Intelligent self-adaptive cleaning system and method for power battery components
Patent Information
- Application Number
- CN202611165198.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-03
- Publication Date
- 2026-09-01
AI Technical Summary
[0003]本申请提供了一种动力电池构件的智能自适应清洗系统及方法，主要用于解决现有清洗设备在面对复杂深盲孔时存在水压盲区，以及恒定高压射流易损伤薄壁基体的技术问题
Smart Images
Figure CN122665804A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of cleaning equipment manufacturing technology, and more specifically, to an intelligent adaptive cleaning system and method for power battery components, which is particularly suitable for adaptive cleaning operations of complex internal cavities in B08B spray cleaning main category and power battery casing cleaning equipment. Background Technology
[0002] In the manufacturing of power battery components, metal debris and cutting fluid can easily remain in the complex deep blind holes on the surface and inside the components. Traditional cleaning equipment generally uses constant pressure blind spraying or preset trajectory robotic arms, which are difficult to adapt to clamping errors or manufacturing tolerances, and pose risks of collision and defects in water pressure control. When dealing with deep blind holes, fluid with fixed water pressure often cannot penetrate to the bottom dead corner due to air resistance inside the hole, forming a water pressure blind zone; if the overall water supply pressure is increased, the constant high-pressure jet sweeping across the thin-walled area can easily cause irreversible deformation or damage. Summary of the Invention
[0003] This application provides an intelligent adaptive cleaning system and method for power battery components, which is mainly used to solve the technical problems of water pressure blind zone when existing cleaning equipment faces complex deep blind holes, and the easy damage of thin-walled substrate by constant high-pressure jet.
[0004] To achieve the above objectives, this application adopts the following technical solution: an intelligent adaptive cleaning system for power battery components, comprising: a frame; a six-axis industrial robotic arm mounted on the frame; a fluid supply system including a main water supply pipeline and an electro-hydraulic proportional servo valve disposed on the main water supply pipeline; a piezoelectric variable diameter spray head assembly installed at the end of the six-axis industrial robotic arm, internally comprising a flexible silicone throat and an annular piezoelectric ceramic actuator surrounding the outer wall; a three-dimensional structured light scanner mounted on the six-axis industrial robotic arm; a high-frequency pressure sensor disposed in the fluid supply system; and a main control unit, respectively connected to the six-axis industrial robotic arm, the electro-hydraulic proportional servo valve, and the annular piezoelectric ceramic actuator. The annular piezoelectric ceramic actuator, the three-dimensional structured light scanner, and the high-frequency pressure sensor are electrically connected. The main control unit is configured to determine the surface shape of the cleaning area based on the point cloud acquired by the three-dimensional structured light scanner, and output a pulse width modulation voltage to drive the annular piezoelectric ceramic actuator to squeeze the flexible silicone throat and change the jet cross-section. The feature vector of the hole feature surface in the point cloud is extracted to guide the six-axis industrial robot arm to coaxially feed and center. The back pressure change rate is obtained by time differentiation calculation of the back pressure acquired by the high-frequency pressure sensor. When the back pressure change rate is greater than the first warning threshold, the opening of the electro-hydraulic proportional servo valve is reduced and a retraction command is sent to the six-axis industrial robot arm.
[0005] Specifically, the main control unit is configured to: disconnect the power supply to the annular piezoelectric ceramic actuator in response to the cleaning area surface being identified as a flat thin-walled surface, so that the flexible silicone throat is in an expanded state, and reduce the opening of the electro-hydraulic proportional servo valve to spray a low-pressure umbrella-shaped water flow; and energize the annular piezoelectric ceramic actuator in response to the cleaning area surface being identified as a deep blind hole, so that the flexible silicone throat contracts to form a high-pressure water needle jet.
[0006] Specifically, the main control unit is configured to extract the feature vector of the hole feature surface when the surface shape of the cleaning area is identified as a deep blind hole. This includes: identifying the local neighborhood point set of the deep blind hole through the point cloud; solving the covariance matrix of the local neighborhood point set; performing principal component analysis on the covariance matrix to extract the minimum eigenvalue; using the eigenvector corresponding to the minimum eigenvalue as the normal vector reference; and, in conjunction with the inverse kinematics model of the six-axis industrial robot, planning a coaxial interpolation path based on the normal vector reference to control the six-axis industrial robot to carry the piezoelectric variable diameter spray head assembly deep into the deep blind hole to perform coaxial interpolation operations.
[0007] Specifically, the main control unit is equipped with a proportional-integral-derivative (PID) controller. The main control unit is configured to continuously monitor the sudden change trend of the back pressure change rate after performing time derivative calculation on the back pressure. In response to the sudden increase in the back pressure change rate exceeding the first warning threshold, the underlying limiting logic of the PID controller is triggered to reduce the opening of the electro-hydraulic proportional servo valve.
[0008] Specifically, the step of sending a retraction command to the six-axis industrial robot arm is configured as follows: while triggering the underlying limiting logic of the proportional-integral-derivative controller, a backpressure differential linkage interlock control closed loop is established; based on the backpressure differential linkage interlock control closed loop, the current downward command of the six-axis industrial robot arm is interrupted; based on the previously extracted feature vector, a retraction trajectory is generated along the opposite direction of the normal vector of the hole feature surface; and the six-axis industrial robot arm is controlled to perform a reverse retraction operation according to the retraction trajectory.
[0009] This application also provides an intelligent adaptive cleaning method for power battery components, applied to the aforementioned intelligent adaptive cleaning system for power battery components. The method includes: acquiring a point cloud obtained by scanning the power battery component using a three-dimensional structured light scanner; determining the surface shape of the cleaning area based on the point cloud; in response to the surface shape of the cleaning area being identified as a deep blind hole, extracting the minimum eigenvector of the covariance matrix of the hole feature surface in the point cloud to guide the coaxial feed and centering of a six-axis industrial robotic arm; outputting a pulse width modulation voltage to a ring piezoelectric ceramic actuator to induce radial strain, thereby squeezing a flexible silicone throat to form a water needle jet; during the cleaning of the deep blind hole, acquiring the back pressure of the fluid inside the hole collected by a high-frequency pressure sensor; performing a differential calculation on the back pressure in the time dimension to obtain the back pressure change rate; in response to the back pressure change rate being greater than a first warning threshold, triggering an attenuation signal to reduce the opening of the electro-hydraulic proportional servo valve, and sending a retraction command to the six-axis industrial robotic arm in the opposite direction of the normal vector.
[0010] Specifically, after determining the surface shape of the cleaning area based on the point cloud, the method further includes: in response to recognizing that the surface shape of the cleaning area is a flat thin-walled surface, disconnecting the pulse width modulation voltage output to the annular piezoelectric ceramic actuator, so that the flexible silicone throat returns to its original inner diameter size; and simultaneously outputting a pressure reduction command to reduce the opening of the electro-hydraulic proportional servo valve, and outputting an umbrella-shaped water flow.
[0011] Specifically, the step of performing a time-dimensional differential calculation on the recoil back pressure to obtain the back pressure change rate includes: extracting back pressure acquisition values within two consecutive high-frequency sampling periods on the time axis; calculating the difference between the back pressure acquisition values within the two consecutive high-frequency sampling periods; dividing the difference between the back pressure acquisition values by the time span of the high-frequency sampling period to obtain a single differential result; and performing a moving average filter on the single differential result within multiple consecutive periods to obtain the smoothed back pressure change rate.
[0012] Specifically, the step of triggering the attenuation signal to reduce the opening of the electro-hydraulic proportional servo valve and sending a retraction command to the six-axis industrial robot in the opposite direction of the normal vector includes: setting multi-level warning judgment conditions, the multi-level warning judgment conditions including a first warning threshold and a second warning threshold greater than the first warning threshold; in response to the back pressure change rate being greater than the first warning threshold and less than the second warning threshold, triggering only the attenuation signal to reduce the opening of the electro-hydraulic proportional servo valve; in response to the back pressure change rate being greater than the second warning threshold, reducing the control opening of the electro-hydraulic proportional servo valve to zero and calling the inverse kinematics model to generate a normal reverse displacement matrix; and driving the six-axis industrial robot to perform a rigid retraction action according to the normal reverse displacement matrix.
[0013] This application also provides a computer device, including: a memory for storing a computer program; and a processor electrically connected to the memory for executing the computer program to implement the above-described intelligent adaptive cleaning method for power battery components. Attached Figure Description
[0014] Figure 1 This is a schematic diagram of the overall structure of the intelligent adaptive cleaning system for power battery components provided in the embodiments of this application. Figure 2 This is a cross-sectional view of the piezoelectric variable diameter spray head assembly provided in the embodiments of this application. Figure 3 This is a block diagram of the logical structure of the central control unit provided in the embodiments of this application. Figure 4 This is a flowchart of the intelligent adaptive cleaning method for power battery components provided in the embodiments of this application. Figure 5 This is a partially enlarged view of the piezoelectric actuated variable diameter section provided in the embodiments of this application.
[0015] Explanation of reference numerals in the attached figures
[0016] In the diagram: 101-Frame, 102-Six-axis industrial robotic arm, 103-Conveying and positioning platform, 104-Power battery component, 105-Storage tank, 106-High-frequency water pump, 107-Main water supply pipeline, 108-Electro-hydraulic proportional servo valve, 109-High-pressure flexible hose, 110-Piezoelectric variable diameter spray head assembly, 111-3D structured light scanner, 112-High-frequency pressure sensor, 113-Multi-stage filtration device, 114-Six-dimensional torque sensor, 115-Central main control unit, 201-Stainless steel protective shell, 202-Flow stabilizing chamber, 203-Piezoelectric actuation variable diameter section, 204-Nozzle outlet, 301-Point cloud processing module, 302-Motion control module, 303-Fluid control module, 304-Back pressure monitoring module, 501-Flexible silicone throat, 502-Annular piezoelectric ceramic actuator. Detailed Implementation
[0017] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.
[0018] like Figure 1As shown, this application embodiment provides an intelligent adaptive cleaning system for power battery components. The system encompasses multiple physical execution stages, including multi-axis linkage mechanical execution, multi-sensor perception, and fluid power supply. At the system's underlying architecture, a frame 101 is stably mounted on a ground workbench, and a six-axis industrial robotic arm 102 providing multi-degree-of-freedom spatial movement is mounted on the frame 101. A conveying and positioning platform 103 is configured below the frame 101, equipped with positioning pins and hydraulic clamps to support and rigidly hold the power battery components 104 to be cleaned. To perform the cleaning action, a piezoelectric variable-diameter spray head assembly 110 is installed at the end flange of the six-axis industrial robotic arm 102 via a mechanical quick-change device with anti-detachment pin holes. This configuration of the mechanical support base ensures the consistency of the workpiece's spatial position during batch cleaning operations, while limiting the nozzle from loosening or shifting due to fluid recoil.
[0019] like Figure 1 As shown, the cleaning system is equipped with a fluid supply system to provide power for the washing action. This fluid supply system includes a storage tank 105 for storing pure cleaning fluid, a high-frequency water pump 106 connected to the storage tank 105, and a main water supply pipeline 107 responsible for pipeline transmission. A core flow and pressure regulating element, namely an electro-hydraulic proportional servo valve 108, is connected in series on the main water supply pipeline 107. The main water supply pipeline 107 is made of high-strength stainless steel seamless pipe with a metal braided shielding layer, and its end extends to the end of the robotic arm via a high-pressure flexible hose 109, forming a physical connection with the piezoelectric variable diameter spray head assembly 110. Simultaneously, to prevent damage to the precision valve core from small particles, a multi-stage filtration device 113 is also connected in series on the main water supply pipeline 107. The multi-stage filtration device 113 is installed between the high-frequency water pump 106 and the electro-hydraulic proportional servo valve 108, intercepting impurities through an internal stainless steel coarse filter and a micron-level fine filter element. By combining these fluid supply systems, a high-purity, dynamically adjustable pressure source power can be continuously provided to the downstream spray components.
[0020] To endow the aforementioned mechanical and fluid structures with intelligent environmental perception capabilities, the system also integrates a multi-sensor system and a processing center. For example... Figure 1As shown, a 3D structured light scanner 111 with a top-down view of the working surface is installed above the frame 101. This scanner uses infrared laser to project striped structured light and combines it with binocular vision to acquire high-precision spatial topography point clouds of the power battery component 104 in a non-contact manner. Simultaneously, a high-frequency pressure sensor 112 with a sampling frequency reaching the 10 kilohertz level is installed in the area of the main water supply pipeline 107 near the piezoelectric variable diameter spray head assembly 110. For physical wrist impact protection, a six-dimensional torque sensor 114 is also installed between the six-axis industrial robotic arm 102 and the piezoelectric variable diameter spray head assembly 110. All sensors and actuators in the system are electrically connected to the central control unit 115 via the industrial Ethernet bus protocol. It should be noted that, to ensure closed-loop system operation, the system base is also equipped with a system power supply module, a servo driver for driving the motor, a pulse-width modulation voltage drive amplifier circuit for generating high-frequency control signals, and a cleaning waste liquid collection and recovery module at the bottom.
[0021] like Figure 2 and Figure 5 As shown in the partial cross-sectional view and enlarged view, this embodiment features a specialized design for the internal microscopic physical structure of the piezoelectric variable diameter spray head assembly 110. The piezoelectric variable diameter spray head assembly 110 is externally equipped with a stainless steel protective shell 201 providing physical impact resistance. Inside the stainless steel protective shell 201, a flow stabilizing chamber 202, a piezoelectric actuated variable diameter section 203, and a nozzle outlet 204 for jet constraint and guidance are sequentially arranged along the fluid flow path. Specifically, the piezoelectric actuated variable diameter section 203 abandons the traditional mechanical needle valve; its core includes an internal highly elastic flexible silicone throat 501 and multiple sets of annular piezoelectric ceramic actuators 502 tightly surrounding its outer wall. The central control unit 115 is configured to drive the annular piezoelectric ceramic actuators 502 by continuously adjusting the output pulse-width modulation voltage. Based on the inverse piezoelectric effect physical characteristics, the annular piezoelectric ceramic actuators 502 generate radial strain along their material polarization direction when receiving voltage excitation. By applying radial strain to compress the flexible silicone throat 501 inward, the silicone material can be forced to elastically contract. Through this physical coupling between the flexible material and the piezoelectric ceramic, the system can steplessly change the flow cross-sectional area within a millisecond response time, achieving instantaneous switching of the jet morphology.
[0022] In one alternative implementation, combined with Figure 3The logical structure diagram shown indicates that the central control unit 115 integrates a point cloud processing module 301, a motion control module 302, a fluid control module 303, and a back pressure monitoring module 304. During cleaning operations, the control unit performs differentiated fluid drive by identifying the surface shape of the area. Specifically, when the point cloud processing module 301 analyzes the point cloud data collected by the 3D structured light scanner 111 and determines that the surface shape of the current cleaning area is identified as a flat, thin-walled surface, the fluid control module 303 rapidly reduces the opening of the electro-hydraulic proportional servo valve 108 to lower the basic water supply pressure in the pipeline and simultaneously disconnects the pulse width modulation voltage supplied to the annular piezoelectric ceramic actuator 502. In this zero-level state, the flexible silicone throat 501 fully expands due to its own tension, causing the water flow to form a low-pressure, umbrella-shaped water jet through the large-diameter nozzle outlet 204. This mode effectively reduces the concentrated impact force of the water hammer effect and prevents physical impact damage to the rigid areas of the thin-walled component.
[0023] Conversely, in response to the surface shape of the cleaning area being identified as a deep blind hole, the fluid control module 303 sends a current command to increase the opening of the electro-hydraulic proportional servo valve 108 to increase the overall pipeline water pressure. Simultaneously, a high duty cycle energizing voltage is applied to the annular piezoelectric ceramic actuator 502. This voltage causes the piezoelectric ceramic to contract inward, compressing the flexible silicone throat 501. Through the physical compression effect of a large flow rate of water combined with the micro-aperture, the system outputs a high-pressure micro-jet water jet from the nozzle outlet 204. This water jet has the ability to break through the hydrodynamic barriers of the air resistance dead zones within the deep hole.
[0024] In the aforementioned deep blind hole alignment operation, to prevent the high-pressure water jet from slipping or cutting the internal hole wall, the central control unit 115 executes strict spatial normal pair extraction logic during the point cloud processing stage. Specifically, the point cloud processing module 301 first identifies the local neighborhood point set belonging to the target deep blind hole through the discrete point cloud. Then, to eliminate the fitting bias caused by local surface roughness noise, it constructs and solves the covariance matrix of this local neighborhood point set, based on the following fundamental mathematical expression:
[0025]
[0026] Where C represents the covariance matrix data obtained from solving the target hole region, which is a third-order symmetric matrix; N represents the total number of three-dimensional spatial points contained in the local neighborhood point set, and its value usually depends on the scanner resolution; P i Represents the three-dimensional coordinate vector of the i-th spatial point in the local neighborhood point set; This represents the three-dimensional coordinate vector of the centroid of all local neighborhood point sets. The main control unit performs principal component analysis on this covariance matrix to extract the eigenvector corresponding to the smallest eigenvalue. The system uses the eigenvector corresponding to the smallest eigenvalue as the normal vector reference close to the physical geometric axis. The motion control module 302, combined with the inverse kinematics model of the six-axis industrial robot arm 102, generates a homogeneous transformation displacement compensation matrix based on this normal vector reference and plans a smooth coaxial interpolation path. By controlling the six-axis industrial robot arm 102 to move along this interpolation path, it can ensure that the central axis of the piezoelectric variable diameter spray head assembly 110 and the axis of the deep blind hole to be cleaned achieve a strict coaxial feed alignment operation, and constrain the attitude yaw angle and pitch angle error when cutting into the hole interval within the safe tolerance range.
[0027] To address the transient water pressure damage to the pipe walls caused by the accumulation of debris and residual fluid during deep blind hole cleaning, this application further designs a deep hole anti-backpressure protection interlock mechanism based on the above embodiments. During deep blind hole cleaning, the backpressure monitoring module 304 utilizes a high-frequency pressure sensor 112 with a sampling rate at the 10 kHz level to acquire real-time backpressure data transmitted to the pipeline after the fluid inside the hole is obstructed and rebounds. To accurately detect early signs of blockage, the main control unit performs rigorous time-dimension differential calculations on this backpressure. The specific formula for obtaining the backpressure change rate is:
[0028]
[0029] Among them, R p This represents the calculated rate of change of recoil back pressure. This represents the back pressure difference extracted within the current high-frequency sampling period. Its physical quantity is at the kilopascal level, and its single-step difference fluctuation range can be between 1 kPa and 5 kPa. The time step of the differential sampling of the system is represented by a step size of 0.0001 seconds, which can be taken as small as 0.0001 seconds given the 10 Hz characteristics of the high-frequency sensor. After continuously obtaining the back pressure change rate, the system will continuously monitor the sudden change trend of the back pressure change rate. Since the main control unit is equipped with a proportional-integral-derivative controller, when the system determines that the back pressure change rate is greater than the internally preset first warning threshold, it indicates that the flow channel at the bottom of the deep blind hole is acutely blocked by chips. At this time, the underlying limiting logic of the proportional-integral-derivative controller will be forcibly triggered, and a forced attenuation signal will be output to instantly reduce the opening of the electro-hydraulic proportional servo valve 108, cutting off the overload of the water supply pipeline. At the same time as sending the signal to reduce the valve opening attenuation, the system establishes a back pressure differential linkage interlock control closed loop, forcibly interrupting the previously uncompleted downward movement command of the six-axis industrial robot arm 102. The motion control module 302 immediately generates a retraction trajectory based on the aforementioned extracted normal feature vector, in the opposite direction of the normal vector of the hole feature surface. The six-axis industrial robotic arm 102 is driven to perform a rigid retraction action according to the reverse retraction trajectory. Through differential prediction using a very short time window before the occurrence of fluid dynamic hysteresis and software / hardware retraction interlocking, the system can proactively retract and cut off the jet source the instant high-pressure stall occurs within the hole, improving the success rate of overpressure flushing protection. If extreme dimensional deviations cause direct collision between the nozzle and the wall, the excessive force measured by the aforementioned six-dimensional torque sensor 114 will unconditionally trigger a mechanical emergency stop, forming a supplementary fallback defense.
[0030] One embodiment of this application also provides an intelligent adaptive cleaning method for power battery components. This method operates on the aforementioned base equipped with multi-source sensors and a piezoelectric variable-diameter component. Figure 4As shown in the flowchart, this intelligent adaptive cleaning method specifically covers the following core operational steps: Step S401: Obtain a high-precision 3D point cloud by scanning the power battery component placed on the worktable using a 3D structured light scanner. Using a low-level point cloud classification algorithm, determine whether the surface of the current cleaning area is a flat, thin-walled surface or a deep blind hole with a high aspect ratio based on the point cloud data. Step S402: In response to the cleaning area surface being identified as a deep blind hole, the system enters a fine alignment sequence. Specifically, the system extracts the minimum eigenvector of the covariance matrix of the hole feature surface in the point cloud as the approaching normal vector axis to guide the six-axis industrial robotic arm to achieve precise coaxial feed and centering operations. Step S403: During the centering process, a calculated pulse-width modulated voltage is output to the annular piezoelectric ceramic actuator on the outer wall, causing it to undergo polarized radial strain. This physical pressure significantly compresses the internal flexible silicone throat, forcibly concentrating the water flow and forming a penetrating water needle jet at the nozzle outlet that reaches the bottom of the hole. In step S404, during continuous high-pressure deep blind hole cleaning, to prevent potential risks of air resistance or debris blockage within the hole, the system initiates a parallel environmental resistance monitoring process. It acquires the backpressure signal from the reverse transmission of obstructed fluid within the hole, collected by a high-frequency pressure sensor on the main pipeline; performs differential calculation on the backpressure sequence over time to obtain the backpressure change rate reflecting the abrupt slope. A threshold crossing judgment is performed on the calculated change rate. If the backpressure change rate exceeds the system's first warning threshold, the safety logic determines that the deep hole is at risk of bursting. The main controller immediately triggers an emergency stop attenuation signal via the communication bus to reduce or even cut off the opening of the electro-hydraulic proportional servo valve, thereby blocking pressurization at its source; and simultaneously sends a retraction command to the six-axis industrial robotic arm, rapidly withdrawing the tool in the opposite direction to the original positioning normal vector.
[0031] In another embodiment, a more in-depth divergence and expansion of the specific control branch of the above method is provided. After the surface shape recognition and determination in step S401, a flat surface processing branch is also included in parallel: in response to the recognition that the surface shape of the cleaning area is a large-area continuous flat thin-walled surface, the system actively disconnects the high-frequency pulse width modulation voltage output to the annular piezoelectric ceramic actuator. This operation causes the flexible silicone throat to lose external stress pressure and quickly recover to its original unfolded inner diameter size by relying on the material memory properties. During this period, the main controller synchronously outputs a pressure reduction control command to the electro-hydraulic proportional servo valve to reduce the mechanical opening, outputting a large-area, low-impact umbrella-shaped water flow at the nozzle tip, effectively adapting to the rapid and gentle cleaning requirements of a large, weakly rigid flat surface.
[0032] In another embodiment, the control method for generating strain by energizing the piezoelectric ceramic component is explained in detail. The step of outputting a pulse-width modulation voltage to form a water jet in step S403 specifically employs dynamic nonlinear tracking logic, including: during the continuous compression of the flexible silicone throat by the annular piezoelectric ceramic actuator, the current voltage duty cycle parameter of the annular piezoelectric ceramic actuator is monitored in real time in the background; the duty cycle value of the pulse-width modulation voltage is dynamically adjusted to compensate for the hysteresis effect, ensuring that the actual deformation of the flexible silicone throat and the duty cycle value always have a nonlinear correspondence; in response to the duty cycle value steadily climbing to reach the internally preset target peak value, the actual flow cross-sectional area within the flexible silicone throat is forcibly compressed to a preset minimum proportion of the original unfolded cross-sectional area, thereby forming a micro-water jet flow state with strong peeling ability.
[0033] In another embodiment, the calculation method for backpressure is robustly extended. In step S404, which involves performing time-dimensional differential calculation on the backpressure to obtain the rate of change of backpressure, a filtering and smoothing operation is employed to eliminate false alarm spikes caused by high-frequency oscillations in the water flow and sensor noise. This includes: extracting the raw backpressure values from two consecutive high-frequency sampling periods on the time axis; calculating the difference between the backpressure values from these two periods; directly dividing the difference by the time span of the high-frequency sampling period to obtain a single differential result; and instead of using this single result directly, performing a moving average filtering process on the single differential results from multiple continuously stored historical periods to obtain the true rate of change of backpressure after removing high-frequency noise contamination and smoothing the curve.
[0034] In another embodiment, a more flexible and tiered protection logic setting is provided for the decompression and tool retraction process after back pressure exceeds the limit. This trigger-attenuation tool retraction step specifically includes: setting multi-level warning judgment ladder conditions within the control system. This condition system includes the aforementioned first warning threshold and a second warning threshold whose value is significantly greater than the first warning threshold. During operation, if the smoothed back pressure change rate is greater than the first warning threshold but still less than the more dangerous second warning threshold, it is determined to be a moderate chip removal problem. At this time, only a unilateral attenuation signal is triggered to reduce the opening of the electro-hydraulic proportional servo valve and lower the water supply pressure, without interrupting the deep-reaching action of the robotic arm. If the back pressure change rate rises sharply to exceed the second warning threshold, it is determined to be a complete blockage danger. The main control will immediately intervene, returning the control opening of the electro-hydraulic proportional servo valve to zero, and within the same cycle, calling the homogeneous inverse kinematics model to generate a normal reverse displacement matrix. By directly driving the underlying servo system of the six-axis industrial robotic arm, the arm body is controlled to perform a rigid reverse tool retraction action according to the normal reverse displacement matrix.
[0035] This application also provides a computer device for carrying out the actual digital operation of the aforementioned control logic in a physical machine. This computer device exists as the core host computer or industrial controller within the system base. Its internal hardware backbone includes a processor and a memory electrically connected to the processor via a high-speed internal bus. This memory is allocated as a non-volatile storage pool for computer-readable instructions, used to load and store compiled computer programs. When the device is powered on and put into operation, the central processing unit is invoked to read and cyclically execute the instruction set in the computer program, driving the physical electromechanical ports at various external levels. This enables the entire platform to automatically implement and execute the intelligent adaptive cleaning method for the power battery components and the corresponding various logical judgments and interlocking tool retraction control actions detailed in the foregoing embodiments.
[0036] This application provides an intelligent adaptive cleaning system and method for power battery components. By combining spatial normal alignment technology extracted from three-dimensional structured light principal components with fluid control using pulse-width voltage to drive piezoelectric ceramic radial extrusion throats, a highly dynamic and smooth adjustment of the cleaning water column morphology is achieved. This structure, combined with a back pressure differential monitoring anti-impact mechanism, solves the water pressure blind zone problem caused by air resistance in the complex flow channels and deep holes of new energy battery trays. Based on a control mechanism that links multi-level back pressure thresholds with servo valves, the system can accurately control the water pressure output limits and is supplemented by reverse tool retraction protection. This limits the irreversible physical deformation and stamping damage of constant high-pressure jets to thin-walled and fragile porous substrates, improving the adaptive fault tolerance rate of intelligent cleaning operations under multi-batch tolerances and the yield rate of the final product.
Claims
1. An intelligent adaptive cleaning system for power battery components, characterized in that, include: frame; A six-axis industrial robotic arm is mounted on the frame; a fluid supply system includes a main water supply pipeline and an electro-hydraulic proportional servo valve installed on the main water supply pipeline; A piezoelectric variable-diameter spray head assembly is installed at the end of the six-axis industrial robotic arm, and internally includes a flexible silicone throat and an annular piezoelectric ceramic actuator surrounding the outer wall; a 3D structured light scanner is installed on the six-axis industrial robotic arm; a high-frequency pressure sensor is configured in the fluid supply system; a main control unit is electrically connected to the six-axis industrial robotic arm, the electro-hydraulic proportional servo valve, the annular piezoelectric ceramic actuator, the 3D structured light scanner, and the high-frequency pressure sensor; the main control unit is configured to determine the surface shape of the cleaning area based on the point cloud acquired by the 3D structured light scanner, output a pulse width modulation voltage to drive the annular piezoelectric ceramic actuator to squeeze the flexible silicone throat and change the jet cross-section; extract the feature vector of the hole feature surface in the point cloud to guide the six-axis industrial robotic arm to coaxially feed and center; perform time differential calculation on the back pressure acquired by the high-frequency pressure sensor to obtain the back pressure change rate, and when the back pressure change rate is greater than a first warning threshold, reduce the opening of the electro-hydraulic proportional servo valve and send a retraction command to the six-axis industrial robotic arm.
2. The intelligent adaptive cleaning system for power battery components as described in claim 1, characterized in that, The main control unit is configured to: in response to the cleaning area surface being identified as a flat thin-walled surface, disconnect the power supply to the annular piezoelectric ceramic actuator, causing the flexible silicone throat to be in an expanded state, and reduce the opening of the electro-hydraulic proportional servo valve to spray a low-pressure umbrella-shaped water flow; in response to the cleaning area surface being identified as a deep blind hole, energize the annular piezoelectric ceramic actuator, causing the flexible silicone throat to contract and form a high-pressure water needle jet.
3. The intelligent adaptive cleaning system for power battery components as described in claim 2, characterized in that, The main control unit is configured to extract the feature vector of the hole feature surface when the surface shape of the cleaning area is identified as a deep blind hole. Specifically, this includes: identifying the local neighborhood point set of the deep blind hole through the point cloud; solving the covariance matrix of the local neighborhood point set; performing principal component analysis on the covariance matrix to extract the minimum eigenvalue; using the eigenvector corresponding to the minimum eigenvalue as the normal vector reference; and, in conjunction with the inverse kinematics model of the six-axis industrial robot, planning a coaxial interpolation path based on the normal vector reference, for controlling the six-axis industrial robot to carry the piezoelectric variable diameter spray head assembly deep into the deep blind hole to perform coaxial interpolation operations.
4. The intelligent adaptive cleaning system for power battery components as described in claim 1, characterized in that, The main control unit is equipped with a proportional-integral-derivative (PID) controller. The main control unit is configured to continuously monitor the sudden change trend of the back pressure change rate after performing time derivative calculation on the back pressure. In response to the sudden increase in the back pressure change rate exceeding the first warning threshold, the underlying limiting logic of the PID controller is triggered to reduce the opening of the electro-hydraulic proportional servo valve.
5. The intelligent adaptive cleaning system for power battery components as described in claim 4, characterized in that, The step of sending a retraction command to the six-axis industrial robot arm is configured as follows: simultaneously triggering the underlying limiting logic of the proportional-integral-derivative controller, establishing a backpressure differential linkage interlock control closed loop; based on the backpressure differential linkage interlock control closed loop, interrupting the current downward movement command of the six-axis industrial robot arm; generating a retraction trajectory along the opposite direction of the normal vector of the hole feature surface based on the previously extracted feature vector; and controlling the six-axis industrial robot arm to perform a reverse retraction operation according to the retraction trajectory.
6. A smart adaptive cleaning method for power battery components, characterized in that, An intelligent adaptive cleaning system for power battery components as described in claim 1, the method comprising: acquiring a point cloud obtained by scanning the power battery component using a three-dimensional structured light scanner; determining the surface shape of the cleaning area based on the point cloud; in response to the surface shape of the cleaning area being identified as a deep blind hole, extracting the minimum eigenvector of the covariance matrix of the hole feature surface in the point cloud, which is used to guide the coaxial feed centering of a six-axis industrial robotic arm; outputting a pulse width modulation voltage to a ring piezoelectric ceramic actuator to induce radial strain, which is used to compress a flexible silicone throat to form a water needle jet; during the cleaning of the deep blind hole, acquiring the back pressure of the fluid inside the hole collected by a high-frequency pressure sensor; performing a differential calculation on the back pressure in the time dimension to obtain the back pressure change rate; in response to the back pressure change rate being greater than a first warning threshold, triggering an attenuation signal to reduce the opening of the electro-hydraulic proportional servo valve, and sending a retraction command to the six-axis industrial robotic arm in the opposite direction of the normal vector.
7. The intelligent adaptive cleaning method for power battery components as described in claim 6, characterized in that, After determining the surface shape of the cleaning area based on the point cloud, the method further includes: in response to recognizing that the surface shape of the cleaning area is a flat thin-walled surface, disconnecting the pulse width modulation voltage output to the annular piezoelectric ceramic actuator, so that the flexible silicone throat returns to its original inner diameter size; and simultaneously outputting a pressure reduction command to reduce the opening of the electro-hydraulic proportional servo valve, and outputting an umbrella-shaped water flow.
8. The intelligent adaptive cleaning method for power battery components as described in claim 6, characterized in that, The step of performing a time-dimensional differential calculation on the recoil back pressure to obtain the back pressure change rate includes: extracting back pressure acquisition values within two consecutive high-frequency sampling periods on the time axis; calculating the difference between the back pressure acquisition values within the two consecutive high-frequency sampling periods; dividing the difference between the back pressure acquisition values by the time span of the high-frequency sampling period to obtain a single differential result; and performing a moving average filter on the single differential result within multiple consecutive periods to obtain the smoothed back pressure change rate.
9. The intelligent adaptive cleaning method for power battery components as described in claim 8, characterized in that, The step of using the trigger attenuation signal to reduce the opening of the electro-hydraulic proportional servo valve and sending a retraction command to the six-axis industrial robot in the opposite direction of the normal vector specifically includes: setting multi-level warning judgment conditions, the multi-level warning judgment conditions including a first warning threshold and a second warning threshold greater than the first warning threshold; in response to the back pressure change rate being greater than the first warning threshold and less than the second warning threshold, triggering only the attenuation signal to reduce the opening of the electro-hydraulic proportional servo valve; in response to the back pressure change rate being greater than the second warning threshold, reducing the control opening of the electro-hydraulic proportional servo valve to zero and calling the inverse kinematics model to generate a normal reverse displacement matrix; driving the six-axis industrial robot to perform a rigid retraction action according to the normal reverse displacement matrix.
10. A computer device, characterized in that, include: Memory, used to store computer programs; A processor, electrically connected to the memory, is used to execute the computer program to implement the intelligent adaptive cleaning method for power battery components as described in any one of claims 6 to 9.