Adaptive path planning control system based on environmental feature perception

Abstract

The present application relates to the technical field of cleaning path planning, and discloses an adaptive path planning control system based on environment feature perception, comprising: firstly, obtaining environment feature parameters of a surface to be cleaned, the parameters at least including temperature field parameters, near-wall flow field parameters and contact angle feature parameters; then, identifying a stable locking zone on the surface to be cleaned based on the environment feature parameters, and dividing the surface into multiple zone units with the locking zone as a boundary. The cleaning sequence is determined according to the topological relationship between the stable locking zone and the zone units, so as to ensure the continuity and area coverage of the cleaning process. Subsequently, the cleaning direction in each zone unit is determined according to the temperature field parameters, so that the cleaning path is consistent with the surface heat distribution; then, the cleaning row spacing of each zone unit is determined based on the environment feature parameters, so as to realize adaptive adjustment of the row spacing and prevent repeated cleaning or omission. Finally, a special cleaning operation is performed on the stable locking zone, so as to further eliminate liquid residues and recontamination.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cleaning path planning, more particularly, it relates to an adaptive path planning control system based on environmental feature perception. BACKGROUND

[0002] With the wide application of glass curtain walls and high-rise building facades, automatic cleaning technology has gradually become an important means of building maintenance. However, the aerodynamic environment and thermal environment of the near-vertical facade are very complex, and local negative pressure zones and wind speed amplification phenomena often occur at the corner and edge regions. At the same time, the metal frame and connecting parts of the building form a thermal bridge effect after being heated, resulting in a significant temperature gradient along the horizontal direction of the glass surface. Under this non-uniform temperature field and airflow condition, the surface tension of the cleaning liquid changes in space, thereby generating capillary flow along the temperature gradient direction of the surface. This flow will push the liquid to migrate from the high-temperature zone to the low-temperature zone and form liquid accumulation at the corner or edge region. In traditional cleaning path generation algorithms, the path distribution is usually based on the geometric coverage principle or graph segmentation, without considering the redistribution of the cleaning liquid on the vertical surface affected by temperature and airflow. Therefore, when cleaning is performed in a fixed direction, the liquid is deviated by wind shear force and surface tension gradient during the downward process, and is easily concentrated in certain fixed strip-shaped regions, making the cleaned area be polluted again, and the cleaning path planning loses timeliness.

[0003] Specifically, on the near-vertical cleaning surface, the flow and evaporation behavior of the cleaning liquid are affected by the following factors:

[0004] External aerodynamic effect: the wind speed enhancement at the corner and edge regions leads to uneven distribution of surface shear stress, causing the liquid to drift horizontally.

[0005] Surface thermal difference driving: the temperature gradient caused by the thermal bridge changes the surface tension of the liquid, forming a surface tension driven flow along the temperature difference direction.

[0006] Wetting hysteresis limitation: the difference in wetting performance between glass and metal boundaries causes contact line pinning effect, and part of the liquid is locked in a specific position to form a stable residual liquid zone.

[0007] When the above three effects are superimposed on the glass facade with an inclination angle of 60° to 90°, stable band-shaped accumulation zones appear in the liquid flow path. These accumulation zones have spatial fixity and time hysteresis, and will form new pollution sources after the cleaning path is covered. With the changes of external parameters such as sunlight and wind speed, the position of the liquid band will also suddenly migrate, making the traditional fixed path planning unable to adapt in real time, resulting in the misalignment of the cleaning path and the liquid migration band. SUMMARY

[0008] The application provides an adaptive path planning control system based on environmental feature perception, which solves the technical problem proposed in the background art: when performing facade operation, the cleaning path cannot be adjusted according to the migration state of the liquid, thereby causing the cleaning coverage to decrease and the recontamination phenomenon to continuously occur.

[0009] The application provides an adaptive path planning control system based on environmental feature perception, which includes:

[0010] A data acquisition module acquires environmental feature parameters of a surface to be cleaned, and the environmental feature parameters at least include temperature field parameters, near-wall flow field parameters and contact angle feature parameters.

[0011] A stable lock belt extraction module identifies a stable lock belt on the surface to be cleaned based on the environmental feature parameters.

[0012] A segmentation module divides the surface to be cleaned into a plurality of belt domain units with the stable lock belt as a boundary.

[0013] A cleaning sequence module determines the cleaning sequence of each belt domain unit according to the topological relationship between the stable lock belt and the belt domain unit.

[0014] A cleaning direction module determines the cleaning direction in each belt domain unit based on the temperature field parameters of the surface to be cleaned.

[0015] A cleaning row distance module determines the cleaning row distance in each belt domain unit based on the environmental feature parameters of the surface to be cleaned.

[0016] A cleaning module performs a cleaning operation on the stable lock belt.

[0017] The beneficial effects of the application include: a multi-field coupling model of the environment is constructed by using the temperature field, the near-wall flow field and the contact angle feature parameters, which can accurately identify the liquid convergence and retention position and generate a cleaning partition structure with stable boundary constraints. Compared with the traditional plane path planning, the application can dynamically determine the comprehensive influence of the Marangoni effect, airflow shear and contact line hysteresis on the liquid film movement, so that the cleaning path and row distance are adaptively adjusted according to the physical field distribution, thereby significantly reducing the overlap of the strip saliva and the uncovered area, improving the continuity and uniformity of the high-angle glass surface cleaning. At the same time, the application performs a guard trajectory and step constraint at the lock belt, which can effectively prevent liquid backflow and recontamination, and realize the overall optimal control of cleaning efficiency, energy consumption and surface cleanliness on a complex inclined surface. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a module diagram of the adaptive path planning control system based on environmental feature perception of the application. DETAILED DESCRIPTION

[0019] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide insight into the subject matter described herein, and the functions and arrangements of the elements discussed can be changed, without departing from the scope of the subject matter described in this specification. Various examples can omit, substitute, or add various procedures or components as appropriate, or in appropriate combination. Also, features described with respect to some examples can be combined in other examples.

[0020] As Figure 1 illustrated, the adaptive path planning control system based on environmental feature perception includes:

[0021] A data acquisition module acquires environmental feature parameters of the surface to be cleaned, the environmental feature parameters including at least temperature field parameters, near-wall flow field parameters, and contact angle feature parameters;

[0022] A stable lock zone extraction module identifies a stable lock zone on the surface to be cleaned based on the environmental feature parameters;

[0023] A segmentation module divides the surface to be cleaned into a plurality of zone units with the stable lock zone as a boundary;

[0024] A cleaning sequence module determines a cleaning sequence of each zone unit according to a topological relationship between the stable lock zone and the zone units;

[0025] A cleaning direction module determines a cleaning direction in each zone unit based on the temperature field parameters of the surface to be cleaned;

[0026] A cleaning row distance module determines a cleaning row distance in each zone unit based on the environmental feature parameters of the surface to be cleaned;

[0027] A cleaning module performs a cleaning operation on the stable lock zone.

[0028] In an embodiment of the present application, the environmental feature parameters of the surface to be cleaned, including at least temperature field parameters, near-wall flow field parameters, and contact angle feature parameters, are acquired by:

[0029] A coordinate system is established, in which a horizontal coordinate is a horizontal distance from a vertical edge of a facade, and a vertical coordinate is a distance along the direction of gravity;

[0030] An infrared image of the surface to be cleaned is acquired, and a temperature field is obtained from the infrared image, the temperature field being a distribution of temperature on a plane formed by the horizontal coordinate and the vertical coordinate;

[0031] A partial derivative of the temperature field along the horizontal coordinate is taken as a horizontal temperature gradient;

[0032] A free-stream wind speed is measured;

[0033] The corner near-wall wind speed distribution is calculated according to the incoming wind speed, including: the incoming wind speed multiplied by a preset corner opening amount coefficient, and then multiplied by a negative attenuation exponential raised to the power of (1 plus the ratio of the horizontal coordinate to a preset attenuation distance);

[0034] The near-wall shear stress distribution is calculated according to the corner near-wall wind speed distribution, including: 0.5 multiplied by the air density, multiplied by a preset near-wall friction coefficient, and then multiplied by the square of the corner near-wall wind speed distribution;

[0035] The advancing contact angle and the receding contact angle are measured;

[0036] The contact line hysteresis is calculated according to the advancing contact angle and the receding contact angle, including: the cosine value of the receding contact angle minus the cosine value of the advancing contact angle;

[0037] The temperature field and the temperature horizontal gradient are taken as temperature field parameters;

[0038] The corner near-wall wind speed distribution and the near-wall shear stress distribution are taken as near-wall flow field parameters;

[0039] The advancing contact angle, the receding contact angle, and the contact line hysteresis are taken as contact angle characteristic parameters.

[0040] The coordinate system is a reference for describing the position of the surface to be cleaned. The horizontal coordinate refers to the horizontal distance from a point on the surface to be cleaned (such as a glass curtain wall) to the vertical edge of the surface (such as the vertical frame of the curtain wall), with a unit of meters. The vertical coordinate refers to the distance of the point along the direction of gravity (i.e., the vertical direction), with a unit of meters, and the bottom of the surface is usually taken as the origin.

[0041] The infrared image is an image obtained by shooting the surface to be cleaned by an infrared thermal imager. Each pixel point in the image corresponds to the temperature information of a position on the surface. It is necessary to ensure that the image covers the entire area to be cleaned and has no dead angle. The resolution of the infrared image should be ≥320×240 pixels, corresponding to a pixel spacing of ≤2 cm×2 cm on the actual surface, to ensure that the error of the subsequent temperature horizontal gradient calculation is ≤0.5°C / m.

[0042] The temperature field refers to the collection of temperature values corresponding to all positions (determined by the horizontal coordinate and the vertical coordinate) on the surface to be cleaned, reflecting the spatial distribution of the surface temperature, with a unit of degrees Celsius. For example, the temperature at a horizontal distance of 0.5 meters and a vertical distance of 2 meters is 28 degrees Celsius.

[0043] The temperature transverse gradient refers to the rate of change of temperature along the transverse coordinate direction, reflecting the amount of temperature change per 1 meter of transverse position change, with the unit of degree Celsius per meter. For example, a transverse gradient of -2 degrees Celsius per meter indicates that the temperature decreases by 2 degrees Celsius per meter away from the vertical edge along the transverse direction. The temperature transverse gradient is selected as a parameter to specifically capture the transverse temperature difference caused by heat bridges (such as vertical frames), which is the driving source of the Malmquist effect, avoiding redundant data caused by omnidirectional gradient calculation. Specifically, the temperature transverse gradient is equal to the temperature of the right adjacent point minus the temperature of the point, divided by the transverse distance between the two points (usually taking the actual transverse distance corresponding to the infrared image pixels, such as 0.02 meters).

[0044] The incoming wind speed refers to the speed of natural wind that is not blocked by the facade in the environment where the facade to be cleaned is located, with the unit of meters per second. It needs to be measured by a wind speed meter at an unobstructed place within 10 meters of the facade, for more than 5 minutes, and the average value is taken.

[0045] The corner near-wall wind speed distribution refers to the wind speed of each transverse position near the corner area (such as the windward corner of the curtain wall) of the facade to be cleaned, with the unit of meters per second, reflecting the decay law of the corner wind speed with the transverse distance. The preset corner amplification coefficient value range is 1.2-1.8, 1.6-1.8 for right-angle curtain wall corners, and 1.2-1.4 for round-angle curtain wall corners (radius ≥ 5 cm), which is set according to the experimental data of corner airflow acceleration coefficient in building wind engineering. The preset decay distance value range is 0.3-1.0 meters, 0.3-0.5 meters for curtain wall height ≤ 10 meters, 0.6-0.8 meters for 10-20 meters, and 0.9-1.0 meters for > 20 meters, which is set according to the conclusion of wind tunnel experiment that near-wall airflow decay length is positively correlated with building height. Specifically, the corner near-wall wind speed is equal to the incoming wind speed multiplied by the preset corner amplification coefficient, and then multiplied by (1 plus the transverse coordinate divided by the preset decay distance), and the result is taken to the negative decay index power (the decay index is usually 1.5-2.0, determined by the geometric complexity of the curtain wall corner, 2.0 for right-angle and 1.5 for round-angle).

[0046] The near-wall shear stress distribution refers to the shear force intensity generated by air flowing through the surface to be cleaned on the surface liquid, with the unit of pascal, reflecting the driving force size of airflow on liquid transverse drift. The preset near-wall friction coefficient value range is 0.012-0.015, 0.012-0.013 for smooth glass surface without residue after cleaning, and 0.014-0.015 for glass surface with slight stains without cleaning, which is set according to the standard value of friction coefficient of smooth solid surface in aerodynamics. The air density is 1.205 kg / m3 at 25°C under standard atmospheric pressure. Specifically, the near-wall shear stress is equal to 0.5 times the air density, multiplied by the preset near-wall friction coefficient, and then multiplied by the square of the corner near-wall wind speed.

[0047] The advancing contact angle is the angle between the droplet and the surface at the contact line when the droplet is moving along the spreading direction, unit is degree, reflecting the difficulty of liquid spreading on the surface. The measurement needs to use the sessile drop method, slowly drop the cleaning liquid on the surface, read the data when pushing the droplet to spread; the measurement environment conditions are temperature 25±2 degrees Celsius, relative humidity 40%-60%, to avoid the evaporation of the liquid caused by too high temperature and the dryness of the surface caused by too low humidity, ensuring that the measurement error is ≤2 degrees.

[0048] The receding contact angle is the angle between the droplet and the surface at the contact line when the droplet is moving along the shrinking direction, unit is degree, reflecting the difficulty of liquid shrinking on the surface. The measurement needs to use the sessile drop method, slowly suck back part of the liquid after the advancing contact angle measurement, read the data when pushing the droplet to shrink; the measurement environment conditions are consistent with the advancing contact angle (temperature 25±2 degrees Celsius, relative humidity 40%-60%), ensuring that the measurement error is ≤2 degrees.

[0049] The contact line hysteresis is a parameter for measuring the resistance of the cleaning liquid to the movement of the contact line on the surface to be cleaned, without unit, the larger the value, the more difficult the contact line moves, and the more likely the liquid stays. Specifically, the contact line hysteresis is equal to the cosine value of the receding contact angle minus the cosine value of the advancing contact angle (the angle needs to be converted to the cosine value first, such as 60 degrees cosine value is 0.5, 80 degrees cosine value is 0.1736).

[0050] In an embodiment of the present application, based on the environmental characteristic parameters, a stable locking zone on the surface to be cleaned is identified, comprising:

[0051] The Marangoni shear stress is calculated, including: the surface tension temperature coefficient multiplied by the temperature transverse gradient;

[0052] The thread saliva width is measured;

[0053] The unit length retention force is calculated, including: the surface tension of the cleaning liquid multiplied by the contact line hysteresis;

[0054] The transverse resultant force function is formed, including: (Marangoni shear stress plus near-wall shear stress distribution) multiplied by thread saliva width, and then subtracting the unit length retention force;

[0055] On each fixed vertical coordinate, the transverse position function is solved, which is that the transverse resultant force function is equal to zero and the partial derivative of the transverse resultant force function along the transverse coordinate is less than zero;

[0056] The transverse position function is taken as the spatial trajectory of the stable locking zone.

[0057] Marangoni shear stress is the tangential stress of the free surface of the cleaning liquid caused by the temperature gradient, with the unit of Pascal, reflecting the driving force of the surface tension gradient on the liquid. Among them, the surface tension temperature coefficient is a parameter describing the change of the surface tension of the cleaning liquid with temperature, with the unit of Newton per meter per degree Celsius, and the value range of water-based glass cleaner (common cleaning liquid) at 25 degrees Celsius is-0.00015 to-0.0002 Newton per meter per degree Celsius. The negative sign indicates that the surface tension decreases with the increase of temperature, which is set according to the physical experimental data of the liquid surface. Specifically, the Marangoni shear stress is equal to the absolute value of the surface tension temperature coefficient multiplied by the transverse temperature gradient (since the surface tension temperature coefficient is negative, the absolute value is taken to ensure that the stress direction matches the temperature gradient direction).

[0058] The width of the strip is the transverse length of the non-continuous strip liquid formed by the cleaning liquid on the surface to be cleaned, with the unit of meter, reflecting the transverse dimension of the liquid aggregation. High-definition industrial camera (resolution ≥1920×1080 pixels) is needed to take pictures of the surface liquid film, and the edge of the strip is extracted by image recognition software (such as OpenCV) to calculate the transverse distance between the edges. When measuring, the middle of the strip should be focused (to avoid the influence of edge deformation on accuracy), and the average value of three measurements is taken as the final width.

[0059] The unit length retention force is the retention resistance of the contact line of the cleaning liquid caused by wetting hysteresis, with the unit of Newton per meter, which prevents the transverse migration of the liquid. Among them, the surface tension of the cleaning liquid is the attractive force between the surface molecules of the liquid, with the unit of Newton per meter, and the value range of water-based glass cleaner at 25 degrees Celsius and relative humidity of 40%-60% is 0.07 to 0.073 Newton per meter, which can be measured by a surface tension meter. Specifically, the unit length retention force is equal to the surface tension of the cleaning liquid multiplied by the contact line hysteresis difference.

[0060] The transverse force function is a function describing the total driving force and resistance of the cleaning liquid in the transverse position, with the unit of Newton per meter, and the positive value indicates that the liquid has a tendency to migrate along the transverse direction, and the negative value indicates that the liquid is retained. It should be noted that the total transverse driving force (both taking positive values, because the directions are consistent, pointing to the liquid convergence area) is the sum of the Marangoni shear stress and the near-wall shear stress distribution. Specifically, the transverse force function is equal to (Marangoni shear stress plus near-wall shear stress) multiplied by the width of the strip, and then minus the unit length retention force.

[0061] The partial derivative of the transverse force function along the transverse coordinate is a parameter describing the rate of change of the transverse force with the transverse position, with the unit of Newton per square meter, reflecting the stability of the force. The partial derivative is less than zero, indicating that when the liquid deviates from the transverse position, the force will pull it back, ensuring the stability of the position. Specifically, the partial derivative of the transverse force function along the transverse coordinate is equal to the difference between the forces of the adjacent two transverse positions divided by the transverse distance between the two positions (the transverse distance is 0.01 meters, to ensure the calculation accuracy).

[0062] The transverse position function refers to a set of transverse positions at each fixed vertical height (such as a vertical coordinate of 0.5 m, 1 m) that satisfy the conditions of zero resultant force and a partial derivative less than zero, in units of meters, reflecting the transverse positions of stable convergence of the liquid at different heights. Specifically, within the transverse coordinate range (0 to the transverse width of the curtain wall), the interval where the resultant force changes from positive to negative is first determined, and then the range is gradually narrowed within the interval to find the position where the resultant force is close to zero (error ≤ 0.001 N / m) and the partial derivative is less than zero.

[0063] The spatial trajectory of the stable locking band is a curve formed by connecting the transverse position functions corresponding to all vertical heights, reflecting the band-shaped area of stable convergence of the cleaning liquid on the surface. The trajectory needs to be continuous and have no breakpoints. If there is no transverse position at a certain vertical height that meets the conditions, there is no stable locking band at that height.

[0064] In an embodiment of the present application, the surface to be cleaned is divided into a plurality of band domain units with the stable locking band as the boundary, including:

[0065] Determining the number of stable locking bands;

[0066] At each fixed vertical coordinate, the transverse position functions of the stable locking bands are sorted in ascending order to obtain the sorted stable locking band transverse position functions;

[0067] Determining the area of the surface to be cleaned;

[0068] Defining the geometric set of each stable locking band as the set of points in the area of the surface to be cleaned whose transverse coordinates are equal to the corresponding sorted stable locking band transverse position functions;

[0069] Dividing the band domain unit set, including:

[0070] The first band domain unit is the set of points in the area of the surface to be cleaned whose transverse coordinates are less than the transverse position function of the first sorted stable locking band;

[0071] The middle band domain unit is the set of points in the area of the surface to be cleaned whose transverse coordinates are between the transverse position functions of adjacent two sorted stable locking bands;

[0072] The last band domain unit is the set of points in the area of the surface to be cleaned whose transverse coordinates are greater than the transverse position function of the last sorted stable locking band;

[0073] The area of the surface to be cleaned is the union of all band domain units and the geometric set of all stable locking bands, and each band domain unit is mutually exclusive.

[0074] The number of stable locking bands refers to the total number of independent horizontal position functions that satisfy the horizontal force function equal to zero and the partial derivative less than zero on the surface to be cleaned. Specifically, count the number of horizontal position functions that exist at all vertical coordinates (covering the full height of the curtain wall), and take the highest frequency number (must exist in more than 80% of the vertical coordinates to avoid false judgments caused by local missing), for example, if there are 2 horizontal position functions at 90% of the vertical coordinates, the number is 2; if only locally there are 3, and the frequency is less than 20%, then ignore the local, and take the high-frequency number.

[0075] The sorted horizontal position function refers to the ordered horizontal position function group formed by arranging all stable locking bands in ascending order of horizontal coordinate value at the same vertical coordinate. For example, at a vertical coordinate z=2 meters, the horizontal positions of 3 locking bands are 0.4 meters, 0.9 meters, and 1.5 meters, sorted in ascending order as 0.4 meters, 0.9 meters, and 1.5 meters, respectively. Each vertical coordinate is sorted according to this rule to ensure the order of the horizontal position.

[0076] The area of the surface to be cleaned refers to the physical range of the curtain wall that needs to be cleaned, defined by the horizontal and vertical coordinates. Specifically, the horizontal boundary range of the surface to be cleaned is from the horizontal coordinate 0 meters of the left vertical edge of the curtain wall to the horizontal coordinate of the right vertical edge (measured by a tape measure, denoted as the actual total width of the curtain wall); the vertical boundary range is from the vertical coordinate 0 meters at the bottom of the curtain wall to the vertical coordinate at the top (measured by a tape measure, denoted as the actual total height of the curtain wall); the final area is a rectangular range from horizontal 0 meters to the actual total width, and vertical 0 meters to the actual total height.

[0077] The geometric set of stable locking bands refers to the set of all spatial points corresponding to a single sorted horizontal position function within the cleaning area. For example, the first sorted horizontal position function is y1(z), and its geometric set is all points (y1(z), z) that satisfy the horizontal coordinate = y1(z), the vertical coordinate z between 0 meters and the actual total height, and y1(z) between 0 meters and the actual total width, ensuring that all points in the set are within the cleaning area.

[0078] The belt domain set is an independent cleaning area formed by dividing the area to be cleaned according to the sorted locking belt transverse position function, and each unit does not contain the locking belt (the locking belt is used as a boundary alone). For example, if there are two sorted transverse position functions y1(z) and y2(z), the first unit is the set of points with a transverse coordinate < y1(z) and a vertical coordinate between 0 meters and the actual total vertical height; the middle unit (there is no middle unit here, and two locking belts correspond to two interval units) does not exist; the last unit is the set of points with a transverse coordinate > y2(z) and a vertical coordinate between 0 meters and the actual total vertical height; if there are three locking belts, there is one middle unit (the transverse coordinate is between y2(z) and y3(z)).

[0079] The union condition means that all points of the area to be cleaned belong to a certain belt domain unit or a certain geometric set of stable locking belts, with no missing points; the disjoint condition means that there are no overlapping points between any two belt domain units (i.e., a point belongs to only one belt domain unit), for example, the maximum transverse coordinate of the first unit is y1(z), and the minimum transverse coordinate of the last unit is y2(z), y1(z) < y2(z), and there is no overlap between them.

[0080] In an embodiment of the present application, the cleaning order of each belt domain unit is determined according to the topological relationship between the stable locking belt and the belt domain unit, including:

[0081] For any two adjacent belt domain units, respectively denoted as the first belt domain unit and the second belt domain unit;

[0082] Determine the stable locking belt corresponding to the common boundary of the first belt domain unit and the second belt domain unit;

[0083] Denote the two sides of the common boundary as the first side and the second side, respectively, and take the position with a transverse coordinate equal to the transverse position function of the stable locking belt minus a very small displacement on the first side, and take the position with a transverse coordinate equal to the transverse position function of the stable locking belt plus a very small displacement on the second side; wherein the very small displacement is taken as a fixed fraction of the infrared image pixel spacing;

[0084] Calculate the unit length transverse force of the first side position and the second side position, respectively;

[0085] Take the closed interval of the intersection of the vertical projections of the first belt domain unit and the second belt domain unit as the effective vertical interval;

[0086] Calculate the directionality measure, including: integrating (the unit length transverse force of the first side minus the unit length transverse force of the second side) over the effective vertical interval;

[0087] When the directionality measure is greater than zero, generate a directed edge of the first belt domain unit pointing to the second belt domain unit;

[0088] generating a directed edge from the second band unit to the first band unit when the directivity measure is less than zero;

[0089] constructing a directed graph, vertices of the directed graph being the band units in the set of band units, edges of the directed graph being the generated directed edges;

[0090] defining a level value of each band unit, the level value being a maximum value of the number of edges in all directed chains ending with the band unit;

[0091] determining a cleaning order of the band units in an order of the level values from small to large;

[0092] for a plurality of band units with the same level value, taking any one of the band units as a target band unit, calculating a minimum normal distance from the target band unit to any stable locking band, the minimum normal distance being a minimum value of absolute values of differences between transverse coordinates of points in the target band unit and transverse position functions of the stable locking band at corresponding vertical coordinates;

[0093] determining a cleaning order of the band units with the same level value in an order of the minimum normal distances from large to small.

[0094] Two adjacent band units refer to two band units sharing the same stable locking band as a common boundary on the surface to be cleaned, without other band units between the two. For example, band unit A (transverse < y1(z)) and band unit B (transverse between y1(z) and y2(z)) share the stable locking band y1(z) as a common boundary, and are therefore adjacent units, denoted as first and second band units respectively.

[0095] The stable locking band corresponding to the common boundary refers to the stable locking band separating the two adjacent band units after sorting, and the geometric set of the locking band is the common boundary of the two units. For example, the common boundary of the first unit (transverse < y1(z)) and the second unit (transverse between y1(z) and y2(z)) corresponds to the first stable locking band (transverse position function y1(z)) after sorting.

[0096] The first side is the side of the common boundary facing the first band unit, and the second side is the side facing the second band unit. For example, the common boundary is y1(z), the first unit is transverse < y1(z), and the first side is the left side of y1(z) (facing the first unit), and the second side is the right side of y1(z) (facing the second unit). The positions of the two sides need to be within the band unit to ensure that the calculated resultant force can reflect the force characteristics of the corresponding unit.

[0097] The minimum displacement is a small lateral distance for taking points on both sides of the common boundary, in meters, which needs to be much smaller than the strip width (to avoid exceeding the unit range) and ensures that the points are taken within the strip domain unit. Specifically, the actual pixel spacing of the infrared image is first obtained (e.g., 0.02 meters), and the fixed fraction is set to 1 / 5 to 1 / 10 (depending on the pixel spacing accuracy, the smaller the spacing, the larger the fraction can be), and the minimum displacement is equal to the pixel spacing of the infrared image multiplied by the fixed fraction (for example, the pixel spacing is 0.02 meters, the fixed fraction is 1 / 5, and the minimum displacement is 0.004 meters).

[0098] The first side position is a specific spatial point on the first side of the common boundary, with coordinates (stable locking band lateral position function - minimum displacement, vertical coordinate z); the second side position is a specific spatial point on the second side of the common boundary, with coordinates (stable locking band lateral position function + minimum displacement, vertical coordinate z), and z needs to be within the vertical range of the two strip domain units.

[0099] The unit length lateral force is the total lateral force per unit length of the cleaning liquid at the first side and second side positions, in Newton per meter, used to judge the migration trend of the liquid on both sides.

[0100] The effective vertical interval is a continuous vertical coordinate interval that belongs to both the first and second strip domain units, in meters, and only in this interval is the directional measure calculated to ensure that the result reflects the vertical range common to both units. Specifically, the vertical projection of the first strip domain unit (vertical coordinate range, e.g., 0.5 meters to 10 meters) and the vertical projection of the second strip domain unit (e.g., 0.5 meters to 10 meters) are first determined, and the overlapping part of the vertical range of the two is taken (if the first is 0.5-10 meters and the second is 1-9 meters, the overlapping part is 1-9 meters), and the closed interval (including endpoints) of this overlapping part is the effective vertical interval.

[0101] The directional measure is the core indicator for judging the cleaning order of adjacent strip domain units, in Newton (the integral result of meters cancels out per meter), a positive value indicates that the liquid migrates from the first unit to the second unit, and a negative value indicates the opposite. Specifically, the effective vertical interval is divided into several small sections with a step size of 0.1 meters, and the average value of (first side force - second side force) at both endpoints of each small section is taken, multiplied by the small section length (0.1 meters), and the sum of all small section results is the directional measure; the step size can be adjusted according to the vertical range, and the smaller the range, the smaller the step size (e.g., 0.05 meters) to ensure the accuracy of the integral.

[0102] The directed edge is a directed line segment representing the priority of cleaning the strip domain unit, with the arrow pointing to the unit that needs to be cleaned later (to avoid the unit being cleaned first being contaminated by the migrated liquid of the unit cleaned later). For example, if the directional measure > 0, the liquid migrates from the first unit to the second unit, so the first unit is cleaned first and the second unit is cleaned later, generating a directed edge from the first to the second.

[0103] The directed graph is a graph composed of domain units (vertices) and directed edges (cleaning priority relations), which is used to visually show the cleaning sequence logic of all units, without vertex isolation (all units are vertices) and without circular edges (to avoid cleaning sequence contradictions, such as A→B and B→A).

[0104] The level value is a numerical value representing the cleaning sequence of the domain unit (the larger the value, the later the cleaning), with a unit of strips (the number of edges), and reflects the length of the longest path from a unit without an incoming edge (cleaned first) to the unit. Specifically, identify the unit without an incoming edge in the directed graph (the level value is initially 0); then find all units that can be reached from the unit without an incoming edge via 1 directed edge, with a level value of 1; then find units that can be reached via 2 directed edges, with a level value of 2; and so on, until all units have a level value, and the final level value is the maximum number of edges in all reachable paths (for example, unit C can be reached via A→B→C (2 edges) or A→C (1 edge), with a level value of 2).

[0105] The target domain unit refers to a single unit in the domain unit with the same level value, for which the minimum normal distance needs to be calculated, which is used to determine the cleaning sequence of units in the same level, and the notation can be arbitrary, and other units in the same level are calculated in turn after one unit is calculated.

[0106] The minimum normal distance is the shortest transverse distance between the target domain unit and the stable locking zone (since the locking zone extends vertically, the normal direction is transverse), with a unit of meters, and the larger the value, the farther the unit is from the locking zone, which needs to be cleaned first (to reduce the pollution of the migrated liquid to the cleaned unit). Specifically, sample points are taken in the target domain unit, with a transverse distance of 0.05 meters and a vertical distance of 0.1 meters, calculate the absolute value of the difference between the transverse coordinate of each sampling point and the vertical coordinate of all stable locking zone transverse position functions, and select the minimum value from all absolute values, which is the minimum normal distance (the number of sampling points needs to be ≥50 to ensure coverage of the entire unit and avoid local extreme value misjudgment).

[0107] In an embodiment of the present application, based on the temperature field parameters of the surface to be cleaned, the cleaning direction in each domain unit is determined, including:

[0108] For each position in the temperature field, a temperature gradient vector is calculated, the first component of the temperature gradient vector is the partial derivative of the temperature field along the transverse coordinate, and the second component of the temperature gradient vector is the partial derivative of the temperature field along the vertical coordinate;

[0109] The temperature gradient module of each position is calculated, which is the module of the temperature gradient vector at the corresponding position;

[0110] The isothermal normal unit vector of each position is calculated, including: the temperature gradient vector at the corresponding position is divided by the temperature gradient module at the position;

[0111] The isothermic tangential unit vector of each position is calculated, the first component of the isothermic tangential unit vector is the negative value of the second component of the isothermic normal unit vector, and the second component of the isothermic tangential unit vector is the first component of the isothermic normal unit vector;

[0112] For each band element, the isothermic tangential unit vector of each position in the band element is taken as the cleaning direction of the corresponding position in the band element to form a cleaning direction field in each band element.

[0113] The temperature gradient vector is a vector describing the direction and intensity of temperature change in a certain position in the horizontal and vertical directions, with a unit of degrees Celsius per meter. The first component reflects the horizontal temperature change rate, and the second component reflects the vertical temperature change rate. The partial derivative of the temperature field along the vertical coordinate is specifically equal to the temperature of the adjacent point above minus the temperature of the point at a certain vertical position, and then divided by the vertical distance between the two points (taking the actual vertical distance corresponding to the infrared image pixel, such as 0.02 meters). The overall calculation of the temperature gradient vector needs to ensure that the two components correspond to the same spatial position (the same horizontal and vertical coordinates).

[0114] The temperature gradient module is the size of the temperature gradient vector, with a unit of degrees Celsius per meter, reflecting the overall intensity of temperature change at that position (irrespective of direction). The larger the module value, the more intense the temperature change. Specifically, the temperature gradient module is equal to the square of the first component of the temperature gradient vector plus the square of the second component, and the arithmetic square root of the sum is taken. If the calculation result is less than 0.01 degrees Celsius per meter (considered as uniform temperature), the module value is taken as 0.01 degrees Celsius per meter (to avoid errors in subsequent division operations).

[0115] The isothermic normal unit vector is a vector along the temperature gradient direction with a length of 1, without unit, and the direction points to the direction of temperature rise (since the gradient vector points to the direction of temperature rise), which is used for subsequent derivation of the cleaning direction along the isothermic line. If the temperature gradient module is 0.01 degrees Celsius per meter (uniform temperature), the isothermic normal unit vector is taken as the vertical upward unit vector by default (first component 0, second component 1), ensuring the uniqueness of the direction. Specifically, the first component of the isothermic normal unit vector is equal to the first component of the temperature gradient vector divided by the temperature gradient module, and the second component is equal to the second component of the temperature gradient vector divided by the temperature gradient module. After calculation, it is necessary to verify whether the square sum of the components is close to 1 (error ≤0.001) to ensure that it is a unit vector.

[0116] The isothermic tangential unit vector is a vector along the tangent direction of the isothermic line with a length of 1, no unit, perpendicular to the isothermic normal unit vector (to ensure that the cleaning direction is along the isothermic line and reduce liquid migration caused by temperature gradient), and the direction is counterclockwise rotation of 90 degrees around the temperature rising direction. If the isothermic normal unit vector is vertical upward (when the temperature is uniform), the first component of the isothermic tangential unit vector is -1 and the second component is 0 (horizontal left), which ensures the uniformity of the direction. Specifically, the first component of the isothermic tangential unit vector is equal to the negative second component of the isothermic normal unit vector, and the second component is equal to the first component of the isothermic normal unit vector; for example, the first component of the normal vector is 0.6 and the second component is 0.8, the first component of the tangential vector is -0.8 and the second component is 0.6.

[0117] The cleaning direction field is a collection of cleaning directions at all positions in the zone unit, reflecting the spatial distribution of the cleaning direction in the unit, and needs to ensure that the direction changes continuously at adjacent positions (no sudden change, to avoid frequent turning of the equipment). If the temperature gradient modulus in a certain area in the unit is ≤0.01 degrees Celsius per meter (uniform temperature), the cleaning direction in the area is uniformly set to the default value of the isothermic tangential unit vector (horizontal left), to ensure that the direction field is not chaotic.

[0118] In an embodiment of the present application, based on the environmental characteristic parameters of the surface to be cleaned, the cleaning row distance in each zone unit is determined, including:

[0119] The dynamic viscosity of the cleaning liquid, the effective width of the cleaning head, the downlink speed of the cleaning head, and the single downlink stroke height are obtained;

[0120] The total lateral shear stress at each position is calculated, including the Marangoni shear stress at the corresponding position plus the near-wall shear stress at the position;

[0121] The lateral velocity at each position is calculated, including a preset empirical coefficient multiplied by the total lateral shear stress, then multiplied by the saliva thickness, and the result divided by the dynamic viscosity of the cleaning liquid;

[0122] The single downlink stroke time is calculated, including the single downlink stroke height divided by the downlink speed of the cleaning head;

[0123] The lateral drift at each position is calculated, including the lateral velocity at the corresponding position multiplied by the single downlink stroke time;

[0124] For each zone unit, the maximum lateral drift in the zone unit is determined, which is the maximum value of the lateral drift at all positions in the zone unit;

[0125] The cleaning row distance of each zone unit is determined, which is not greater than the effective width of the cleaning head minus the maximum lateral drift of the corresponding zone unit.

[0126] The dynamic viscosity of the cleaning liquid is a physical quantity that describes the flow resistance of the cleaning liquid, with units of pascal-seconds, and reflects the ease of flow of the liquid under shear. The water-based glass cleaner (a common cleaning liquid) has a value range of 0.001 to 0.0015 pascal-seconds at 25±2 degrees Celsius, which needs to be measured using a rotational viscometer. Ensure that the liquid is free of air bubbles and the temperature is stable (avoid temperature fluctuations affecting the viscosity) during measurement.

[0127] The effective width of the cleaning head is the actual length of the cleaning head that plays a cleaning role, with units of meters, i.e., the effective contact width of the rubber wiper strip of the cleaning head (1-2 millimeters of wear allowance needs to be deducted). Specifically, use a tape measure to measure along the horizontal direction of the wiper strip, take the average of the measurement values of 3 evenly distributed points (such as the left end, the middle, and the right end), and take the average as the final effective width. The typical value is 0.15 to 0.3 meters (adapted to the conventional size of curtain wall cleaning equipment).

[0128] The downstroke speed of the cleaning head is the vertical movement speed of the equipment along the surface to be cleaned, with units of meters per second. A speed that is too high can result in incomplete cleaning (the liquid is not completely wiped off), and a speed that is too low can reduce efficiency. The set range is 0.01 to 0.03 meters per second, with water-based glass cleaner at 0.02 meters per second, which is set based on the wiping ability of the cleaning head wiper strip (wiper strip hardness of 50-60 Shore A) and the evaporation rate of the cleaning liquid.

[0129] The single downstroke height is the distance of one continuous vertical downstroke of the equipment, with units of meters, to avoid frequent turning (to improve efficiency). Specifically, take 1 to 1.5 times the effective width of the cleaning head (e.g., 0.2 meters for the effective width, 0.2 to 0.3 meters for the stroke height), and if there are protruding structures (such as keels) on the curtain wall, the stroke height should not exceed 1 / 2 of the distance between the protrusions (to avoid missing cleaning due to crossing the protrusions).

[0130] The total lateral shear stress is the total driving force intensity that drives the cleaning liquid to drift laterally, with units of pascals, reflecting the combined driving effect of air flow and temperature gradient on the liquid. Specifically, the total lateral shear stress is equal to the corresponding location's Marangoni shear stress plus the near-wall shear stress at that location.

[0131] The lateral velocity is the movement speed of the cleaning liquid under the action of the total lateral shear stress, with units of meters per second, reflecting the speed of the liquid's lateral drift. The preset empirical coefficient has a value range of 0.1 to 0.3, with water-based glass cleaner at 0.2, which is set based on the ratio of the liquid's viscosity to the surface tension (the greater the viscosity, the smaller the coefficient) and experimental data; the saliva thickness is the vertical thickness of the saliva, with units of meters, which is measured using a laser thickness gauge at the middle of the saliva (to avoid edge deformation), taking the average of 3 measurements, with a typical value of 0.0005 to 0.001 meters. Specifically, the lateral velocity is equal to the preset empirical coefficient multiplied by the total lateral shear stress, then multiplied by the saliva thickness, and the result is divided by the dynamic viscosity of the cleaning liquid.

[0132] Single downstroke time is the time taken by the device to complete a single downstroke, in seconds, used to calculate the lateral drift of the liquid in that time period. Specifically, single downstroke time is equal to single downstroke height divided by downstroke speed of the cleaning head.

[0133] Lateral drift is the distance that the cleaning liquid moves in the lateral direction during a single downstroke of the device, in meters, reflecting the degree to which the liquid deviates from the cleaning path. Specifically, lateral drift is equal to the lateral velocity at the corresponding position multiplied by the single downstroke time.

[0134] Maximum lateral drift is the maximum value of the lateral drift at all positions within the belt domain, in meters, used to determine the safe cleaning pitch (to ensure that adjacent paths are not missed). Specifically, the lateral drift at each sampling point is calculated by uniformly sampling every 0.05 meters laterally and every 0.1 meters vertically within the belt domain (the number of sampling points ≥ 30, covering the entire domain), and the maximum value of all the drifts is selected as the maximum lateral drift.

[0135] Cleaning pitch is the lateral distance between two adjacent cleaning paths, in meters, which needs to ensure that the liquid can still be covered by the adjacent path after drifting (no missed cleaning). If the effective width of the cleaning head minus the maximum lateral drift ≤ 0.01 meters (too narrow and easy to repeat cleaning), then the cleaning pitch is taken as 0.01 meters (the minimum safe distance). Specifically, the cleaning pitch is not greater than the effective width of the cleaning head minus the maximum lateral drift of the corresponding belt domain.

[0136] In an embodiment of the present application, the cleaning operation is performed on the stable locking belt, including:

[0137] Defining the center trajectory of each stable locking belt, which is a trajectory with a lateral coordinate equal to the lateral position function of the corresponding stable locking belt and a vertical coordinate that changes synchronously with the lateral position function;

[0138] Calculating the first derivative of the lateral position function of each stable locking belt;

[0139] Calculating the tangent unit vector of each stable locking belt, including: first calculating the square of 1 plus the first derivative of the lateral position function of the stable locking belt, taking the square root of the result as the denominator, the first component of the tangent unit vector is the first derivative of the lateral position function of the stable locking belt, and the second component is 1, then dividing the two components by the denominator;

[0140] Calculating the normal unit vector of each stable locking belt, including: taking the square root of the sum of 1 and the square of the first derivative of the lateral position function of the stable locking belt as the denominator, the first component of the normal unit vector is 1, and the second component is the negative of the first derivative of the lateral position function of the stable locking belt, then dividing the two components by the denominator;

[0141] extracting the transverse component of the normal unit vector of each stable locking band;

[0142] defining a neighborhood set of each stable locking band, which is a set of all points within the surface to be cleaned whose absolute value of the difference between the transverse coordinate and the transverse position function of the corresponding stable locking band is not greater than the neighborhood half-width;

[0143] calculating the normal drift of each position in the neighborhood set relative to the corresponding stable locking band, including: the transverse drift of the corresponding position multiplied by the absolute value of the transverse component of the normal unit vector of the stable locking band at that position;

[0144] determining the guard offset distance of each stable locking band at each vertical coordinate, which is the maximum value of the normal drift of all positions in the neighborhood set of the corresponding stable locking band;

[0145] generating two guard offset trajectories for each stable locking band, including: taking the center trajectory of the corresponding stable locking band as the reference, offsetting in the positive and negative directions of the normal unit vector respectively by the guard offset distance at the corresponding vertical coordinate to obtain the two guard offset trajectories;

[0146] calculating the second derivative of the transverse position function of each stable locking band;

[0147] calculating the curvature of the center trajectory of each stable locking band, including: first calculating the square of 1 plus the first derivative of the transverse position function of the stable locking band, taking the two-thirds power of the result as the denominator, and then using the absolute value of the second derivative of the transverse position function of the stable locking band as the numerator, dividing the numerator by the denominator to obtain the curvature at the corresponding position;

[0148] determining the maximum curvature of the center trajectory of each stable locking band, and setting an upper limit for the geometric approximation error;

[0149] calculating the discrete step length when cleaning each stable locking band, specifically, 2 times the upper limit of the geometric approximation error, divided by the maximum curvature of the corresponding stable locking band, taking the square root of the result, and the discrete step length when cleaning is not greater than the square root.

[0150] The stable locking band is the geometric center line of the stable locking band, with a unit of meters (coordinate unit), reflecting the spatial trend of the locking band, ensuring that the center is taken as the reference to cover both sides when cleaning. The trajectory needs to satisfy that each vertical coordinate corresponds to a unique transverse coordinate, with no intersection or breakpoint (if there is no transverse position function for a vertical coordinate, there is no center trajectory for that segment).

[0151] The first derivative of the transverse position function of the stable locking strip is a parameter that describes the transverse variation rate of the center trajectory at a certain vertical coordinate, has no unit, and reflects the degree of inclination of the trajectory (a positive derivative indicates that the transverse coordinate increases with the vertical, and a negative derivative indicates the opposite). Specifically, take two adjacent vertical coordinates (with a spacing of 0.05 meters to ensure calculation accuracy), and obtain the transverse position function values corresponding to the vertical coordinates, respectively. The first derivative is equal to the transverse position value of the upper vertical coordinate minus the transverse position value of the lower vertical coordinate divided by the spacing of the two vertical coordinates. If the adjacent transverse position values are unchanged (the derivative is close to 0), the first derivative is 0.

[0152] The tangential unit vector is a vector along the tangent direction of the center trajectory with a length of 1, has no unit, and is used to control the direction of the cleaning device along the trajectory (to ensure that the path follows the trajectory). After calculation, it is necessary to verify whether the square of the first component plus the square of the second component is close to 1 (error ≤0.001) to ensure that it is a unit vector. If the first derivative is 0 (the trajectory is horizontal), the first component of the tangential unit vector is 0 and the second component is 1 (vertically upward). Specifically, first calculate 1 plus the square of the first derivative, and take the arithmetic square root of the sum as the denominator. The first component of the tangential unit vector is equal to the first derivative divided by the denominator, and the second component is equal to 1 divided by the denominator.

[0153] The normal unit vector of the stable locking strip is a vector perpendicular to the tangent direction of the center trajectory with a length of 1, has no unit, and is used to determine the direction of the guard sweeping offset trajectory (pointing to both sides of the trajectory). After calculation, the unit vector characteristics (the sum of the squares of the components is close to 1) are also verified. If the first derivative is 0 (the trajectory is horizontal), the first component of the normal unit vector is 1 and the second component is 0 (horizontally to the right). Specifically, the denominator is the same as that of the tangential unit vector. The first component of the normal unit vector is equal to 1 divided by the denominator, and the second component is equal to -the first derivative divided by the denominator.

[0154] The transverse component of the normal unit vector of the stable locking strip is the first component of the normal unit vector (corresponding to the transverse coordinate direction), has no unit, and reflects the transverse direction of the guard sweeping offset (positive for right and negative for left). It is directly extracted from the normal unit vector without additional calculation, and the sign needs to be recorded after extraction to determine the offset direction.

[0155] The neighborhood set of the stable locking strip is the area around the stable locking strip that needs to be swept, ensuring that the liquid drifting towards the locking strip is covered. The neighborhood half-width is 1.5 times the width of the strip, which is set according to the maximum range of liquid drifting towards the locking strip. The vertical range of the neighborhood set is consistent with the center trajectory (covering the full vertical length of the locking strip).

[0156] The normal drift amount of each position in the neighborhood set relative to the corresponding stable locking band is the drift distance of the liquid in the neighborhood along the normal unit vector direction, in meters, reflecting the range that needs to be covered by the guard sweep. Multiplying the absolute value of the transverse component ensures that the drift amount is positive (only the distance needs to be considered, not the direction). Specifically, the normal drift amount is equal to the transverse drift amount at the corresponding position multiplied by the absolute value of the transverse component of the normal unit vector at that position.

[0157] The guard sweep bias distance of the stable locking band at each vertical coordinate is the bias length of the guard sweep trajectory relative to the center trajectory, in meters, ensuring that the guard sweep trajectory can cover all the drifted liquid in the neighborhood. Specifically, within the neighborhood set, uniformly sample every 0.02 meters horizontally and every 0.05 meters vertically (the number of sampling points ≥ 20, covering the entire neighborhood), calculate the normal drift amount of each sampling point, and take the maximum value as the guard sweep bias distance at that vertical coordinate.

[0158] The guard sweep bias trajectory is the cleaning path extending along both sides of the center trajectory, in meters (coordinate units), with the two trajectories covering the drifted liquid in the positive and negative normal directions, respectively, ensuring that there is no liquid residue in the locking band and the surrounding area. When biasing, it is necessary to ensure that the vertical coordinates of each trajectory are consistent with the center trajectory, and the horizontal coordinates are the center trajectory horizontal coordinates ± guard sweep bias distance × normal unit vector transverse component (the sign of the component determines the direction).

[0159] The second derivative of the transverse position function of the stable locking band is a parameter that describes the rate of change of the first derivative, with no units, reflecting the change in the degree of curvature of the center trajectory (the larger the absolute value of the second derivative, the more dramatic the change in trajectory curvature). Specifically, take the first derivative values of two adjacent vertical coordinates, and the second derivative is equal to (the first derivative value of the upper vertical coordinate minus the first derivative value of the lower vertical coordinate) divided by the interval between the two vertical coordinates (consistent with the vertical interval of the first derivative, taking 0.05 meters); if the adjacent first derivative values are unchanged (the second derivative is close to 0), then the second derivative is taken as 0.

[0160] The curvature of the center trajectory of the stable locking band is a parameter that describes the degree of curvature of the center trajectory, with a unit of 1 / meter, and the larger the curvature, the more severe the bending of the trajectory (a smaller discrete step is required to ensure path adherence). Specifically, first calculate 1 plus the square of the first derivative, and take the square root of the sum to the third power (i.e., first calculate the square root of the sum, then square it and multiply it by the original sum) as the denominator; the numerator is the absolute value of the second derivative; the curvature is equal to the numerator divided by the denominator.

[0161] The maximum curvature of the stable locking center trajectory is the maximum value of the curvature of the center trajectory in the full vertical range, with a unit of 1 / m, which is used to calculate the cleaning discrete step length (the maximum curvature corresponds to the minimum step length). Specifically, in the vertical range of the center trajectory, take a curvature value every 0.1 m (cover the full trajectory), and select the maximum value from all curvature values as the maximum curvature; if all curvature values are ≤0.1 (1 / m, the trajectory is close to a straight line), the maximum curvature is 0.1.

[0162] The upper limit of the geometric approximation error is the maximum allowed deviation of the device cleaning path from the center trajectory, with a unit of meters, which ensures that the path fits the trajectory (no obvious deviation). The value range is 0.001 to 0.003 meters, and 0.002 meters is taken for conventional curtain cleaning. The error upper limit cannot exceed twice the device positioning accuracy (avoiding the inability to achieve it).

[0163] The discrete step length is the moving step length of the device along the center trajectory or the guard trajectory when cleaning, with a unit of meters. The smaller the step length, the more the path fits the trajectory, but the lower the efficiency. If the calculated square root is ≤0.005 meters (too small to affect efficiency), the discrete step length is 0.005 meters (the minimum safe step length); specifically, the discrete step length is not greater than the arithmetic square root of (2 times the geometric approximation error upper limit, divided by the maximum curvature).

[0164] It should be noted that the device can be a robot or other device.

[0165] The above describes the embodiments of the present embodiment, but the present embodiment is not limited to the specific embodiments described above, which are only illustrative and not limiting. Those skilled in the art can make many forms under the inspiration of the present embodiment, which are all within the protection of the present embodiment.

Claims

1. An adaptive path planning control system based on environmental feature perception, characterized in that, The method comprises the following steps: a data acquisition module acquires environmental characteristic parameters of the surface to be cleaned, the environmental characteristic parameters at least including temperature field parameters, near-wall flow field parameters and contact angle characteristic parameters; a stable locking zone extraction module identifies a stable locking zone on the surface to be cleaned based on the environmental characteristic parameters, and the stable locking zone extraction module comprises the following steps: calculating the Marangoni shear stress, including: the surface tension temperature coefficient multiplied by the temperature transverse gradient; measuring the width of the dewetting line; calculating the unit length retention force, including: the surface tension of the cleaning liquid multiplied by the contact line hysteresis difference; forming a transverse resultant force function, including: the sum of the Marangoni shear stress and the near-wall shear stress distribution multiplied by the width of the dewetting line, and then subtracting the unit length retention force; solving, at each fixed vertical coordinate, a transverse position function for which the transverse resultant force function is equal to zero and the partial derivative of the transverse resultant force function with respect to the transverse coordinate is less than zero; taking the transverse position function as the spatial trajectory of the stable locking zone; a segmentation module divides the surface to be cleaned into a plurality of zone units with the stable locking zone as the boundary; a cleaning sequence module determines the cleaning sequence of each zone unit according to the topological relationship between the stable locking zone and the zone units, and the cleaning sequence module comprises the following steps: for any two adjacent zone units, denoted as a first zone unit and a second zone unit; determining the stable locking zone corresponding to the common boundary of the first zone unit and the second zone unit; denoting the two sides of the common boundary as a first side and a second side, respectively, and taking the position on the first side as the transverse position function of the stable locking zone minus a very small displacement, and taking the position on the second side as the transverse position function of the stable locking zone plus a very small displacement; wherein the very small displacement is taken as a fixed fraction of the infrared image pixel spacing; calculating the unit length transverse resultant force of the first side position and the second side position, respectively; taking the closed interval of the intersection of the vertical projections of the first zone unit and the second zone unit as an effective vertical interval; calculating a directionality measure, including: integrating the difference between the unit length transverse resultant force of the first side and the unit length transverse resultant force of the second side over the effective vertical interval; when the directionality measure is greater than zero, generating a directed edge from the first zone unit to the second zone unit; when the directionality measure is less than zero, generating a directed edge from the second zone unit to the first zone unit; constructing a directed graph, the vertices of the directed graph being the zone units in the zone unit set, and the edges of the directed graph being the generated directed edges; defining a level value for each zone unit, the level value being the maximum value of the number of edges in all directed chains ending with the zone unit; determining the cleaning sequence of each zone unit in order of the level value from small to large; for a plurality of zone units with the same level value, taking any one of them as a target zone unit, calculating the minimum normal distance from the target zone unit to any stable locking zone, the minimum normal distance being the minimum value of the absolute values of the differences between the transverse coordinates of the points in the target zone unit and the transverse position function of the stable locking zone at the corresponding vertical coordinates; determining the cleaning sequence of the zone units with the same level value in order of the minimum normal distance from large to small; a cleaning direction module determines the cleaning direction in each zone unit based on the temperature field parameters of the surface to be cleaned. The cleaning path module determines the cleaning path in each zone unit based on the environmental characteristic parameters of the surface to be cleaned. The cleaning module performs cleaning operations on the stable locking zone.

2. The environmental feature awareness based adaptive path planning control system of claim 1, wherein, Obtain the environmental characteristic parameters of the surface to be cleaned, including at least temperature field parameters, near-wall flow field parameters, and contact angle characteristic parameters, including: A coordinate system is established, in which the horizontal coordinate is the horizontal distance from the vertical edge of the facade, and the vertical coordinate is the distance along the gravity direction; Obtain the infrared image of the surface to be cleaned, and obtain the temperature field from the infrared image, which is the distribution of temperature on the plane composed of the horizontal coordinate and the vertical coordinate; The partial derivative of the temperature field along the horizontal coordinate is taken as the temperature horizontal gradient; Measure the incoming flow speed; Calculate the corner near-wall wind speed distribution based on the incoming flow speed, including: the incoming flow speed multiplied by a preset corner release amount coefficient, then multiplied by 1 plus the negative attenuation exponential power of the sum of the horizontal coordinate and the preset attenuation distance; Calculate the near-wall shear stress distribution based on the corner near-wall wind speed distribution, including: 0.5 times the air density, multiplied by the preset near-wall friction coefficient, then multiplied by the square of the corner near-wall wind speed distribution; Measure the advancing contact angle and the receding contact angle; Calculate the contact line hysteresis based on the advancing contact angle and the receding contact angle, including: the cosine of the receding contact angle minus the cosine of the advancing contact angle; Take the temperature field and the temperature horizontal gradient as the temperature field parameters; Take the corner near-wall wind speed distribution and the near-wall shear stress distribution as the near-wall flow field parameters; Take the advancing contact angle, the receding contact angle, and the contact line hysteresis as the contact angle characteristic parameters.

3. The environmental feature awareness based adaptive path planning control system of claim 2, wherein, Divide the surface to be cleaned into multiple zone units with the stable locking zone as the boundary, including: Determine the number of stable locking zones; Sort the horizontal position functions of each stable locking zone in ascending order at each fixed vertical coordinate to obtain the sorted stable locking zone horizontal position functions; Determine the area of the surface to be cleaned; Define the geometric set of each stable locking zone as the set of points in the surface to be cleaned area whose horizontal coordinates are equal to the corresponding sorted stable locking zone horizontal position functions; Divide the zone unit set, including: The first zone unit is the set of points in the surface to be cleaned area whose horizontal coordinates are less than the horizontal position function of the first sorted stable locking zone; The middle zone unit is the set of points in the surface to be cleaned area whose horizontal coordinates are between the horizontal position functions of adjacent sorted stable locking zones; The last zone unit is the set of points in the surface to be cleaned area whose horizontal coordinates are greater than the horizontal position function of the last sorted stable locking zone; The surface to be cleaned area is the union of all zone units and the geometric set of all stable locking zones, and each zone unit is mutually exclusive.

4. The environmental feature awareness based adaptive path planning control system of claim 3, wherein, Determine the cleaning direction in each zone unit based on the temperature field parameters of the surface to be cleaned, including: For each position in the temperature field, calculate the temperature gradient vector, the first component of which is the partial derivative of the temperature field along the horizontal coordinate, and the second component of which is the partial derivative of the temperature field along the vertical coordinate; Calculate the temperature gradient modulus of each position, which is the modulus of the temperature gradient vector at the corresponding position; calculating an isotherm normal unit vector of each position, including: dividing a temperature gradient vector of the corresponding position by a temperature gradient module of the position; calculating an isotherm tangential unit vector of each position, a first component of the isotherm tangential unit vector being a negative value of a second component of the isotherm normal unit vector, and a second component of the isotherm tangential unit vector being a first component of the isotherm normal unit vector; for each zone unit, taking the isotherm tangential unit vectors of the positions in the zone unit as the cleaning directions of the corresponding positions in the zone unit, to form a cleaning direction field in each zone unit.

5. The environmental feature awareness based adaptive path planning control system of claim 4, wherein, determining a cleaning row distance in each zone unit based on the environmental characteristic parameters of the surface to be cleaned, including: obtaining a dynamic viscosity of the cleaning liquid, an effective width of the cleaning head, a downlink speed of the cleaning head, and a single downlink stroke height; calculating a total shear stress in the transverse direction of each position, including: adding a Marangoni shear stress of the corresponding position to a near-wall shear stress of the position; calculating a transverse velocity of each position, including: multiplying a preset empirical coefficient by the total shear stress in the transverse direction, multiplying the result by the strip thickness, and dividing the result by the dynamic viscosity of the cleaning liquid; calculating a single downlink stroke time, including: dividing the single downlink stroke height by the downlink speed of the cleaning head; calculating a transverse drift of each position, including: multiplying the transverse velocity of the corresponding position by the single downlink stroke time; for each zone unit, determining a maximum transverse drift in the zone unit, the maximum transverse drift being the maximum value among the transverse drifts of all positions in the zone unit; determining a cleaning row distance for each zone unit, the cleaning row distance being not greater than the effective width of the cleaning head minus the maximum transverse drift of the corresponding zone unit.

6. The environmental feature awareness based adaptive path planning control system of claim 5, wherein, performing a cleaning operation on the stable locking bands, including: defining a center trajectory of each stable locking band, the center trajectory being a trajectory with a transverse coordinate equal to a transverse position function of the corresponding stable locking band and a vertical coordinate changing synchronously with the transverse position function; calculating a first derivative of the transverse position function of each stable locking band; calculating a tangential unit vector of each stable locking band, including: first calculating a sum value of 1 plus the square of the first derivative of the transverse position function of the stable locking band, taking a square root of the sum value as a first denominator, taking a first component of the tangential unit vector as the first derivative of the transverse position function of the stable locking band, and taking a second component as 1, and then dividing the two components by the first denominator respectively; calculating a normal unit vector of each stable locking band, including: taking a first component of the normal unit vector as 1, and taking a second component as a negative value of the first derivative of the transverse position function of the stable locking band, and then dividing the first component and the second component by the first denominator respectively; extracting a transverse component of the normal unit vector of each stable locking band; defining a neighborhood set of each stable locking band, the neighborhood set being: a set of all points in the surface to be cleaned whose absolute value of the difference between the transverse coordinate and the transverse position function of the corresponding stable locking band is not greater than a neighborhood half-width; calculating a normal drift of the corresponding stable locking band for each position in the neighborhood set, including: multiplying the transverse drift of the corresponding position by the absolute value of the transverse component of the normal unit vector of the stable locking band at the position; determining a guard offset distance of each stable locking belt at each vertical coordinate, the guard offset distance being a maximum value of normal drifts of all positions in a neighborhood set of the corresponding stable locking belt; generating two guard offset trajectories of each stable locking belt, including: taking the center trajectory of the corresponding stable locking belt as a reference, offsetting in the positive direction and the negative direction of the normal unit vector respectively by the guard offset distance at the corresponding vertical coordinate to obtain the two guard offset trajectories; calculating the second derivative of the lateral position function of each stable locking belt; calculating the curvature of the center trajectory of each stable locking belt, including: first calculating a sum value of 1 plus the square of the first derivative of the stable locking belt lateral position function, taking the square root of three of the sum value as a second denominator, and then using the absolute value of the second derivative of the stable locking belt lateral position function as a numerator, and dividing the numerator by the second denominator to obtain the curvature of the corresponding position; determining the maximum curvature of the center trajectory of each stable locking belt, and setting an upper limit of the geometric approximation error; calculating the discrete step length of each stable locking belt during cleaning, the calculation process being: 2 times the upper limit of the geometric approximation error, divided by the maximum curvature of the corresponding stable locking belt, taking the square root of the result, and the discrete step length during cleaning being not greater than the square root.

Images