dimensional location data of the area to be cleaned, and collecting real-time working parameters to determine the overlap and spacing parameters of the cleaning trajectory, thereby improving cleaning efficiency and cleaning quality.

[0057] This invention divides the surface of the hollow fiber membrane bundle into multiple cleaning sub-regions and prioritizes them according to the distribution density of pollutants to determine the cleaning order. This allows for differentiated cleaning strategies for areas with different levels of contamination, effectively improving the targeting of cleaning and resource utilization.

[0058] This invention determines the cleaning trajectory curve based on the coordinate points of the cleaning path and converts it into motion control commands, which are then sent to the cleaning execution device. This achieves intelligent and automated control of the cleaning process, reduces manual intervention, lowers the difficulty of operation and labor costs, and improves the safety and reliability of cleaning operations. Attached Figure Description

[0059] Figure 1 This is a flowchart illustrating the intelligent trajectory planning method for the hollow fiber membrane universal cleaning head according to an embodiment of the present invention.

[0060] Figure 2 This is a schematic diagram of the process for determining the trajectory spacing parameters according to an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0063] Figure 1 This is a flowchart illustrating the intelligent trajectory planning method for the hollow fiber membrane universal cleaning head according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0064] Obtain the installation parameter information of the hollow fiber membrane bundle, and generate three-dimensional cleaning area location data of the hollow fiber membrane bundle based on the installation parameter information;

[0065] Real-time operating parameters of the hollow fiber membrane universal cleaning head are collected, the effective cleaning range of the cleaning nozzle is calculated based on the real-time operating parameters, the overlap degree of adjacent cleaning trajectories is determined based on the effective cleaning range, and the trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface are generated based on the overlap degree.

[0066] Based on the three-dimensional location data of the area to be cleaned, the surface of the hollow fiber membrane bundle is divided into multiple cleaning sub-regions. The cleaning sub-regions are prioritized according to the pollutant distribution density of each cleaning sub-region to determine the cleaning order.

[0067] Based on the effective cleaning range, the overlap degree, and the trajectory spacing parameters, the cleaning path coordinate points of the multiple cleaning sub-regions are calculated according to the cleaning sequence, and the cleaning trajectory curve is determined based on the cleaning path coordinate points.

[0068] The cleaning trajectory curve is converted into motion control commands for the hollow fiber membrane universal cleaning head, and the motion control commands are sent to the cleaning execution device.

[0069] In one optional implementation, installation parameter information of the hollow fiber membrane bundle is obtained, and three-dimensional cleaning area location data of the hollow fiber membrane bundle is generated based on the installation parameter information, including:

[0070] A strain sensing array is constructed on the surface of a hollow fiber membrane bundle support, and the strain sensing array collects support deformation data of the hollow fiber membrane bundle during the installation process based on the installation parameter information. The installation stress distribution of the hollow fiber membrane bundle is calculated based on the support deformation data.

[0071] Based on the stress distribution of the hollow fiber membrane bundle installation, a capacitive sensing array is constructed on the surface of the hollow fiber membrane bundle. The dielectric constant change data between the hollow fiber membrane filaments and the capacitive sensing array is collected through the capacitive sensing array, and the three-dimensional spatial coordinates of the hollow fiber membrane filaments are determined based on the dielectric constant change data.

[0072] Data on the surface charge distribution of hollow fiber membrane filaments are collected. Based on the charge distribution data and the stress distribution of the hollow fiber membrane bundle installation, the stress deformation area of ​​the hollow fiber membrane filament installation is determined. The stress magnitude and stress direction of the stress deformation area are determined. Based on the stress magnitude and stress direction, three-dimensional morphology data of the hollow fiber membrane bundle surface is generated.

[0073] Spatial registration is performed between the three-dimensional spatial coordinates of the hollow fiber membrane filaments and the three-dimensional morphology data of the hollow fiber membrane bundle surface to generate three-dimensional cleaning area location data of the hollow fiber membrane bundle.

[0074] In a specific embodiment of the method of the present invention, installation parameter information of the hollow fiber membrane bundle is obtained, including relevant data such as membrane bundle model, installation position, installation angle, and fixing point position. This parameter information provides basic data support for subsequent processing.

[0075] When constructing a strain sensing array on the surface of the hollow fiber membrane bundled support, flexible strain sensors, such as graphene-based flexible strain sensors, are selected and attached to key stress points on the support surface in a 10×10 matrix arrangement. Each sensor measures 5mm×5mm, has a thickness of 0.2mm, and a sensitivity of 0.02% / N. The strain sensing array is connected to a data acquisition module via wires, with a sampling frequency set to 100Hz to monitor the deformation of the support during installation in real time. During the installation of the hollow fiber membrane bundles, the strain sensing array collects deformation data at various points on the support. For example, during installation, a maximum strain value of 2.5% is measured at a certain fixed point of the support, indicating that this point is subjected to significant stress.

[0076] When calculating the stress distribution of hollow fiber membrane bundle installation based on the collected support deformation data, the deformation data was processed using the principles of elasticity. A virtual model of the support structure was established, and material property parameters, such as those for a stainless steel support with an elastic modulus of 210 GPa, were incorporated to convert the collected deformation data into stress distribution data. The calculation results show that the stress concentration at the upper connection of the support reaches 75 MPa, while the stress distribution in the bottom fixed area is relatively uniform, approximately 30 MPa.

[0077] When constructing a capacitive sensing array based on the stress distribution of hollow fiber membrane bundles, the sensor density in stress concentration areas was increased according to the aforementioned stress distribution data. Using flexible printed circuit technology, an 8×12 capacitive sensing array was constructed on the surface of the hollow fiber membrane bundles, with each capacitor unit having an area of ​​10mm×10mm and a dielectric layer thickness of 0.5mm. The capacitive sensing array operated under a 1MHz AC signal, achieving a sensitivity of 0.01pF change.

[0078] When acquiring data on the dielectric constant variation between the hollow fiber membrane filaments and the capacitive sensing array, the capacitance value of each sensing unit is continuously monitored. When the distance or dielectric properties between the hollow fiber membrane filaments and the capacitive sensors change, the corresponding capacitance value changes accordingly. In actual measurements, the baseline capacitance value in the normal region is 5.2 pF, while when the membrane filaments shift or deform, the capacitance value can vary to the range of 4.8-5.5 pF. The capacitance distribution map of the entire array is recorded, generating a spatial capacitance distribution matrix.

[0079] When determining the three-dimensional spatial coordinates of hollow fiber membrane filaments based on dielectric constant variation data, the spatial position of the filaments is calculated using triangulation based on the relationship between capacitance and distance. For an 8×12 sensor array, a total of 96 spatial sampling points are acquired. After data interpolation processing, a high-precision spatial position mapping is formed. Through an inversion algorithm, the capacitance value variation is converted into a distance parameter, thereby determining the three-dimensional coordinates of each membrane filament. For example, within a certain test area, the spatial distribution range of the membrane filaments is ±50mm on the x-axis, ±75mm on the y-axis, and 0-120mm on the z-axis, with a spatial positioning accuracy of ±0.5mm.

[0080] When collecting surface charge distribution data of hollow fiber membrane filaments, a high-sensitivity surface potentiometer array was used, with a measurement range of ±500V and a resolution of 0.1V. Twenty-five measurement points were set on the membrane filament surface to form a surface charge distribution map. The measurement results showed that the surface potential in the normal area ranged from -5V to +5V, while abnormal potential values ​​of -15V to +20V appeared in the stress-deformed area.

[0081] When determining the installation stress deformation region based on charge distribution data and membrane bundle installation stress distribution, the two sets of data are spatially registered and subjected to correlation analysis. A pattern recognition algorithm is applied to identify regions with abnormal charge distribution and stress concentration; these regions typically exhibit twisting, bending, or excessive stretching of the membrane fibers. Stress deformation regions are then marked by setting thresholds (stress threshold of 60 MPa and charge anomaly threshold of ±15 V).

[0082] When determining the magnitude and direction of stress in the stress deformation zone during installation, the principal stress values ​​and their directions are calculated using tensor analysis. For a typical stress deformation zone, the principal stress value reaches 65 MPa, and the stress direction forms a 35° angle with the membrane fiber axis, indicating that there is significant shear stress in this area, which can easily lead to the accumulation of contaminants on the membrane fiber surface or structural damage.

[0083] When generating three-dimensional morphological data of the hollow fiber membrane bundle surface based on stress magnitude and direction, the aforementioned stress information is integrated to construct a stress morphology map of the membrane bundle surface. Using the stress-deformation relationship, the deformation state of the membrane fibers under stress is predicted, generating surface micromorphological data. The morphological data resolution reaches 10 μm, clearly displaying features such as wrinkles and deformations on the membrane fiber surface.

[0084] When spatially registering the three-dimensional spatial coordinates of the hollow fiber membrane filaments with the three-dimensional morphological data of the hollow fiber membrane bundle surface, feature point matching and spatial transformation techniques are employed to establish the correspondence between the two sets of data. The registration accuracy is controlled within ±0.2 mm to ensure accurate correspondence between morphological information and spatial location. The resulting three-dimensional location data of the hollow fiber membrane bundle's area to be cleaned contains accurate spatial location information and surface morphological features, providing a basis for subsequent precise cleaning.

[0085] In one optional implementation, real-time operating parameters of the hollow fiber membrane universal cleaning head are collected; the effective cleaning range of the cleaning nozzle is calculated based on the real-time operating parameters; the overlap degree of adjacent cleaning trajectories is determined based on the effective cleaning range; and the trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface are generated based on the overlap degree, including:

[0086] A jet pressure sensor array is arranged around the cleaning nozzle, and a sensitivity calibration coefficient of the jet pressure sensor array is generated based on the real-time operating parameters.

[0087] The sensitivity calibration coefficient is used to enable the jet pressure sensor array to collect pressure field distribution data of the cleaning jet, calculate the jet pressure field gradient threshold based on the pressure field distribution data, and determine the effective cleaning range of the cleaning nozzle based on the jet pressure field gradient threshold.

[0088] The cleaning nozzle is moved along a preset trajectory, and the pressure field data of the adjacent trajectory jet collected by the jet pressure sensor array is calibrated based on the sensitivity calibration coefficient. The spatial change rate of the pressure field data of the adjacent trajectory jet is calculated.

[0089] The pressure superposition intensity matrix is ​​determined based on the spatial change rate and the jet pressure field gradient threshold. The pressure superposition intensity matrix is ​​spatially mapped to the boundary of the jet pressure isosurface to identify the distribution location of the cleaning dead zone and the over-cleaning zone. The overlap of adjacent cleaning trajectories is adjusted according to the distribution location.

[0090] The overlap degree of the adjacent cleaning trajectories and the effective cleaning range are correlated and calculated to generate trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface.

[0091] This specific implementation method collects real-time operating parameters and calculates the effective cleaning range through a jet pressure sensor array, thereby determining the overlap degree of adjacent cleaning trajectories and generating trajectory spacing parameters for uniform cleaning.

[0092] Figure 2 This is a schematic diagram illustrating the process of determining the trajectory spacing parameters according to an embodiment of the present invention. Figure 2As shown, in determining the trajectory spacing parameters, a jet pressure sensor array is first uniformly arranged around the circumference of the cleaning nozzle. This array consists of eight miniature pressure sensors, installed at 45-degree intervals along the outer wall of the cleaning nozzle, to collect pressure field distribution data of the cleaning jet. For the cleaning nozzle, its real-time operating parameters are collected, including water supply pressure, flow rate, nozzle rotation speed, and moving speed. The cleaning nozzle is placed on a standard test platform, and under different water supply pressures ranging from 0.5 MPa to 2.5 MPa, the output signal of the jet pressure sensor array is collected and compared with the reading of a standard pressure gauge to calculate the sensitivity calibration coefficient matrix K of the sensor array. For example, when the water supply pressure is 1.5 MPa, the obtained calibration coefficient K values ​​are [1.03, 0.98, 1.05, 0.97, 1.02, 0.99, 1.01, 0.96], corresponding to the calibration coefficients of the eight sensors.

[0093] Using a calibrated jet pressure sensor array, the jet pressure field distribution within a 360-degree range around the cleaning nozzle is measured while it is stationary. During measurement, pressure values ​​are measured every 5 mm within a radial region ranging from 10 mm to 500 mm from the nozzle, forming jet pressure field distribution data P(r,θ). The jet pressure field gradient value G(r,θ) is obtained by calculating the pressure difference between adjacent measurement points divided by the distance. A gradient threshold Gth is set to 0.05 MPa / cm. When the pressure gradient G(r,θ) is less than the threshold Gth, the jet pressure change in that area is considered gradual and falls within the effective cleaning range. Using this method, the effective cleaning range of the cleaning nozzle is determined to be an irregular region, reaching a maximum of 380 mm in the main jet direction and a maximum of 120 mm perpendicular to the main jet direction.

[0094] The cleaning nozzle is moved along a preset horizontal straight trajectory at a speed of 10 cm / s. During this movement, jet pressure field data is continuously collected using a calibrated jet pressure sensor array. When the cleaning nozzle moves along a second trajectory parallel to the first trajectory, jet pressure field data is collected again. The initial distance between the two trajectories is set to 90% of the effective cleaning range, i.e., 108 mm. For the collected pressure field data P1(x,y) and P2(x,y) from the two trajectories, the spatial rate of change of the pressure field at spatial coordinates (x,y) V(x,y) is calculated, which is expressed as the ratio of the pressure difference between adjacent sampling points to the distance.

[0095] The spatial rate of change V(x,y) is compared with the jet pressure field gradient threshold Gth to construct a pressure superposition intensity matrix S(x,y). When V(x,y) is less than 0.8Gth, it is identified as a cleaning dead zone and assigned the value S(x,y)=0; when V(x,y) is greater than 1.2Gth, it is identified as an over-cleaning region and assigned the value S(x,y)=2; the remaining regions are considered appropriate cleaning regions and assigned the value S(x,y)=1. This method identifies the distribution locations of cleaning dead zones and over-cleaning regions. The results show that the initially set trajectory spacing of 108mm results in a cleaning dead zone of approximately 15mm width in the middle region.

[0096] Based on the analysis results of the pressure superposition intensity matrix S(x,y), the overlap of adjacent cleaning trajectories was adjusted. Reducing the trajectory spacing from 108mm to 95mm and testing again, along with analysis of the pressure superposition intensity matrix, revealed a significant reduction in cleaning dead zones, but also the appearance of a small number of over-cleaning areas. Further fine-tuning of the trajectory spacing to 100mm resulted in neither obvious cleaning dead zones nor over-cleaning areas; the proportion of regions with S(x,y)=1 in the pressure superposition intensity matrix exceeded 95%, indicating optimal cleaning uniformity at this point.

[0097] The optimal overlap degree was correlated with the effective cleaning range, resulting in a trajectory spacing parameter D = 100 mm. Furthermore, considering the impact of different cleaning head movement speeds on the cleaning effect, a correlation between speed and trajectory spacing was established: when the movement speed is 5 cm / s, the optimal trajectory spacing is 90 mm; when the movement speed is 15 cm / s, the optimal trajectory spacing is 110 mm. These parameters constitute the trajectory spacing parameter set for uniform cleaning of the hollow fiber membrane bundle surface, and appropriate parameter combinations can be selected according to actual cleaning requirements.

[0098] The trajectory spacing parameters obtained above were applied to actual hollow fiber membrane bundle cleaning operations. Test results showed that the cleaning uniformity was improved by 37%, the cleaning efficiency was increased by 25%, and the risk of wear on the membrane surface was significantly reduced, thus extending the service life of the membrane module.

[0099] In one optional implementation, a pressure superposition intensity matrix is ​​determined based on the spatial rate of change and the jet pressure field gradient threshold. The pressure superposition intensity matrix is ​​then spatially mapped to the boundary of the jet pressure isosurface to identify the distribution locations of cleaning dead zones and over-cleaning zones. The overlap between adjacent cleaning trajectories is adjusted based on these distribution locations, including:

[0100] A three-dimensional tensor matrix is ​​generated based on the spatial rate of change, and the three-dimensional tensor matrix is ​​projected onto a two-dimensional plane to form a spatial rate of change distribution matrix.

[0101] The spatial rate of change distribution matrix is ​​compared with the jet pressure field gradient threshold to determine the coordinates of the boundary points where the spatial rate of change exceeds the jet pressure field gradient threshold, and a pressure superposition intensity matrix is ​​generated based on the boundary point coordinates.

[0102] The boundary of the jet pressure isosurface is fitted to generate a continuous and smooth jet pressure distribution surface. The pressure superposition intensity matrix is ​​then registered with the jet pressure distribution surface in spatial coordinates to generate a three-dimensional distribution map of the cleaning pressure field.

[0103] Based on the three-dimensional distribution map of the cleaning pressure field, calculate the spatial coordinates of the zero-value region and the peak region of the pressure field, and map the spatial coordinates onto the surface of the hollow fiber membrane to form a distribution map of the cleaning dead zone and the over-cleaning region.

[0104] The ratio of the area of ​​the dead cleaning zone to the area of ​​the over-cleaning zone is calculated based on the distribution map of the dead cleaning zone and the over-cleaning zone, and the overlap of adjacent cleaning trajectories is adjusted based on the ratio.

[0105] In this embodiment, the spatial rate of change data of the jet pressure field and a preset jet pressure field gradient threshold are first acquired. The spatial rate of change of the jet pressure field represents the degree of pressure change in three-dimensional space, and is usually characterized by the first derivative or gradient value of the pressure field. For example, under a certain test condition, a high-precision pressure sensor array can be used to collect pressure data at grid points spaced 5 mm apart on the surface of the hollow fiber membrane to form a spatial pressure distribution matrix.

[0106] Based on the collected pressure data, a three-dimensional tensor matrix is ​​generated. Each element of this three-dimensional tensor matrix represents the rate of change of the pressure field at the corresponding spatial location. For example, in the test, the generated three-dimensional tensor matrix has a size of 100×100×50, covering a spatial area of ​​500mm×500mm×250mm. This three-dimensional tensor matrix is ​​then projected onto a two-dimensional plane using the maximum value projection method to form a spatial rate of change distribution matrix with a size of 100×100.

[0107] The spatial rate of change distribution matrix is ​​compared with a preset jet pressure field gradient threshold. In practical applications, this threshold can be set to 0.5 kPa / mm, indicating that a pressure change exceeding 0.5 kPa per millimeter is considered an excessive gradient, potentially leading to uneven cleaning. The coordinates of boundary points where the spatial rate of change exceeds the threshold are determined through this comparison. For example, 285 boundary points were identified in the test data, and their coordinates were recorded as follows: Based on these boundary point coordinates, a pressure superposition intensity matrix is ​​generated, which reflects the cumulative pressure intensity experienced by each region during the cleaning process.

[0108] A surface fitting process is performed on the boundary of the jet pressure isosurface to generate a continuous and smooth jet pressure distribution surface. In this embodiment, a radial basis function is used for surface fitting, with a spline smoothing parameter of 0.85 and 500 control points, to fit a continuous pressure distribution surface. This surface accurately represents the continuous spatial distribution characteristics of the jet pressure, achieving a fitting accuracy of 96.8% of the original discrete data.

[0109] The pressure superposition intensity matrix and the jet pressure distribution surface are spatially registered. During the registration process, an iterative nearest-point algorithm is used, with a maximum number of iterations set to 200, a convergence threshold of 0.001 mm, and the registration error controlled within 0.05 mm. Through registration, the pressure superposition intensity information is correlated with the actual jet pressure distribution, generating a three-dimensional distribution map of the cleaning pressure field.

[0110] Based on the three-dimensional distribution map of the cleaning pressure field, the spatial coordinates of the zero-value region and the peak region of the pressure field were calculated. The zero-value region was defined as the area with pressure below 0.2 kPa, representing insufficient cleaning force; the peak region was defined as the area with pressure above 2.5 kPa, representing excessive cleaning force. In the test sample, 32 clusters of zero-value regions were identified, with a total area of ​​125 cm²; and 18 clusters of peak regions were identified, with a total area of ​​78 cm².

[0111] These spatial coordinates are mapped onto the surface of the hollow fiber membrane to form a distribution map of the cleaning dead zone and the over-cleaning zone. The mapping process uses the nearest distance projection method to project the pressure field feature points in three-dimensional space onto the hollow fiber membrane surface model. In this embodiment, the hollow fiber membrane surface model is represented by a mesh composed of 80,000 triangular facets, and the projection accuracy is controlled within 0.02 mm.

[0112] Based on the distribution map of cleaning dead zones and over-cleaning areas, the ratio of the area of ​​the cleaning dead zone to the area of ​​the over-cleaning area was calculated. In the test case, the area of ​​the cleaning dead zone was 125 cm², and the area of ​​the over-cleaning area was 78 cm², with a ratio of approximately 1.6. The overlap of adjacent cleaning trajectories was adjusted based on this ratio. When the ratio is greater than 1.5, it indicates that there are many cleaning dead zones, and the trajectory overlap needs to be increased; when the ratio is less than 0.5, it indicates that there are many over-cleaning areas, and the trajectory overlap needs to be decreased.

[0113] With an initial overlap of 30%, the ratio obtained through the above analysis was 1.6, so the overlap was adjusted to 42%. After adjustment, the cleaning effect was evaluated again. The cleaning dead zone area decreased to 58 cm², the over-cleaned area area increased slightly to 85 cm², and the ratio dropped to 0.68, which is within the ideal range (0.5~1.5), indicating that the cleaning uniformity has been significantly improved.

[0114] By implementing the above technical methods, dead zones and over-cleaning areas in the hollow fiber membrane cleaning process can be accurately identified, and cleaning uniformity can be improved by optimizing the overlap between adjacent cleaning trajectories. In actual application tests, using the optimized cleaning trajectory parameters for hollow fiber membrane cleaning increased membrane flux recovery by 18%, shortened cleaning time by 23%, and reduced cleaning media consumption by 15%, significantly improving the cleaning efficiency and service life of hollow fiber membranes.

[0115] In one optional implementation, the hollow fiber membrane bundle surface is divided into multiple cleaning sub-regions based on the three-dimensional location data of the area to be cleaned. The cleaning sub-regions are then prioritized according to the contaminant distribution density of each sub-region to determine the cleaning order, including:

[0116] The surface feature point coordinates of the hollow fiber membrane bundle are extracted from the three-dimensional location data of the area to be cleaned, and the surface of the hollow fiber membrane bundle is divided into multiple cleaning sub-regions based on the surface feature point coordinates.

[0117] Calculate the boundary point set for each of the cleaning sub-regions, and determine the area and perimeter of each of the cleaning sub-regions based on the boundary point set;

[0118] Collect images of pollutant distribution on the surface of hollow fiber membrane bundles, extract gray values ​​from the pollutant distribution images, map the gray values ​​to the boundary point set corresponding to the cleaning sub-region, calculate the pollutant distribution density of each cleaning sub-region based on the boundary point set, and generate pollution load indexes for each cleaning sub-region based on the pollutant distribution density and the area.

[0119] Based on the boundary point set, the common boundary coordinates of adjacent cleaning sub-regions are identified, the connectivity length of the common boundary coordinates is calculated, and the ratio of the connectivity length to the perimeter is determined as the location association strength.

[0120] The pollutant distribution density gradient between adjacent cleaning sub-regions is calculated based on the location correlation strength. The cleaning sub-regions are then prioritized based on the pollutant distribution density gradient and the pollution load index to determine the cleaning order.

[0121] A specific implementation method for dividing the surface of the hollow fiber membrane bundle into multiple cleaning sub-regions based on the three-dimensional location data of the area to be cleaned, achieves effective cleaning of contaminants on the surface of the hollow fiber membrane bundle through a series of technical steps.

[0122] First, three-dimensional location data of the area to be cleaned from the hollow fiber membrane bundle is acquired. This data can be obtained through a 3D laser scanner, depth camera, or other 3D imaging equipment. Surface feature point coordinates are extracted from this 3D data. A surface feature detection algorithm is used during the extraction process. This algorithm identifies salient features by calculating the normal vector and curvature changes of each point in the 3D point cloud. For example, in one practical application, 25,000 surface feature points were extracted from the 3D point cloud. These feature points accurately describe the surface geometry of the hollow fiber membrane bundle.

[0123] Based on the extracted surface feature point coordinates, a region growing partitioning algorithm was used to divide the surface of the hollow fiber membrane bundle into multiple cleaning sub-regions. This algorithm uses feature points as seed points and expands the region based on the similarity of the Euclidean distance and the angle between the normal vectors of the points. In practice, when the Euclidean distance between adjacent points is less than 5 mm and the difference in the angle between their normal vectors is less than 15 degrees, these points are assigned to the same sub-region. Using this method, a hollow fiber membrane bundle with a diameter of 60 cm and a height of 120 cm was divided into 48 cleaning sub-regions.

[0124] For each defined cleaning sub-region, its boundary point set is calculated. Boundary point set identification employs contour tracking technology, determining boundaries by detecting changes in point density between adjacent regions. The boundary points of each sub-region are sorted to form closed boundary curves. Taking a typical sub-region as an example, its boundary point set contains approximately 350 three-dimensional coordinate points, which precisely define the boundary contour of the sub-region.

[0125] Based on the established set of boundary points, the area and perimeter of each cleaning sub-region are calculated. The area is calculated by projecting the 3D surface onto its principal plane and then applying polygon area calculation methods. The perimeter is obtained by summing the Euclidean distances between adjacent points in the boundary point set. In the example, cleaning sub-region numbered 12 has an area of ​​245 square centimeters and a perimeter of 68 centimeters.

[0126] High-resolution industrial cameras were used to acquire images of contaminant distribution on the surface of hollow fiber membrane bundles. During acquisition, to ensure image quality, a ring-shaped LED light source was used to provide uniform illumination, and the camera resolution was set to 4096×3072 pixels. The acquired images underwent preprocessing, including noise removal and contrast enhancement, to highlight contaminant features.

[0127] For the preprocessed image of pollutant distribution, its grayscale values ​​are extracted. During extraction, an image grayscale conversion algorithm is used to convert the color image into a 256-level grayscale image. The grayscale value ranges from 0 to 255, where lower grayscale values ​​represent more heavily polluted areas, and higher grayscale values ​​represent cleaner areas.

[0128] The extracted grayscale values ​​are mapped to the corresponding set of boundary points for each cleaning sub-region. The mapping process uses a 3D-to-2D projection transformation to establish the correspondence between the boundary point set in 3D space and the 2D image plane. The mapping accuracy is ensured by calibration using a calibration plate, with the positioning error controlled within 2 mm.

[0129] Based on the mapped set of boundary points and their corresponding gray values, the contaminant distribution density of each cleaning sub-region is calculated. The calculation method involves averaging the gray values ​​of all points within the sub-region and converting them according to a pre-defined gray value-contaminant density lookup table. For example, in one test, the average gray value of cleaning sub-region numbered 8 was 78, corresponding to a contaminant distribution density of 6.5 g / m².

[0130] Pollution load indices for each cleaning sub-region are generated based on pollutant distribution density and area. These indices are calculated by multiplying the pollutant distribution density by the sub-region area, reflecting the total pollutant content of the sub-region. For example, cleaning sub-region number 15 has a pollutant distribution density of 4.2 g / m² and an area of ​​320 square centimeters, resulting in a calculated pollution load index of 13.44 g / m².

[0131] The identification process uses boundary point sets to identify the common boundary coordinates of adjacent cleaning sub-regions. By comparing the coordinates of boundary point sets in different sub-regions, points are considered to be located on a common boundary when the distance between two points is less than a preset threshold (e.g., 1 mm). For example, two adjacent sub-regions numbered 23 and 24 have a total of 42 common boundary points, forming a common boundary approximately 15 cm long.

[0132] The connectivity length of the common boundary coordinates is calculated, and the ratio of this length to the perimeter of the sub-region is determined as the location association strength. In practical applications, the perimeter of sub-region numbered 17 is 52 cm, and the length of its common boundary with adjacent sub-regions is 18 cm, therefore its location association strength is 0.346.

[0133] Based on the location association strength, the pollutant distribution density gradient between adjacent cleaning sub-regions is calculated. The calculation method is to divide the difference in pollutant distribution density between adjacent regions by the common boundary connectivity length. For example, if the pollutant distribution densities of two adjacent sub-regions numbered 31 and 32 are 7.8 g / m² and 5.3 g / m² respectively, and the common boundary connectivity length is 12 cm, the calculated pollutant distribution density gradient is 0.21 g / m² / cm².

[0134] Finally, the cleaning sub-regions were prioritized based on the pollutant distribution density gradient and pollution load index to determine the cleaning order. The prioritization strategy was to give higher priority to sub-regions with higher pollution load indices, and when pollution load indices were similar, sub-regions with a larger pollutant distribution density gradient from the highly polluted areas had higher priority. In a practical application case, a priority order from 1 to 48 was determined for 48 cleaning sub-regions, and the cleaning robot cleaned them in this order, effectively improving cleaning efficiency and quality.

[0135] In one optional implementation, based on the effective cleaning range, the overlap degree, and the trajectory spacing parameters, the cleaning path coordinates of the multiple cleaning sub-regions are calculated according to the cleaning sequence, and the cleaning trajectory curve is determined based on the cleaning path coordinates, including:

[0136] Based on the effective cleaning range data, the spatial curvature distribution of the hollow fiber membrane bundle surface is extracted, the effective cleaning range data is projected onto the spatial curvature distribution, the projection deformation coefficient is calculated, the effective cleaning range data is corrected based on the projection deformation coefficient, and the actual coverage area parameter is generated.

[0137] The local trajectory offset is calculated based on the actual coverage area parameter, the overlap degree and the trajectory spacing parameter, and the cleaning path coordinate points are generated based on the local trajectory offset and the projection deformation coefficient.

[0138] The angle between the cleaning nozzle jet direction and the normal vector of the membrane fiber surface is calculated based on the coordinate points of the cleaning path. The effective action distance of the local jet is calculated based on the angle. The area to be optimized is determined based on the effective action distance of the local jet and the projection deformation coefficient.

[0139] Within the defined area, coordinate point optimization criteria are established. Based on these criteria, the density and position of the cleaning path coordinate points are adjusted and optimized to generate a cleaning trajectory curve that satisfies both the requirements of local jet action and surface curvature constraints.

[0140] To accurately calculate the coordinates of the cleaning path and determine the cleaning trajectory curve, the spatial curvature distribution of the hollow fiber membrane bundle surface is first extracted based on the effective cleaning range data. This process involves obtaining point cloud data by performing a 3D scan of the hollow fiber membrane bundle surface, extracting a surface mesh model from the point cloud data, and then calculating the principal curvature and Gaussian curvature of each point in the mesh model. For example, for a cylindrical hollow fiber membrane bundle with a diameter of 200 mm and a height of 1000 mm, the curvature is larger in the top region, while the middle region is approximately cylindrical. A hemispherical structure with a curvature radius of approximately 100 mm at the top and a cylindrical structure in the middle region can be identified.

[0141] When projecting the effective cleaning range data onto a spatial curvature distribution, a ray tracing algorithm is used to calculate the intersection points of the cleaning jet and the membrane bundle surface at different positions of the nozzle. For a setting with an effective nozzle range of 250 mm and a jet radius of 30 mm, when the nozzle is perpendicularly aligned with the membrane surface, a circular coverage area with a diameter of approximately 60 mm is formed in the planar region. However, when projected onto a curved region, the coverage shape will be deformed, and the projection distortion coefficient is calculated by comparing the area before and after projection. For example, in a region with a curvature radius of 100 mm, under the same jet conditions, an elliptical coverage area may be formed, with a major axis of approximately 70 mm and a minor axis of approximately 50 mm, and a projection distortion coefficient of approximately 1.17.

[0142] The actual coverage area parameter is calculated by correcting the effective cleaning range data based on the projection distortion coefficient. For example, for a cleaning jet with a nominal coverage diameter of 60 mm, the actual coverage area in a surface region with a radius of curvature of 100 mm is approximately 2750 square millimeters, rather than the 2827 square millimeters of the non-planar projection. This difference is particularly noticeable in high curvature regions, such as the top edge region of the membrane bundle, where the actual coverage area may be 20% smaller than the theoretically calculated value.

[0143] Based on the actual coverage area parameters, preset overlap and trajectory spacing parameters, local trajectory offsets are calculated. Assuming an overlap of 30%, the standard trajectory spacing in a planar area is 42mm. However, in high-curvature areas, the trajectory spacing will be adjusted according to the projection distortion coefficient. For example, in an area with a curvature radius of 100mm, the trajectory spacing may need to be reduced to 36mm to ensure an effective overlap of 30%. Based on these local trajectory offsets and projection distortion coefficients, a series of cleaning path coordinate points are generated, forming a preliminary cleaning path plan.

[0144] When determining the angle between the cleaning nozzle jet direction and the normal vector of the membrane fiber surface, the angle is calculated by plotting the normal vector of the membrane bundle surface at each coordinate point and then plotting it against the nozzle jet direction vector. Ideally, the nozzle jet direction should be parallel to the surface normal vector, with an angle of 0 degrees, resulting in optimal cleaning. In practice, due to limitations in robotic arm movement, the angle often cannot be maintained at 0 degrees. Therefore, a maximum acceptable angle threshold of 30 degrees is set; areas exceeding this threshold require special handling.

[0145] When calculating the effective range of a local jet based on the included angle, a jet attenuation model is applied. When the included angle is 0 degrees, the effective range can reach 250 mm. As the included angle increases, the effective range decreases; for example, the effective range is approximately 230 mm at a 15-degree angle and approximately 200 mm at a 30-degree angle. Combining these effective range data with the projection distortion coefficient, the area requiring optimization is determined, especially those areas with included angles greater than 20 degrees or projection distortion coefficients greater than 1.5.

[0146] After determining the area requiring optimization, coordinate point optimization criteria were established. These criteria included: ensuring a minimum overlap of 25%; maintaining a distance of no more than 5 mm between adjacent trajectory points; ensuring the angle between the nozzle jet direction and the surface normal vector did not exceed 30 degrees; and achieving a cleaning coverage rate of over 95%. Taking the top area of ​​the membrane bundle as an example, the original plan might have resulted in uneven coverage. By increasing the trajectory point density (from one point every 10 mm to one point every 3 mm) and adjusting the nozzle angle (to make the jet direction closer to the surface normal vector), the optimized coverage rate increased from 85% to 98%.

[0147] Finally, the density and position of the coordinate points along the cleaning path are adjusted and optimized. The density of coordinate points is increased in high-curvature areas and appropriately decreased in low-curvature areas. For example, a larger point spacing (approximately 8 mm) can be used in the cylindrical region in the middle of the membrane bundle; while in high-curvature areas such as the top and connections, the point spacing may need to be reduced to 2 mm to ensure accurate coverage. Through this adaptive density adjustment, a cleaning trajectory curve that satisfies both the requirements of local jet action and surface curvature constraints is ultimately generated, ensuring effective cleaning of the entire hollow fiber membrane bundle surface. This improves cleaning uniformity by approximately 40% and reduces cleaning time by approximately 25%.

[0148] In one optional implementation, a coordinate point optimization criterion is established within the said region. Based on this criterion, the density and position of the cleaning path coordinate points are adjusted and optimized to generate a cleaning trajectory curve that satisfies both local jet action requirements and surface curvature constraints. This includes:

[0149] Calculate the topological connection features between the coordinate points of the cleaning path within the area, compare the topological connection features with the minimum turning radius of the trajectory, and determine the cleaning sub-regions that need to be optimized for coordinate point optimization.

[0150] Extract the local surface geometric features in the sub-region to be optimized for cleaning, calculate the principal curvature and Gaussian curvature of the local surface, determine the coordinate point density weight coefficient based on the numerical distribution of the principal curvature and Gaussian curvature, and combine the density weight coefficient with the coordinate points of the cleaning path to generate the coordinate point optimization criterion.

[0151] Based on the coordinate point optimization criteria, position information of newly added coordinate points and coordinate points to be deleted is generated in the cleaning sub-region to be optimized. The position information is input into the coordinate point self-organizing optimization equation based on the fluid diffusion principle to calculate the spatial position correction of each cleaning path coordinate point in the cleaning sub-region to be optimized.

[0152] Reconstruct the cleaning path coordinate points according to the spatial position correction amount, extract the direction features of the reconstructed coordinate point sequence, and generate a cleaning trajectory curve that meets the requirements of local jet action and surface curvature constraints.

[0153] In a specific example of implementing the present invention, for a given area range, it is necessary to first calculate the topological connection features between the cleaning path coordinate points. This step is achieved by calculating the spatial distance and angle between adjacent coordinate points. Specifically, for each coordinate point Pi in the area, calculate the angle θi formed by it and the adjacent points Pi-1 and Pi+1, as well as the connection line lengths li,i-1 and li,i+1. When the angle θi is less than a preset threshold (e.g., 120 degrees) or any connection line length exceeds a set value (e.g., 5 mm), record this point as a potential optimization point. By performing the above calculations on all coordinate points, determine the set of points whose topological connection features do not meet the requirements.

[0154] Next, compare the calculated topological connection features with the minimum turning radius of the trajectory. Assume that the minimum turning radius of the cleaning device is R = 10 mm. For each potential optimization point Pi, calculate its curvature radius ri = 1 / κi, where κi is the curvature of this point. If ri < R, then this point needs to be optimized and adjusted. By clustering analysis, these points that need to be optimized are grouped into several sub-regions to be optimized for cleaning. For example, optimize points with a spatial distance less than 15 mm into the same sub-region.

[0155] For each determined sub-region to be optimized for cleaning, the present invention extracts the local surface geometric features. The specific implementation method is: select dense sampling points within the sub-region, obtain the local surface equation through the surface fitting method, and calculate the principal curvatures κ1 and κ2 at each sampling point, as well as the Gaussian curvature K = κ1 × κ2. For example, within a certain sub-region to be optimized for cleaning, at the sampling point p1, the principal curvature κ1 = 0.05 mm^-1, κ2 = 0.02 mm^-1, then the Gaussian curvature K = 0.001 mm^-2.

[0156] According to the numerical distribution of the principal curvature and the Gaussian curvature, the present invention determines the coordinate point density weight coefficient. Specifically, design a weight function w = f(|κ1|,|κ2|,|K|) such that the weight value in the region with a larger curvature is higher. For example, the calculation method of w = 0.3×(|κ1|+|κ2|)+0.4×|K| can be used. At the above sampling point p1, the weight coefficient is w = 0.3×(0.05 + 0.02)+0.4×0.001 = 0.0214. Normalize the weight coefficients of all points within the entire sub-region to obtain the final density weight coefficient distribution map.

[0157] The coordinate point optimization criterion is generated by combining the density weighting coefficients mentioned above with the coordinate points along the cleaning path. In areas with higher weighting coefficients, the coordinate points should be denser; in areas with lower weighting coefficients, the coordinate points can be appropriately sparser. For example, if the baseline density is set to d0 = 1 point / square millimeter, then in a region with a weight of w, the local density is set to d = d0 × (1 + 5w). This way, the coordinate point density will increase significantly in high-curvature areas such as edges or corners, while maintaining a lower density in flat areas.

[0158] Based on the aforementioned coordinate point optimization criteria, this invention generates positional information for newly added and deleted coordinate points within the sub-region to be optimized and cleaned. Specifically, for regions with a density lower than the required local density, new points are inserted between two adjacent points to bring the local density to the required level; for regions with a density higher than the required level, redundant points are marked as points to be deleted. For example, in a sub-region with high curvature, if the original point density is 0.8 points / square millimeter and the required density is 1.2 points / square millimeter, then 50% more points need to be added; while in a flat region, if the original density is 1.5 points / square millimeter and the required density is 1.0 points / square millimeter, then approximately 33% of the points need to be deleted.

[0159] The aforementioned location information is input into a coordinate point self-organizing optimization equation based on the principle of fluid diffusion to calculate the spatial position correction for each cleaning path coordinate point. In this step, the coordinate points are treated as particles in the fluid, with repulsive and attractive forces existing between them. An iterative calculation is performed to reach an equilibrium state. Specifically, for each point Pi, the resultant force Fi is calculated, and the displacement vector Δxi is determined based on Fi. After multiple iterations (e.g., 50 times), when the displacement of all points is less than a threshold (e.g., 0.01 mm), equilibrium is considered achieved, and the cumulative displacement of each point at this point is recorded as the spatial position correction.

[0160] The coordinates of the cleaning path are reconstructed based on the calculated spatial position correction. For each original coordinate point Pi, its new position is Pi' = Pi + Δxi; simultaneously, new points are inserted in areas where density needs to be increased, and marked points are deleted in redundant areas. After reconstruction, the directional features of the coordinate point sequence are extracted, including the tangent direction and normal direction at each point.

[0161] Finally, based on the aforementioned directional characteristics, a cleaning trajectory curve that satisfies the requirements of local jet action and surface curvature constraints is generated. Specifically, cubic spline interpolation is used to connect the optimized sequence of coordinate points, ensuring that the tangent direction of the generated trajectory curve at each point is consistent with the calculated directional characteristics, the curvature satisfies the minimum turning radius constraint, and the interaction angle between the jet and the surface is within an effective range (e.g., 30°-90°). For example, at a certain surface corner, the jet angle before optimization might be 15°, which is adjusted to 45° after optimization, significantly improving the cleaning effect.

[0162] The cleaning trajectory curve generated by the above method can effectively adapt to the geometric features of complex surfaces, ensure the smoothness of the cleaning equipment trajectory during operation, and meet the angle requirements of local jet action, thereby improving cleaning efficiency and quality.

[0163] The hollow fiber membrane universal cleaning head trajectory intelligent planning system of this invention includes:

[0164] The first unit is used to acquire the installation parameter information of the hollow fiber membrane bundle and generate three-dimensional cleaning area location data of the hollow fiber membrane bundle based on the installation parameter information.

[0165] The second unit is used to collect real-time operating parameters of the hollow fiber membrane universal cleaning head, calculate the effective cleaning range of the cleaning nozzle based on the real-time operating parameters, determine the overlap degree of adjacent cleaning trajectories based on the effective cleaning range, and generate trajectory spacing parameters for uniform cleaning of the hollow fiber membrane bundle surface based on the overlap degree.

[0166] The third unit is used to divide the hollow fiber membrane bundle surface into multiple cleaning sub-regions according to the three-dimensional location data of the area to be cleaned, and to prioritize the cleaning sub-regions according to the pollutant distribution density of each cleaning sub-region to determine the cleaning order.

[0167] The fourth unit is used to calculate the cleaning path coordinates of the multiple cleaning sub-regions according to the effective cleaning range, the overlap degree, and the trajectory spacing parameters, and to determine the cleaning trajectory curve based on the cleaning path coordinates.

[0168] The fifth unit is used to convert the cleaning trajectory curve into motion control commands for the hollow fiber membrane universal cleaning head, and to send the motion control commands to the cleaning actuator.

[0169] A third aspect of the present invention provides an electronic device, comprising:

[0170] processor;

[0171] Memory used to store processor-executable instructions;

[0172] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0173] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0174] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0175] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent planning of a trajectory of a universal cleaning head for a hollow fiber membrane, characterized in that, The method comprises the following steps: acquiring installation parameter information of the hollow fiber membrane bundle, generating three-dimensional to-be-cleaned region position data of the hollow fiber membrane bundle based on the installation parameter information; acquiring real-time working parameters of the universal cleaning head of the hollow fiber membrane, calculating an effective cleaning range of the cleaning nozzle according to the real-time working parameters, determining an adjacent cleaning track overlap degree based on the effective cleaning range, and generating a track spacing parameter for uniform cleaning of the surface of the hollow fiber membrane bundle according to the overlap degree; dividing the surface of the hollow fiber membrane bundle into a plurality of cleaning sub-regions according to the three-dimensional to-be-cleaned region position data, prioritizing the cleaning sub-regions according to the pollution distribution density of each cleaning sub-region, and determining a cleaning sequence; calculating cleaning path coordinate points of the plurality of cleaning sub-regions according to the effective cleaning range, the overlap degree, and the track spacing parameter, and determining a cleaning track curve based on the cleaning path coordinate points according to the cleaning sequence; converting the cleaning track curve into a motion control instruction of the universal cleaning head of the hollow fiber membrane, and issuing the motion control instruction to a cleaning execution device.

2. The method of claim 1, wherein, The method comprises the following steps: constructing a strain sensing array on the surface of the hollow fiber membrane bundle support, making the strain sensing array collect support deformation data of the hollow fiber membrane bundle during installation based on the installation parameter information, and calculating the installation stress distribution of the hollow fiber membrane bundle according to the support deformation data; constructing a capacitive sensing array on the surface of the hollow fiber membrane bundle based on the installation stress distribution of the hollow fiber membrane bundle, collecting dielectric constant change data between the hollow fiber membrane wire and the capacitive sensing array through the capacitive sensing array, and determining the three-dimensional space coordinates of the hollow fiber membrane wire according to the dielectric constant change data; collecting surface charge distribution data of the hollow fiber membrane wire, determining the installation stress deformation region of the hollow fiber membrane wire based on the charge distribution data and the installation stress distribution of the hollow fiber membrane bundle, determining the stress size and stress direction of the installation stress deformation region, and generating three-dimensional topographic data of the surface of the hollow fiber membrane bundle based on the stress size and the stress direction; spatially registering the three-dimensional space coordinates of the hollow fiber membrane wire and the three-dimensional topographic data of the surface of the hollow fiber membrane bundle to generate three-dimensional to-be-cleaned region position data of the hollow fiber membrane bundle.

3. The method of claim 1, wherein, The method comprises the following steps: circumferentially arranging a jet pressure sensing array on the cleaning nozzle, and generating a sensitivity calibration coefficient of the jet pressure sensing array based on the real-time working parameters; The sensitivity calibration coefficient is used to collect pressure field distribution data of the cleaning jet flow by the jet pressure sensing array, a jet pressure field gradient threshold is calculated according to the pressure field distribution data, and the effective cleaning range of the cleaning nozzle is determined based on the jet pressure field gradient threshold; The cleaning nozzle is moved along a preset track, adjacent track jet pressure field data collected by the jet pressure sensing array is calibrated based on the sensitivity calibration coefficient, and a spatial variation rate of the adjacent track jet pressure field data is calculated; A pressure superposition intensity matrix is determined according to the spatial variation rate and the jet pressure field gradient threshold, the pressure superposition intensity matrix is spatially mapped with a jet pressure isosurface boundary, distribution positions of a cleaning dead zone and an over-cleaning area are identified, and an adjacent cleaning track overlap degree is adjusted according to the distribution positions. The adjacent cleaning track overlap degree and the effective cleaning range are associated and calculated to generate a track spacing parameter for uniform cleaning of a hollow fiber membrane bundle surface.

4. The method of claim 3, wherein, A pressure superposition intensity matrix is determined according to the spatial variation rate and the jet pressure field gradient threshold, the pressure superposition intensity matrix is spatially mapped with a jet pressure isosurface boundary, distribution positions of a cleaning dead zone and an over-cleaning area are identified, and an adjacent cleaning track overlap degree is adjusted according to the distribution positions, including: A three-dimensional tensor matrix is generated according to the spatial variation rate, and the three-dimensional tensor matrix is projected onto a two-dimensional plane to form a spatial variation rate distribution matrix; The spatial variation rate distribution matrix is compared with the jet pressure field gradient threshold to determine boundary point coordinates at which the spatial variation rate exceeds the jet pressure field gradient threshold, and a pressure superposition intensity matrix is generated based on the boundary point coordinates; The jet pressure isosurface boundary is subjected to surface fitting to generate a continuous and smooth jet pressure distribution surface, the pressure superposition intensity matrix is subjected to spatial coordinate registration with the jet pressure distribution surface to generate a cleaning pressure field three-dimensional distribution map; Based on the cleaning pressure field three-dimensional distribution map, spatial position coordinates of a pressure field zero-value area and a pressure field peak-value area are calculated, and the spatial position coordinates are mapped to a hollow fiber membrane surface to form a distribution map of a cleaning dead zone and an over-cleaning area; A ratio of a cleaning dead zone area to an over-cleaning area is calculated according to the distribution map of the cleaning dead zone and the over-cleaning area, and an adjacent cleaning track overlap degree is adjusted based on the ratio.

5. The method of claim 1, wherein, According to the three-dimensional to-be-cleaned region position data, the surface of the hollow fiber membrane bundle is divided into a plurality of cleaning sub-regions, the cleaning sub-regions are prioritized according to the pollutant distribution density of each cleaning sub-region, and a cleaning sequence is determined, including: Surface feature point coordinates of the hollow fiber membrane bundle are extracted from the three-dimensional to-be-cleaned region position data, and the surface of the hollow fiber membrane bundle is divided into a plurality of cleaning sub-regions based on the surface feature point coordinates; A boundary point set of each cleaning sub-region is calculated, and the area and the perimeter of each cleaning sub-region are determined according to the boundary point set; The pollution distribution image of the hollow fiber membrane bundle surface is collected, the gray value of the pollution distribution image is extracted, the gray value is mapped to the boundary point set corresponding to the cleaning sub-region, the pollution distribution density of each cleaning sub-region is calculated based on the boundary point set, and the pollution load index of each cleaning sub-region is generated according to the pollution distribution density and the area. The common boundary coordinates of adjacent cleaning sub-regions are identified based on the boundary point set, the connected length of the common boundary coordinates is calculated, and the position correlation strength is determined as the ratio of the connected length to the perimeter. The pollution distribution density gradient between adjacent cleaning sub-regions is calculated according to the position correlation strength, the cleaning sub-regions are prioritized based on the pollution distribution density gradient and the pollution load index, and the cleaning sequence is determined.

6. The method of claim 1, wherein, According to the effective cleaning range, the overlap degree and the trajectory spacing parameter, the cleaning path coordinate points of the plurality of cleaning sub-regions are calculated in the cleaning sequence, and the cleaning trajectory curve is determined based on the cleaning path coordinate points, including: According to the effective cleaning range data, the spatial curvature distribution of the hollow fiber membrane bundle surface is extracted, the effective cleaning range data is projected onto the spatial curvature distribution, the projection deformation coefficient is calculated, the effective cleaning range data is corrected according to the projection deformation coefficient, and the actual coverage area parameter is generated; According to the actual coverage area parameter, the overlap degree and the trajectory spacing parameter, the local trajectory offset is calculated, and the cleaning path coordinate points are generated based on the local trajectory offset and the projection deformation coefficient; Based on the cleaning path coordinate points, the included angle between the jet direction of the cleaning nozzle and the surface normal vector of the membrane wire is calculated, the local jet effective action distance is calculated based on the included angle, and the area range that needs to be optimized is determined according to the local jet effective action distance and the projection deformation coefficient; In the area range, the coordinate point optimization criterion is established, the density adjustment and position optimization of the cleaning path coordinate points are performed according to the coordinate point optimization criterion, and the cleaning trajectory curve that meets the local jet action requirement and the surface curvature constraint is generated.

7. The method of claim 6, wherein, In the area range, the coordinate point optimization criterion is established, the density adjustment and position optimization of the cleaning path coordinate points are performed according to the coordinate point optimization criterion, and the cleaning trajectory curve that meets the local jet action requirement and the surface curvature constraint is generated, including: In the area range, the topological connection features between the cleaning path coordinate points are calculated, the topological connection features are compared with the minimum turning radius of the trajectory, and the to-be-optimized cleaning sub-region that needs to be optimized is determined; The local surface geometric features in the to-be-optimized cleaning sub-region are extracted, the principal curvature and Gaussian curvature of the local surface are calculated, the coordinate point density weight coefficient is determined according to the numerical distribution of the principal curvature and the Gaussian curvature, and the coordinate point optimization criterion is generated by combining the density weight coefficient with the cleaning path coordinate points; The position information of the added coordinate points and the to-be-deleted coordinate points in the to-be-optimized cleaning sub-region is generated based on the coordinate point optimization criterion, and the position information is input into a coordinate point self-organizing optimization equation based on fluid diffusion principle to calculate the spatial position correction amount of each cleaning path coordinate point in the to-be-optimized cleaning sub-region; The cleaning path coordinate points are reconstructed according to the spatial position correction amount, the direction features of the coordinate point sequence after reconstruction are extracted, and a cleaning trajectory curve meeting the requirements of local jet action and surface curvature constraint is generated based on the direction features.

8. A hollow fiber membrane universal cleaning head trajectory intelligent planning system for implementing the method according to any one of claims 1-7, characterized in that, Comprise: A first unit for acquiring installation parameter information of a hollow fiber membrane bundle and generating three-dimensional to-be-cleaned region position data of the hollow fiber membrane bundle based on the installation parameter information; A second unit for sampling real-time working parameters of a universal cleaning head of the hollow fiber membrane, calculating an effective cleaning range of the cleaning nozzle based on the real-time working parameters, determining an adjacent cleaning trajectory overlap degree based on the effective cleaning range, and generating a trajectory spacing parameter for uniform cleaning of the surface of the hollow fiber membrane bundle according to the overlap degree; A third unit for dividing the surface of the hollow fiber membrane bundle into a plurality of cleaning sub-regions according to the three-dimensional to-be-cleaned region position data, prioritizing each cleaning sub-region according to the pollutant distribution density of the cleaning sub-region, and determining a cleaning sequence; A fourth unit for calculating cleaning path coordinate points of the plurality of cleaning sub-regions according to the effective cleaning range, the overlap degree, and the trajectory spacing parameter in the cleaning sequence, and determining a cleaning trajectory curve based on the cleaning path coordinate points; A fifth unit for converting the cleaning trajectory curve into a motion control instruction of the universal cleaning head of the hollow fiber membrane, and issuing the motion control instruction to a cleaning execution device.

9. An electronic device, comprising: Comprise: A processor; A memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions are executed by the processor to implement the method of any one of claims 1 to 7.

Images