A coating test robot system

Abstract

This invention discloses a coating testing robot system, comprising: a multi-axis servo motion control unit; a multi-sensor fusion sensing unit; an intelligent path planning unit; a system collaborative optimization unit; and a data processing and analysis unit. The multi-axis servo motion control unit executes an enhanced PID control algorithm; the multi-sensor fusion sensing unit executes a weighing sensor signal processing algorithm and a liquid level monitoring algorithm; the intelligent path planning unit executes a fifth-order polynomial trajectory interpolation algorithm and a dynamic speed optimization algorithm; and the system collaborative optimization unit dynamically adjusts the control parameters of each subsystem based on real-time performance feedback. This invention has the following advantages: this coating testing robot system can effectively solve the technical problems of low accuracy, poor repeatability, and non-quantifiable data in traditional coating testing, achieving the technical goals of coating force control accuracy ±0.05N, motion trajectory repeatability accuracy ±0.1mm, and test result repeatability >99%.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot control technology, and in particular to a coating test robot system. Background Technology

[0002] With the rapid development of Industry 4.0 and intelligent manufacturing, the automation and standardization of coating processes have become important development directions for modern manufacturing. Traditional coating testing mainly relies on manual operation, which faces many challenges in coating quality inspection, process optimization, and production standardization. Currently, the coating equipment commonly found on the market is mainly manual or semi-automatic, and its functions are basically limited to basic brushing operations. It lacks precise force control, real-time monitoring, and intelligent optimization capabilities, making it difficult to meet the needs of modern industry for high-precision and high-consistency coating testing.

[0003] The existing technology has the following problems:

[0004] First, there is the problem of insufficient precision in controlling the brushing force. In traditional manual brushing processes, it is difficult to precisely control the magnitude and distribution of the brushing force, resulting in uneven coating thickness and affecting the consistency of product quality. The brushing force variation range of existing equipment is usually above ±20%, far exceeding the ±5% precision range required by industry standards, which cannot guarantee the stability and reliability of product quality.

[0005] Secondly, there is the problem of poor repeatability of motion trajectories. Manually operated motion trajectories are affected by various factors such as operator skill level, fatigue level, and environmental factors, resulting in poor repeatability and making it difficult to guarantee the comparability and standardization of test results. The trajectory deviation of traditional equipment can reach ±2mm or more, which cannot meet the requirements of high-precision testing and affects the reliability and application value of the test data.

[0006] Third, there is the problem of lack of real-time monitoring capabilities. Traditional methods lack the ability to monitor and adjust key parameters in real time during the painting process. Important parameters such as paint level, brush weight, and painting pressure cannot be effectively monitored, which makes it impossible to achieve process optimization and quality early warning, affecting production efficiency and product quality stability.

[0007] Fourth, there is the problem of insufficient data collection and analysis capabilities. Manual operation makes it difficult to obtain quantitative data on the coating process, lacking objective evaluation and analysis methods for coating quality, and thus unable to provide a scientific basis for product quality improvement and process optimization. Existing equipment lacks complete data recording and analysis functions, making it difficult to support the digitalization and intelligentization requirements of modern manufacturing.

[0008] Fifth, there is the problem of low system integration and poor coordination. In existing technologies, each functional module usually works independently, lacking an effective coordination mechanism, which makes it impossible to maximize the overall system performance. The lack of deep integration between subsystems such as multi-axis motion control, sensor monitoring, and path planning leads to low overall system efficiency and fails to fully leverage the advantages of each subsystem.

[0009] Research at home and abroad shows that although there has been some progress in servo control technology, sensor technology and robot control technology, there is still a technological gap in deeply integrating technologies such as multi-axis collaborative servo control, multi-sensor fusion perception, intelligent path planning and system collaborative optimization to achieve intelligent control of high-precision painting test robots. Summary of the Invention

[0010] The technical problem to be solved by this invention is: how to achieve high-precision multi-axis collaborative control, multi-sensor fusion perception and intelligent optimization in a coating test robot system, maintain high precision, high reliability and strong adaptability in complex industrial environments, while taking into account the real-time performance, accuracy and effectiveness of the system.

[0011] To address the aforementioned technical problems, this invention provides a coating testing robot system. The system includes a multi-axis servo motion control unit, a multi-sensor fusion sensing unit, an intelligent path planning unit, a system collaborative optimization unit, and a data processing and analysis unit.

[0012] The multi-axis servo motion control unit includes a left-side Y-axis servo module, a right-side Y-axis servo module, an X-axis servo motor module, and a Z-axis servo module, used to achieve three-dimensional precision motion control of the brush clamping and weighing device. This unit employs an enhanced PID control algorithm to achieve high-precision coordinated control of multi-axis motion. Through an inter-axis synchronization compensation algorithm, it effectively solves the problem of inter-axis asynchrony caused by traditional independent control, ensuring the coordination and consistency of motion across all axes.

[0013] The multi-sensor fusion sensing unit includes a load cell, a liquid level monitoring device, and a high-definition camera, used for real-time monitoring of brush weight, paint level, and painting process status. This unit executes load cell signal processing algorithms and liquid level monitoring algorithms, achieving accurate measurement and real-time monitoring of key parameters in the painting process through adaptive filtering and multi-sensor data fusion. The load cell employs a strain gauge bridge structure, achieving high-precision measurement of the painting force through signal amplification, filtering, and linearization.

[0014] The intelligent path planning unit generates smooth painting trajectories, ensuring the continuity and consistency of motion. This unit executes a fifth-order polynomial trajectory interpolation algorithm and a dynamic velocity optimization algorithm to guarantee the continuity of position, velocity, and acceleration, achieving a smoother motion trajectory. Through a multi-axis coordinated motion algorithm, the synchronization of motion along each axis is ensured, and the motion speed is optimized in real time based on painting quality requirements and system dynamic performance, achieving the optimal balance between painting quality and efficiency.

[0015] The system collaborative optimization unit coordinates the working states of various subsystems to achieve multi-objective optimization of accuracy, efficiency, and stability. Based on real-time performance feedback, this unit dynamically adjusts the control parameters of each subsystem, establishing a multi-objective collaborative optimization framework for accuracy, efficiency, and stability to optimize the overall system performance. Through a self-learning parameter optimization mechanism, continuous improvement and adaptive enhancement of system performance are achieved.

[0016] The data processing and analysis unit executes adaptive PID control algorithms, sensor signal processing algorithms, and trajectory optimization algorithms. This unit integrates digital signal processing technology and intelligent algorithms to achieve real-time processing, analysis, and optimization of system operation data. By establishing a complete data acquisition and analysis system, it provides a basis for the quantitative evaluation of coating quality and process optimization.

[0017] The core of this invention lies in the deep integration of multi-axis cooperative servo control technology with multi-sensor fusion sensing technology. Through enhanced PID control algorithms and adaptive parameter adjustment mechanisms, high-precision control of complex multi-axis motions is achieved. Simultaneously, a fifth-order polynomial trajectory interpolation algorithm and a dynamic velocity optimization algorithm ensure the smoothness and continuity of the motion trajectory. The system utilizes a multi-objective cooperative optimization framework to achieve coordinated operation of each subsystem and maximize overall performance.

[0018] In summary, the present invention has the following beneficial effects:

[0019] First, it significantly improves the accuracy and repeatability of coating control. Through an enhanced PID control algorithm and adaptive gain adjustment mechanism, the coating force control accuracy is achieved to ±0.05N, which is more than 10 times higher than that of traditional methods. The repeatability accuracy of the motion trajectory is controlled within ±0.1mm, the trajectory tracking accuracy is improved to ±0.05mm, and the repeatability of test results reaches over 99%, significantly improving the consistency and stability of product quality.

[0020] Second, it achieves intelligent sensing capabilities through multi-sensor fusion. By employing signal processing algorithms from weighing sensors and liquid level monitoring algorithms, it enables real-time monitoring and precise measurement of key parameters in the coating process. This results in a 25dB improvement in signal-to-noise ratio, a 3-fold increase in measurement accuracy, an 80% improvement in measurement reliability, and significantly enhanced fault tolerance, providing a reliable data foundation for the system's intelligent control.

[0021] Third, it significantly improves the system's dynamic performance and motion smoothness. Through a fifth-order polynomial trajectory interpolation algorithm, motion smoothness is improved by 60%, mechanical shock and vibration are reduced, and the lifespan of the actuator is extended by 30%. The system response time is less than 10ms, the settling time is less than 50ms, and the dynamic response speed is improved by 35%, meeting the requirements of highly dynamic applications.

[0022] Fourth, it achieves collaborative optimization of overall system performance. Through a multi-objective collaborative optimization framework, the overall system performance is improved by 40%, with each subsystem working collaboratively to avoid global performance loss caused by local optimization. Coating quality consistency is improved by 45%, overall work efficiency is improved by 25%, and it supports customized performance requirements for different application scenarios.

[0023] Fifth, it possesses powerful adaptive learning and optimization capabilities. The system has autonomous learning capabilities, resulting in a performance improvement of over 20% over long-term use, reducing manual parameter tuning workload by 90%, and adapting to equipment aging and environmental changes. Its continuous operating MTBF is greater than 8760 hours, with a high level of automation, reducing manual intervention by over 85%.

[0024] Sixth, it possesses excellent engineering application value and industrialization prospects. The system's modular design supports functional expansion and upgrades, and development and deployment costs are controllable. It has broad application prospects in fields with stringent coating quality requirements, such as automotive manufacturing, aerospace, and shipbuilding. Through standardized coating tests, it can significantly improve product quality consistency and reduce quality costs by 15-20%, providing crucial technical support for the intelligent upgrading of the manufacturing industry. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the coating test robot system of the present invention;

[0026] Figure 2 This is a flowchart of the algorithm collaboration process of the system of the present invention;

[0027] Figure 3 This is a flowchart illustrating the implementation of the enhanced PID control algorithm in this invention.

[0028] Figure 4 This is a flowchart illustrating the implementation of the multi-sensor fusion sensing algorithm in this invention.

[0029] Figure 5 This is a flowchart illustrating the implementation of the fifth-order polynomial trajectory interpolation algorithm in this invention.

[0030] Figure 6 This is a schematic diagram illustrating the working principle of the system collaborative optimization algorithm in this invention.

[0031] Figure 7This is a schematic diagram of the multi-axis cooperative motion control in this invention;

[0032] Figure 8 This is a schematic diagram of the signal processing of the weighing sensor in this invention. Detailed Implementation

[0033] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0034] like Figure 1 , Figure 2 As shown, the coating test robot system provided by the present invention includes a multi-axis servo motion control unit 1, a multi-sensor fusion sensing unit, an intelligent path planning unit, a system collaborative optimization unit, and a data processing and analysis unit.

[0035] The multi-axis servo motion control unit 1 is based on a 32-bit ARM Cortex-A9 processor platform with a main frequency of 1.2GHz. It integrates a floating-point unit and a DSP coprocessor, and is specifically optimized for the execution efficiency of the multi-axis collaborative control algorithm. This unit includes a left Y-axis servo module, a right Y-axis servo module, an X-axis servo motor module, and a Z-axis servo module. Each servo module is equipped with a high-precision incremental encoder with a resolution of 262,144 pulses / revolution and an angle measurement accuracy of 0.005 degrees. The servo motors are permanent magnet synchronous motors with a rated power of 2.3kW and a maximum speed of 3000rpm, featuring high dynamic response and low inertia.

[0036] The multi-sensor fusion sensing unit adopts a distributed architecture design, integrating multiple high-precision sensors. The load cell uses a strain gauge bridge structure, with a measurement range of 0-50N, an accuracy class of 0.02%FS, a temperature coefficient of less than 0.0015%FS / ℃, and excellent linearity and repeatability. The liquid level monitoring device uses ultrasonic ranging, operates at a frequency of 40kHz, has a measurement range of 50-3000mm, an accuracy of ±0.5mm, and temperature compensation. The high-definition camera has a resolution of 1920×1080 pixels, a frame rate of 60fps, and is equipped with autofocus and adaptive lighting functions.

[0037] The intelligent path planning unit is implemented based on a dedicated motion control processor, integrating trajectory planning algorithms and real-time interpolation functions. This unit supports various trajectory types, including straight lines, circular arcs, and spline curves, and possesses speed prediction and dynamic adjustment capabilities. Path planning accuracy reaches 0.01mm, and speed control accuracy is 0.1%, meeting the requirements of high-precision painting tests.

[0038] The system collaborative optimization unit implements a multi-level system optimization mechanism, including adaptive parameter adjustment, performance monitoring and evaluation, and fault diagnosis and handling. This unit establishes a complete performance evaluation system, monitors the system's accuracy, efficiency, and stability indicators in real time, and performs dynamic optimization based on preset objectives.

[0039] The data processing and analysis unit adopts a high-performance embedded computing platform, equipped with an ARM Cortex-A57 quad-core processor with a main frequency of 1.8GHz, 4GB of LPDDR4 memory, and 64GB of eMMC storage. This unit integrates complete data acquisition, processing, storage, and analysis functions, supporting real-time data processing and historical data analysis.

[0040] like Figure 3 As shown, the enhanced PID control algorithm is the core algorithm for system motion control, and its mathematical expression is:

[0041] θ(t)=θ ref (t)+K p (t)×e θ (t)+K i (t)×∫[0,t]e θ (τ)dτ+K d (t)×de θ (t) / dt+F ff (t). Where θ(t) represents the current angular position of the motor, in radians, and signifies the actual angular position of the servo motor rotor, ranging from 0 to 2π radians. θ ref (t) represents the target angular position reference signal, in radians, and signifies the target angle that the system expects the servo motor to reach. It is generated by the upper-level path planning unit. e θ (t)=θ ref (t)-θ meas (t) represents the position error, in radians, and signifies the deviation between the target position and the actual measured position. It is the main input signal for the control algorithm. θ meas (t) represents the actual angular position fed back by the encoder, in radians. It means the actual angular position of the motor rotor obtained by the encoder, with a measurement accuracy of 0.005 degrees.

[0042] K p (t),K i (t),K d (t) represent the adaptive PID gain parameters, where K p (t) represents the proportional gain, which is dimensionless and represents the strength of the response to the current position error. It typically ranges from 10 to 100. i(t) represents the integral gain, in units of 1 / s, which indicates the ability to eliminate historical accumulated errors; its value typically ranges from 0.1 to 10. d (t) represents the differential gain, measured in seconds (s), which indicates the ability to predict and suppress the trend of error changes. Its value typically ranges from 0.001 to 0.1. F ff (t) represents the feedforward compensation term, in radians, which is a feedforward control quantity predicted based on the system model and used to improve the dynamic response performance of the system.

[0043] The mathematical expression for the adaptive gain adjustment mechanism is: K p (t)=K p,base ×[1+α×|e θ (t)|],K i (t)=K i,base ×[1-β×|de θ [(t) / dt|],K d (t)=K d,base ×[1+γ×σ e (t)]. Where K p,base ,K i,base ,K d,base α, β, and γ represent the base gain parameters, determined through system identification and empirical adjustment, with values ​​of 50, 2.5, and 0.05 respectively. α, β, and γ represent adaptive adjustment coefficients, all dimensionless parameters, signifying control over the sensitivity of adaptive adjustment. These are determined through experimental optimization, typically with values ​​of α=0.1, β=0.2, and γ=0.15. σ e (t) represents the error variance, in rad. 2 The significance of this is the statistical dispersion of the position error, which is used to assess the stability of the system.

[0044] like Figure 7 As shown, the multi-axis speed coordination control algorithm ensures the synchronization and smoothness of motion on each axis, and its mathematical expression is: ω e (t)=ω ref,e (t)+K v,e ×[ω ref,e (t)-ω meas,e (t)]+C sync,e (t). Where ω e (t) represents the angular velocity output of the i-th axis, in rad / s. It represents the actual operating angular velocity of the servo motor on the i-th axis, with a value range of -100 to 100 rad / s. ω ref,e (t) represents the target angular velocity along the i-th axis, in rad / s, and signifies the desired running speed along the i-th axis, generated by the path planning algorithm. ω meas,e(t) represents the actual measured angular velocity of the i-th axis, in rad / s, and is the actual angular velocity calculated by the encoder differential.

[0045] K v,e This represents the velocity gain along the i-th axis, in dimensionless form. It signifies the proportional gain for velocity control and typically ranges from 0.5 to 2.0. C sync,e (t) represents the inter-axis synchronization compensation term, with the unit being rad / s. It is a compensation speed used to eliminate asynchronous motion between axes.

[0046] The mathematical expression for the inter-axle synchronization compensation algorithm is:

[0047] C sync,e (t)=Σ[j≠i]K ej ×[ω ref,j (t) / R j -ω ref,e (t) / R e ]. Where K ej R represents the inter-axis coupling gain matrix, a dimensionless parameter that represents the coupling strength of the j-th axis to the i-th axis. It is determined through system modeling and experiments, and its value typically ranges from 0.01 to 0.1. e ,R j These represent the transmission ratios of the i-th and j-th axes, respectively. They are dimensionless parameters and represent the transmission ratio of the reducer. They are determined according to the mechanical design and are usually taken as 50:1 to 200:1.

[0048] For precise control of the coating pressure, the system employs a closed-loop control algorithm based on force feedback, the mathematical expression of which is:

[0049] τ(t)=τ ff (t)+K f ×[F target (t)-F meas (t)]+K i,f ×∫[0,t][F target (τ)-F meas [(τ)]dτ. Where τ(t) represents the motor output torque, in N·m, and signifies the driving torque output by the servo motor, with a value range of 0-20 N·m. ff (t) represents the feedforward torque, with units of N·m. It is the feedforward control torque calculated based on the system model and is used to compensate for known disturbances such as gravity and friction.

[0050] F target (t) represents the target coating force, in N, and signifies the desired coating pressure of the system. It is set according to the coating process requirements and typically ranges from 0.5 to 10 N. meas(t) represents the actual brushing force measured by the load cell, in N, and signifies the brushing contact force measured in real time by the load cell. K f K represents the proportional gain of force control, with units of (N·m) / N. It represents the proportional coefficient from force error to torque output, and its value typically ranges from 0.1 to 1.0. i,f This represents the integral gain of force control, with units of (N·m) / (N·s). It is the integral coefficient from the force error to the torque output, and its value is usually in the range of 0.01-0.5.

[0051] like Figure 8 As shown, the signal processing algorithm for the weighing sensor includes three main steps: signal amplification, filtering, and linearization. The mathematical expression for the signal amplification stage is: W raw (t)=G amp ×V bridge (t). Where W raw (t) represents the original weight signal, in units of V, and signifies the voltage signal after amplification, with a value range of 0-10V. bridge (t) represents the output voltage of the strain gauge bridge, in units of V. It represents the weak voltage signal generated by the strain gauge bridge under stress, and its value is typically in the range of 0-50mV. amp This indicates the amplifier gain, measured in V / V. It represents the signal amplification factor and is determined based on the sensor characteristics. Typically, it is set to 200.

[0052] The signal filtering stage uses a low-pass filter with the transfer function H. lpf (s), with a cutoff frequency set to 50Hz, is used to remove high-frequency noise and interference. The expression for the filtered signal is: W filtered (t)=H lpf (s)×W raw (t). Where W filtered (t) represents the filtered weight signal, in units of V, and signifies the net weight signal after noise removal.

[0053] The digital filtering algorithm uses a second-order Butterworth low-pass filter:

[0054] H lpf (z)=(b0+b1×z -1 +b2×z -2 ) / (1+a1×z -1 +a2×z -2 The filter coefficients are: b0=0.0067, b1=0.0134, b2=0.0067, a1=-1.1430, a2=0.4128. These coefficients were obtained from the analog filter design through bilinear transformation, ensuring the stability and filtering effect of the digital filter.

[0055] The mathematical expression for the signal linearization stage is: W calibrated (t)=A×W filtered (t)+B×W filtered 2 (t)+C. Where W calibrated (t) represents the calibrated weight value in N, signifying the final weight measurement result after linearization. A, B, and C represent calibration coefficients, with units of N / V, N / V, and N / V respectively. 2 These coefficients, N, are determined through standard weight calibration and are typically set to A=5.0, B=0.01, and C=0.

[0056] The liquid level monitoring algorithm employs ultrasonic ranging principles, combined with temperature compensation and multi-point averaging algorithms to achieve high-precision measurement. The mathematical expression for the original ranging calculation is: d raw (t)=(c×t echo (t)) / 2. Where d raw (t) represents the original ranging value in mm, signifying the straight-line distance from the ultrasonic sensor to the liquid surface. c represents the speed of sound, with a standard value of 343 m / s, measured at 20°C and 1 standard atmosphere. t echo (t) represents the echo time, measured in seconds, which is the round-trip time of the ultrasonic wave from transmission to reception, with a measurement accuracy of 1 μs.

[0057] The mathematical expression for temperature compensation calculation is: d compensated (t)=d raw (t)×[1+α temp ×(T(t)-T ref )]. Where d compensated (t) represents the temperature-compensated distance in mm, signifying the corrected distance value after taking temperature effects into account. α temp This represents the temperature compensation coefficient, with units of 1 / ℃, typically 0.0017 / ℃, and signifies the coefficient of sound velocity changing with temperature. T(t) represents the current temperature, with units of ℃, measured in real-time by a temperature sensor, with a measurement accuracy of ±0.5℃. ref This indicates the reference temperature, which is usually set to 20℃.

[0058] The mathematical expression for calculating the liquid level height is: H liquid (t)=H tank -d compensated (t)-H offset H liquid (t) represents the liquid level height, in mm, which means the height of the liquid surface in the paint bucket from the bottom of the bucket. H tank This indicates the total height of the paint bucket, in mm, and is determined based on the paint bucket's specifications; a typical value is 300 mm. Hoffset This indicates the sensor installation offset, in mm, representing the distance between the sensor installation position and the top of the paint bucket, determined through installation measurements.

[0059] The mathematical expression for the multi-point averaging filter algorithm is:

[0060] H averaged (t)=(1 / N)×Σ[i=0,N-1]H liquid (ti×Δt). Where H averaged (t) represents the average filtered liquid level height in mm, signifying the liquid level value reduced by random error through multi-point averaging. N represents the number of averaging points, typically N=5, representing the number of historical data points used in the averaging calculation. Δt represents the sampling interval, usually set to 0.1s, signifying the sampling period for the liquid level data.

[0061] like Figure 5 As shown, the fifth-order polynomial trajectory interpolation algorithm guarantees the smoothness and continuity of the motion trajectory. The mathematical expression of the position function is: q(t) = a0 + a1×t + a2×t 2 +a3×t 3 +a4×t 4 +a5×t 5 Where q(t) represents the position function, in mm or rad, and signifies the position of the motion axis at time t. Its value range is determined by the mechanical limits. a0-a5 represent the polynomial coefficients, determined by boundary conditions, and their physical meanings are the initial position, initial velocity, half of the initial acceleration, etc., respectively.

[0062] The mathematical expression for the velocity function is:

[0063] v(t)=dq(t) / dt=a1+2×a2×t+3×a3×t 2 +4×a4×t 3 +5×a5×t 4 Where v(t) represents the velocity function, with units of mm / s or rad / s, and signifies the velocity of the motion axis at time t. The range of values ​​is limited by the maximum speed of the system.

[0064] The mathematical expression for the acceleration function is:

[0065] a(t)=dv(t) / dt=2×a2+6×a3×t+12×a4×t 2 +20×a5×t 3 Where a(t) represents the acceleration function, with units of mm / s². 2 or rad / s 2 The value of is the acceleration of the motion axis at time t, and its range is limited by the maximum acceleration of the system.

[0066] The boundary condition constraint equations are: q(0) = q start ,q(T)=q end v(0)=v start v(T)=v end , a(0)=a start ,a(T)=a end Where q start ,q end These represent the start and end positions, respectively. start ,v end These represent the initial and final velocities, a and a', respectively. start ,a end These represent the initial and final accelerations, respectively, and T represents the total motion time in seconds.

[0067] like Figure 7 As shown, the multi-axis coordinated motion algorithm employs an interpolation method based on time synchronization. The mathematical expression for the normalized path parameters is: S(t) = S total ×f(t / T total Where S(t) represents the normalized path parameter, ranging from [0,1], and signifies the percentage of motion completed. total This represents the normalized value of the total path length, usually set to 1. f(t / T) total ) represents an S-shaped acceleration / deceleration function, which means a time function that achieves smooth acceleration / deceleration.

[0068] The mathematical expressions for calculating the coordinates of each axis are as follows:

[0069] x i (t)=x istart +(x iend -x istart )×S(t),

[0070] y i (t)=y istart +(y iend -y istart )×S(t),

[0071] z i (t)=z istart +(z iend -z istart )×S(t). Where x i (t),y i (t),z i (t) represents the instantaneous position of each axis, in mm, and signifies the real-time position coordinates of each axis in the three-dimensional coordinate system. istart ,y istart ,z istartrespectively represent the starting positions of each axis, x iend , y iend , z iend respectively represent the ending positions of each axis. T total represents the total motion time, in seconds, and is determined by calculating based on the speed limit and path length.

[0072] The mathematical expression of the S-shaped acceleration and deceleration function is: f(u) = 2u 2 When 0 ≤ u ≤ 0.5, f(u) = 1 - 2(1 - u) 2 When 0.5 < u ≤ 1. Where u represents the normalized time parameter, with a range of [0, 1], and its meaning is the normalized representation of the motion time. This function ensures the smoothness of the motion process and avoids sudden changes and impacts.

[0073] The dynamic speed optimization algorithm optimizes the motion speed in real time according to the painting quality requirements and system dynamic performance. The mathematical expression for calculating the optimized speed is: v optimal (t) = v max × η quality (t) × η dynamic (t). Where v optimal (t) represents the optimized motion speed, in mm / s, and its meaning is the optimal motion speed after comprehensively considering quality and dynamic performance. v max represents the maximum allowable speed, in mm / s, and is determined according to the system performance and safety requirements, usually taking a value of 100 mm / s.

[0074] (t). Where η quality (t) represents the quality constraint factor, a dimensionless parameter, and its meaning is the speed adjustment coefficient based on the painting quality requirements, with a value range of [0, 1]. k quality represents the quality weight coefficient, a dimensionless parameter, usually taking a value of 0.1, and its meaning is the sensitivity of the quality error to the speed influence. F error (t) represents the painting force error, in N, and its meaning is the deviation between the current painting force and the target painting force.

[0075] The mathematical expression of the dynamic constraint factor is: η dynamic (t) = exp(-k dynamic × |a max (t)|). Where η dynamic (t) represents the dynamic constraint factor, a dimensionless parameter, and its meaning is the speed adjustment coefficient based on the system dynamic performance, with a value range of [0, 1]. k dynamic ​​​​​​This represents the dynamic weighting coefficient, a dimensionless parameter, typically set to 0.01, signifying the sensitivity of acceleration to the effect of velocity. max (t) represents the maximum acceleration, in mm / s². 2 The meaning is the maximum acceleration requirement for the current motion.

[0076] like Figure 6 As shown, the system collaborative optimization algorithm employs a multi-objective optimization method to coordinate the overall system performance. The mathematical expression for the overall performance index is: J total =w1×J accuracy +w2×J efficiency +w3×J stability J total This represents the overall performance index, a dimensionless parameter, signifying a quantitative evaluation of the system's comprehensive performance. w1, w2, and w3 represent weighting coefficients, satisfying w1 + w2 + w3 = 1, signifying the importance weight of each sub-objective within the overall objective. Typically, these values ​​are w1 = 0.4, w2 = 0.35, and w3 = 0.25.

[0077] The mathematical expression for the accuracy index is: J accuracy =∫[0,T]|F target (t)-F actual (t)|dt. Where J accuracy This indicates the accuracy index, measured in N·s. It represents the cumulative value of the brushing force error throughout the entire test process; a smaller value indicates higher accuracy. F target (t) represents the target brushing force, F actual (t) represents the actual brushing force, and T represents the total test time.

[0078] The mathematical expression for the efficiency index is: J efficiency =T total / T optimal J efficiency T represents an efficiency index, a dimensionless parameter, meaning the ratio of actual time taken to the theoretical optimal time taken; the closer the value is to 1, the higher the efficiency. total T represents the total actual test time. optimal This represents the theoretically optimal time, with units of seconds.

[0079] The mathematical expression for the stability index is: J stability =σ F / F mean J stability This represents a stability index, a dimensionless parameter, signifying the relative coefficient of variation of the brushing force. A smaller value indicates better stability. σ F This represents the standard deviation of brushing force, expressed in N and F. mean This represents the average brushing force, expressed in N.

[0080] The mathematical expression for the adaptive parameter adjustment algorithm is: K new =K old +η× J(K old ). Where K new ,K old These represent the old and new parameter values, respectively, and represent the gain parameter vector of the PID controller. η represents the learning rate, a dimensionless parameter, typically set to 0.01, which signifies the step size for parameter updates. J(K) represents the performance index gradient. J(K) = ( J / K p , J / K i , J / K d The partial derivative of the performance index with respect to each parameter indicates the direction of parameter adjustment.

[0081] The mathematical expression for learning rate decay is: η(t) = η0 × exp(-α × t). Where η0 represents the initial learning rate, usually taken as 0.01, and α represents the decay coefficient, with the unit being 1 / s, usually taken as 0.001, which represents the rate at which the learning rate decays over time, ensuring the convergence of the algorithm.

[0082] This invention achieves a comprehensive breakthrough in the intelligent control system of the coating test robot in terms of control accuracy, measurement accuracy, system stability, and environmental adaptability through the deep integration of multi-axis collaborative servo control and multi-sensor fusion technology. It provides important technical support for quality inspection and process optimization in modern manufacturing industry and has broad application prospects and significant economic and social benefits.

[0083] To verify the above technical solution, the present invention designs the following calculation process to prove the effectiveness of the coating test robot system.

[0084] I. Test Scenario and System Parameter Settings

[0085] To verify the effectiveness of this invention, the quality inspection of the surface coating of aero-engine blades was used as the test subject, which included characteristics such as high precision requirements, complex curved surface coating, and adaptability to multiple materials. The test environment has typical characteristics such as strict precision requirements (±0.02N force control accuracy), complex geometry (blade twist angle of 15°), and high temperature environment (45° operating temperature).

[0086] 1.1 Basic System Configuration

[0087] Multi-axis servo motion control unit: sampling frequency 2kHz, ARM Cortex-A9; processor clock speed 1.2GHz, 2GB DDR3 memory, 32GB storage; eMMC servo motor configuration: X-axis travel 800mm, Y-axis travel 600mm, Z-axis travel 150mm, rotary axis ±180°, encoder resolution 262144 pulses / revolution; sensor configuration: load cell range 0-15N, resolution 0.001N, response frequency 1kHz; level sensor range 50-250mm, accuracy ±0.3mm; processor configuration: data processing unit ARM Cortex-A57, quad-core, 1.8GHz, 4GB LPDDR4 memory, floating-point operation capability 2.5GFLOPS; communication system: CAN bus communication frequency 1MHz, Ethernet communication bandwidth 100Mbps, real-time control cycle 1ms.

[0088] 1.2 Test Operating Parameters

[0089] Test environment temperature: 45℃±2℃ (simulating engine working environment); relative humidity: 35%±5%; test workpiece: titanium alloy blade, length 120mm, width 80mm, thickness 2.5mm, twist angle 15°; coating parameters: target coating force 4.2N, coating speed 35mm / s, path accuracy ±0.05mm; test coating: high temperature ceramic coating, viscosity 120s (Ford Cup 4), solids content 52%, working temperature range -40℃ to 850℃; test duration: continuous 6 hours, including testing of 150 blade samples, with a test cycle of 2.4 minutes per blade.

[0090] II. Calculation Process of Enhanced PID Control Algorithm

[0091] 2.1 Algorithm Parameter Settings

[0092] Based on actual operating conditions, the parameters of the enhanced PID control algorithm are set as follows: Basic proportional gain K p,base =48 (dimensionless, the basic response strength of the control system to position error) Basic integral gain K i,base =2.8 (unit 1 / s, basic ability to eliminate static error) Basic differential gain K d,base =0.065 (unit: s, basic ability to suppress overshoot and oscillation) Adaptive adjustment coefficient α = 0.15 (dimensionless, sensitivity of proportional gain adaptive adjustment) Adaptive adjustment coefficient β = 0.22 (dimensionless, sensitivity of integral gain adaptive adjustment) Adaptive adjustment coefficient γ = 0.18 (dimensionless, sensitivity of differential gain adaptive adjustment) Sampling period T s =0.5ms (control algorithm execution cycle) error threshold e threshold =0.01rad (Position error threshold for initiating adaptive adjustment)

[0093] 2.2 Data Acquisition

[0094] Taking the position control of the X-axis servo motor at t=14:25:30 as an example, the collected control data is as follows:

[0095] t=0ms: θ ref (0) = 1.2500 rad, θ meas (0) = 1.2485 rad, e θ (0) = 0.0015 rad,

[0096] t=0.5ms: θ ref (0.5) = 1.2518 rad, θ meas (0.5) = 1.2502 rad, e θ (0.5) = 0.0016 rad

[0097] t=1.0ms: θ ref (1.0) = 1.2536 rad, θ meas (1.0) = 1.2521 rad, e θ (1.0) = 0.0015 rad,

[0098] t=1.5ms: θ ref (1.5) = 1.2555 rad, θ meas (1.5) = 1.2541 rad, e θ (1.5) = 0.0014 rad.

[0099] t=2.0ms: θ ref (2.0) = 1.2573 rad, θ meas (2.0) = 1.2562 rad, e θ (2.0) = 0.0011 rad.

[0100] 2.3 Adaptive Gain Calculation Process

[0101] Step 1: Calculate the absolute value of the current position error |e θ (1.0)|=|0.0015|=0.0015rad

[0102] Step 2: Calculate the adaptive proportional gain

[0103] K p (1.0)=K p,base ×[1+α×|e θ (1.0)|]K p (1.0) = 48 × [1 + 0.15 × 0.0015]Kp (1.0) = 48 × [1 + 0.000225]K p (1.0) = 48 × 1.000225 = 48.0108

[0104] Step 3: Calculate the rate of change of error

[0105] de θ (1.0) / dt=[e θ (1.0)-e θ (0.5)] / T s =(0.0015-0.0016) / 0.0005=-0.1rad / s

[0106] Step 4: Calculate the adaptive integral gain

[0107] K i (1.0)=K i,base ×[1-β×|de θ (1.0) / dt|]

[0108] K i (1.0) = 2.8 × [1 - 0.22 × 0.1]K i (1.0) = 2.8 × [1 - 0.022]

[0109] K i (1.0) = 2.8 × 0.978 = 2.7384

[0110] Step 5: Calculate the error variance (based on the 10 most recent sampling points)

[0111] σ e (1.0)=sqrt(sum[(e θ (i)-e mean ) 2 ] / 10)=sqrt(sum[(0.0015-0.00145) 2 ] / 10)=0.000025rad 2

[0112] Step 6: Calculate the adaptive differential gain

[0113] K d (1.0)=K d,base ×[1+γ×σ e (1.0)]

[0114] K d (1.0)=0.065×[1+0.18×0.000025]K d (1.0) = 0.065 × [1 + 0.0000045]

[0115] K d (1.0) = 0.065 × 1.0000045 = 0.0650003

[0116] 2.4 PID Control Output Calculation Process

[0117] Step 1: Calculate the integral term (based on the trapezoidal integral method)

[0118] ∫[0,1.0]e θ (τ)dτ=T s ×[e θ (0)+2×e θ (0.5)+e θ (1.0)] / 2

[0119] ∫[0,1.0]e θ (τ)dτ=0.0005×[0.0015+2×0.0016+0.0015] / 2=0.00000155rad·s

[0120] Step 2: Calculate the feedforward compensation term (based on gravity and friction compensation) F ff (1.0)=τ gravity +τ friction =0.15 + 0.08 = 0.23 rad

[0121] Step 3: Calculate the final PID control output

[0122] θ(1.0)=θ ref (1.0)+K p (1.0)×e θ (1.0)+K i (1.0)×∫[0,1.0]e θ (τ)dτ+K d (1.0)×de θ (1.0) / dt+F ff (1.0)

[0123] θ(1.0)=1.2536+48.0108×0.0015+2.7384×0.00000155+0.0650003×(-0.1)+0.23

[0124] θ(1.0)=1.2536+0.0720162+0.00000424-0.00650003+0.23

[0125] θ(1.0) = 1.54911041 rad.

[0126] III. Calculation Process of Multi-Sensor Fusion Sensing Algorithm

[0127] like Figure 4 As shown:

[0128] 3.1 Algorithm Parameter Settings

[0129] Weighing sensor signal processing parameters: Amplifier gain G amp =250 (V / V, signal amplification factor) Low-pass filter cutoff frequency f c =45Hz (removing high-frequency noise) Linearization calibration coefficient A = 6.2 (N / V, first-order coefficient) Linearization calibration coefficient B = 0.015 (N / V) 2 Linearization calibration coefficient C = -0.05 (N, coefficient of the constant term)

[0130] Liquid level monitoring algorithm parameters: ultrasonic frequency f = 40kHz (ranging carrier frequency), standard sound velocity c = 343m / s (standard sound velocity at 20℃), temperature compensation coefficient α. temp =0.0017 / ℃ (sound speed temperature coefficient) Reference temperature T ref =20℃ (standard reference temperature) Paint bucket height H tank =220mm (total height of paint bucket) Sensor offset H offset =25mm (sensor installation offset) Number of filter points N=5 (number of points for multi-point average filtering) Sampling interval Δt=0.1s (liquid level data sampling period)

[0131] 3.2 Weighing Sensor Signal Processing and Calculation

[0132] Step 1: Raw signal acquisition and amplification. The bridge output voltage at time t=2.5s is measured as follows:

[0133] V bridge (2.5) = 0.0284V

[0134] W raw (2.5)=G amp ×V bridge (2.5) = 250 × 0.0284 = 7.1V

[0135] Step 2: Digital low-pass filtering uses a second-order Butterworth filter.

[0136] H lpf (z)=(0.0067+0.0134×z -1 +0.0067×z -2 ) / (1-1.1430×z -1 +0.4128×z -2 )

[0137] Input signal sequence: W raw (n-2)=7.05V,W raw (n-1) = 7.08 V, W raw (n) = 7.1V

[0138] Output signal sequence:

[0139] W filtered (n-2)=6.95V,

[0140] W filtered (n-1)=6.98V,

[0141] W filtered (n)=0.0067×7.1+0.0134×7.08+0.0067×7.05+1.1430×6.98-0.4128×6.95

[0142] W filtered (2.5)=0.04757+0.09487+0.04724+7.98214-2.86896=5.30286V.

[0143] Step 3: Signal linearization calibration

[0144] W calibrated (2.5) = A × W filtered (2.5)+B×W filtered 2 (2.5)+CW calibrated (2.5) = 6.2 × 5.30286 + 0.015 × (5.30286) 2 +(-0.05)=32.877732+0.421531-0.05=33.249263N.

[0145] 3.3 Liquid Level Monitoring Algorithm Calculation

[0146] Step 1: Echo time measurement of ultrasonic wave emission time: t send =0ms echo reception time: t receive =0.64ms round trip time: t echo (2.5) = 0.64 ms = 0.00064 s.

[0147] Step 2: Calculate the original distance

[0148] d raw (2.5)=(c×t echo (2.5)) / 2=(343×0.00064) / 2=0.109824m=109.824mm

[0149] Step 3: Calculate the current ambient temperature using temperature compensation.

[0150] T(2.5) = 45℃

[0151] d compensated (2.5)=d raw (2.5)×[1+α temp ×(T(2.5)-T ref )]

[0152] d compensated (2.5) = 109.824 × [1 + 0.0017 × (45 - 20)]

[0153] d compensated (2.5) = 109.824 × [1 + 0.0017 × 25]

[0154] d compensated (2.5) = 109.824 × [1 + 0.0425]

[0155] d compensated (2.5)=109.824×1.0425=114.4905mm

[0156] Step 4: Liquid Level Calculation

[0157] H liquid (2.5)=H tank -d compensated (2.5)-H offset H liquid (2.5)=220-114.4905-25=80.5095mm.

[0158] Step 5: Multi-point averaging and filtering of historical liquid level data

[0159] H liquid (2.1) = 80.2 mm, H liquid (2.2) = 80.3 mm, H liquid (2.3) = 80.4 mm, H liquid (2.4) = 80.5 mm

[0160] H averaged (2.5)=(1 / 5)×[80.2+80.3+80.4+80.5+80.5095]=(1 / 5)×401.9095=80.3819mm.

[0161] IV. Algorithm Co-optimization Calculation Process

[0162] 4.1 Calculation of Multi-Objective Performance Indicators

[0163] Step 1: Accuracy index calculation; Test cycle T = 144s (2.4 minutes for a single blade test); Target coating force F target =4.2N; Actual brushing force time series (sampled every 0.1s):

[0164] F actual =[4.18,4.21,4.19,4.22,4.20,...,4.19]N

[0165] J accuracy =∫[0,144]|F target (t)-F actual (t)|dt

[0166] J accuracy =0.1×Σ|4.2-F actual (i)|=0.1×[0.02+0.01+0.01+0.02+0.00+...+0.01]

[0167] J accuracy =0.1×24.5=2.45N·s.

[0168] Step 2: Calculate the theoretical optimal time T for efficiency index optimal =120s (calculated based on maximum speed and shortest path) Actual test time T total =144s; J efficiency =T total / T optimal =144 / 120=1.2.

[0169] Step 3: Calculate the average brushing force F as a stability index mean =(1 / 1440)×ΣF actual (i) = 4.198N;

[0170] Standard deviation of brushing power;

[0171] σ F =sqrt((1 / 1440)×Σ(F actual (i)-F mean ) 2 =0.0085N

[0172] J stability =σ F / F mean =0.0085 / 4.198=0.002025

[0173] Step 4: Calculate the weighting coefficients for the comprehensive performance indicators: w1=0.45 (accuracy weight), w2=0.35 (efficiency weight), w3=0.20 (stability weight).

[0174] J total =w1×J accuracy +w2×J efficiency +w3×J stability J total =0.45×2.45+0.35×1.2+0.20×0.002025J total =1.1025+0.42+0.000405=1.522905.

[0175] 4.2 Adaptive Parameter Optimization Calculation

[0176] Step 1: Calculate the current parameter vector using the performance metric gradient:

[0177] K old =[K p,base =48,K i,base =2.8,K d,base =0.065]

[0178] Parameter perturbation:

[0179] ΔK=0.1;

[0180] J / K p =[J(K p +ΔK)-J(K p )] / ΔK=(1.535-1.523) / 0.1=0.12;

[0181] J / K i =[J(K i +ΔK)-J(K i )] / ΔK=(1.518-1.523) / 0.1=-0.05;

[0182] J / K d =[J(K d +ΔK)-J(K d )] / ΔK=(1.528-1.523) / 0.1=0.05;

[0183] Gradient vector: J(K old = [0.12, -0.05, 0.05].

[0184] Step 2: Calculate the learning rate decay. Initial learning rate η0 = 0.015; decay coefficient α = 0.002s -1Current time t = 3600s (1 hour running time);

[0185] η(3600)=η0×exp(-α×t)=0.015×exp(-0.002×3600)=0.015×exp(-7.2)=0.015×0.00075=0.00001125

[0186] Step 3: Parameter Update Calculation

[0187] K new =K old +η× J(K old )

[0188] K pnew =48 + 0.00001125 × 0.12 = 48 + 0.00000135 = 48.00000135

[0189] K inew =2.8 + 0.00001125 × (-0.05) = 2.8 - 0.0000005625 = 2.7999994375

[0190] K dnew =0.065 + 0.00001125 × 0.05 = 0.065 + 0.0000005625 = 0.0650005625

[0191] Updated parameter vector: K new =[48.00000135,2.7999994375,0.0650005625]

[0192] V. Optimization and Control Effect Verification

[0193] Based on the above calculation results, the system implemented optimized control and compared the system state before and after control.

[0194] 5.1 Variation in brushing force control accuracy

[0195]

[0196] After the optimization control was implemented, the brushing force control accuracy improved from ±0.18N to ±0.025N, an improvement of 86.1%, which fully meets the stringent requirements for the inspection of aero-engine blade coatings.

[0197] 5.2 Motion trajectory accuracy effect

[0198]

[0199] The multi-axis collaborative control algorithm significantly improves the motion accuracy of the system, with the repeatability of each axis increasing by more than 75%, meeting the high-precision requirements for painting complex curved surfaces.

[0200] 5.3 Improvement in Sensor Measurement Accuracy

[0201]

[0202] Multi-sensor fusion algorithms effectively improve the measurement accuracy and reliability of sensors, providing a basis for precise control.

[0203] 5.4 Overall System Performance

[0204]

[0205] 5.5 Economic Benefit Analysis

[0206] Quality cost savings: Scrap rate reduced from 2.8% to 0.5%. Based on an annual production of 100,000 blades, the cost savings from scrap is (100,000 × 0.023 × 1200 yuan / piece) = 2.76 million yuan / year.

[0207] Benefits of increased efficiency: Reduced testing time: 0.4 minutes saved per item, annual labor hours saved: (100,000 × 0.4 minutes × 150 yuan / hour ÷ 60) = 1 million yuan / year;

[0208] Labor cost savings: Operator reduction: from 3 people to 1 person, annual labor cost savings: (2 people × 80,000 yuan / year) = 160,000 yuan / year;

[0209] Equipment maintenance costs are reduced: the failure rate is reduced by 88%, and the annual maintenance cost savings are: (original maintenance cost of 300,000 yuan × 0.88) = 264,000 yuan / year;

[0210] Total annual economic benefits: 276 + 100 + 16 + 26.4 = 418.4 million yuan / year.

[0211] VI. Conclusion

[0212] Through the above calculation process and result verification, the technical benefits of the coating testing robot system of the present invention in industrial application prospects are as follows:

[0213] The enhanced PID control algorithm achieves K through adaptive gain adjustment. p With a fine adjustment of 48.00000135, the control accuracy was improved from ±0.18N to ±0.025N, an increase of 86.1%. The multi-sensor fusion algorithm improved the accuracy of the weighing sensor from ±0.015N to ±0.003N and the accuracy of the liquid level detection from ±0.8mm to ±0.2mm, providing a reliable foundation for high-precision control.

[0214] The multi-objective optimization algorithm achieves the accuracy index J accuracy =2.45 N·s, efficiency index J efficiency =1.2, Stability index J stability The collaborative optimization with a value of 0.002025 yields a comprehensive performance index J. total =1.522905, a 35% improvement compared to before optimization. The adaptive parameter optimization mechanism continuously improves the long-term performance of the system, with parameter fine-tuning accuracy reaching 10. -8 level.

[0215] In the application of aero-engine blade coating inspection, the system achieves a 99.7% success rate, reduces single-piece testing time to 2.4 minutes, lowers system energy consumption by 34.4%, and reduces the failure rate by 88%. The annual economic benefit reaches 4.184 million yuan, with an investment payback period of less than 2 years, demonstrating excellent economic benefits and market prospects.

[0216] The fifth-order polynomial trajectory interpolation algorithm achieves continuous control of position, velocity, and acceleration, improving motion smoothness by 60%. The multi-axis collaborative control algorithm controls the inter-axis synchronization error within ±0.012mm, improving synchronization accuracy by 81.5%. The system has high integration and an automation level of 85%, providing important technical support for intelligent manufacturing.

[0217] This technology has wide application value in fields with stringent coating quality requirements, such as automobile manufacturing, aerospace, shipbuilding, and precision instruments. Through standardized intelligent coating testing, it can significantly improve product quality consistency, reduce quality costs by 15-20%, and promote the digital and intelligent transformation and upgrading of the manufacturing industry.

[0218] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any person skilled in the art can make various changes and modifications without departing from the scope of the technical solution of the present invention, and such changes and modifications should all fall within the scope of protection claimed by the present invention.

Claims

1. A coating testing robot system, characterized in that, include: A multi-axis servo motion control unit, including a left Y-axis servo module, a right Y-axis servo module, an X-axis servo motor module, and a Z-axis servo module, is used to achieve three-dimensional precision motion control of the brush clamping and weighing device; a multi-sensor fusion sensing unit, including a weighing sensor, a liquid level monitoring device, and a high-definition camera, is used to monitor the brush weight, paint level, and painting process status in real time; an intelligent path planning unit is used to generate a smooth painting trajectory to ensure the continuity and consistency of motion; a system collaborative optimization unit is used to coordinate the working states of each subsystem to achieve multi-objective optimization of accuracy, efficiency, and stability; a data processing and analysis unit is used to execute adaptive PID control algorithms, sensor signal processing algorithms, and trajectory optimization algorithms; the multi-axis servo motion control unit executes an enhanced PID control algorithm; the enhanced PID control algorithm is as follows: θ(t)=θ ref (t)+K p (t)×e θ (t)+K i (t)×∫[0,t]e θ (τ)dτ+K d (t)×de θ (t) / dt+F ff (t), where θ(t) is the current angular position of the motor in radians, θ ref (t) represents the target angular position reference signal, in radians, e θ (t)=θ ref (t)-θ meas (t) represents the position error, in radians, θ meas (t) represents the actual angular position fed back by the encoder, in radians, K. p (t),K i (t),K d (t) is the adaptive PID gain parameter, F ff (t) represents the feedforward compensation term, in radians; the adaptive gain adjustment mechanism is: K p (t)=K p,base ×[1+α×|e θ (t)|],K i (t)=K i,base ×[1-β×|de θ [(t) / dt|],K d (t)=K d,base ×[1+γ×σ e [(t)], where K p,base ,K i,base ,K d,base The base gain parameters are α, β, and γ, which are adaptive adjustment coefficients. ∫[0,t]e θ (τ)dτ is the integral term, σ e (t) represents the error variance, in rad. 2 ; The multi-axis servo motion control unit includes a multi-axis speed coordination control algorithm, which is: ω e (t)=ω ref,e (t)+K v,e ×[ω ref,e (t)-ω meas,e (t)]+C sync,e (t), where ω e (t) represents the angular velocity output along the i-th axis, in rad / s, ω ref,e (t) represents the target angular velocity along the i-th axis, in rad / s, ω meas,e (t) represents the actual measured angular velocity along the i-th axis, in rad / s, K v,e C is the velocity gain along the i-th axis. sync,e (t) represents the inter-axis synchronization compensation term, in rad / s; the algorithm for the inter-axis synchronization compensation term is: C sync,e (t)=Σ[j≠i]K ej ×[ω ref,j (t) / R j -ω ref,e (t) / R e ], where K ej R is the inter-axis coupling gain matrix. e ,R j The transmission ratio between the i-th and j-th axes is defined; the multi-sensor fusion sensing unit executes the weighing sensor signal processing algorithm and the liquid level monitoring algorithm; the intelligent path planning unit executes the fifth-order polynomial trajectory interpolation algorithm and the dynamic speed optimization algorithm; the system collaborative optimization unit dynamically adjusts the control parameters of each subsystem based on real-time performance feedback.

2. The coating testing robot system according to claim 1, characterized in that: The multi-sensor fusion sensing unit includes a torque and pressure control algorithm, which is as follows: τ(t)=τ ff (t)+K f ×[F target (t)-F meas (t)]+K i,f ×∫[0,t][F target (τ)-F meas dτ, where τ(t) is the motor output torque in N·m. ff (t) represents the feedforward torque, in N·m, F target (t) represents the target brushing force, in N or F. meas (t) represents the actual brushing force measured by the weighing sensor, in N, K. f For force control proportional gain, the unit is (N·m) / N, K i,f The integral gain is for force control, with units of (N·m) / (N·s); ∫[0,t][F target (τ)-F meas [(τ)]dτ is the force control integral term.

3. The coating testing robot system according to claim 1, characterized in that: The signal processing algorithm for the weighing sensor is as follows: W raw (t)=G amp ×V bridge (t), W filtered (t)=H lpf (s)×W raw (t), W calibrated (t)=A×W filtered (t)+B×W filtered 2 (t)+C, where W raw (t) represents the original weight signal, in units of V. bridge (t) represents the bridge output voltage, in V and G. amp Amplifier gain, in V / V, H lpf (s) is the transfer function of the low-pass filter, with a cutoff frequency of 50Hz, W filtered (t) represents the filtered weight signal, in V and W. calibrated (t) represents the calibrated weight value in N, and A, B, and C are calibration coefficients; a second-order Butterworth low-pass filter is used: H lpf (z)=(b0+b1×z -1 +b2×z -2 ) / (1+a1×z -1 +a2×z -2 ), where the filter coefficients are: b0=0.0067, b1=0.0134, b2=0.0067, a1=-1.1430, a2=0.4128, z -1 This indicates a signal delay of one sampling period, z -2 This indicates a signal delay of two sampling periods.

4. The coating testing robot system according to claim 1, characterized in that: The liquid level monitoring algorithm is: d raw (t)=(c×t echo (t)) / 2,d compensated (t)=d raw (t)×[1+α temp ×(T(t)-T ref )], H liquid (t)=H tank -d compensated (t)-H offset , where d raw (t) represents the original distance measurement value in mm, and c represents the speed of sound, with a standard value of 343 m / s. echo (t) represents the echo time, in seconds (s). compensated (t) represents the temperature-compensated distance in mm, α temp This is the temperature compensation coefficient, in units of 1 / ℃, typically 0.0017 / ℃. T(t) is the current temperature, in units of ℃. ref For reference temperature, typically 20℃, H tank The total height of the paint bucket is in mm (H). offset Sensor installation offset, in mm, H liquid (t) represents the liquid level height in mm; it also includes a multi-point averaging filtering algorithm, which is: H averaged (t)=(1 / N)×Σ[i=0,N-1]H liquid (ti×Δt), where H averaged (t) represents the average filtered liquid level height, N is the number of average points, usually N=5, and Δt is the sampling interval, usually 0.1s; H liquid (t) represents the liquid level height, in mm.

5. The coating testing robot system according to claim 1, characterized in that: The fifth-order polynomial trajectory interpolation algorithm is as follows: q(t)=a0+a1×t+a2×t 2 +a3×t 3 +a4×t 4 +a5×t 5 , v(t)=dq(t) / dt=a1+2×a2×t+3×a3×t 2 +4×a4×t 3 +5×a5×t 4 , a(t)=dv(t) / dt=2×a2+6×a3×t+12×a4×t 2 +20×a5×t 3 , Where q(t) is the position function, in mm or rad, v(t) is the velocity function, in mm / s or rad / s, and a(t) is the acceleration function, in mm / s². 2 or rad / s 2 t represents time; a0-a5 are polynomial coefficients, determined by boundary conditions; the constraint equations included in the boundary conditions are: q(0)=q start ,q(T)=q end v(0)=v start v(T)=v end , a(0)=a start ,a(T)=a end ; where q start ,q end These represent the start and end positions, respectively. start ,v end These represent the initial and final velocities, a and a', respectively. start ,a end These represent the initial and final accelerations, respectively, and T represents the total motion time in seconds.

6. The coating testing robot system according to claim 1, characterized in that: The intelligent path planning unit includes a multi-axis coordinated motion algorithm, which is as follows: S(t)=S total ×f(t / T total ), x i (t)=x istart +(x iend -x istart )×S(t), and i (t)=y istart +(and iend -and istart )×S(t), z i z(t)=z istart +(z iend -z istart )×S(t), where S(t) is the normalized path parameter with a range of [0, 1], and S total represents the normalized value of the total path length, and f(t / T total ) is the S-shaped acceleration and deceleration function, x i (t), y i (t), z i (t) are the instantaneous positions of each axis in mm, and T total is the total motion time in s; x istart , y istart , z istart respectively represent the starting positions of each axis, and x iend , y iend , z iend respectively represent the ending positions of each axis. The S-shaped acceleration and deceleration function is: f(u)=2u 2 When 0 ≤ u ≤ 0.5, f(u)=1 - 2(1 - u) 2 When 0.5 < u ≤ 1, where u represents the normalized time parameter.

7. The coating testing robot system according to claim 1, characterized in that: The dynamic speed optimization algorithm is as follows: v optimal (t)=v max ×η quality (t)×η dynamic (t), η quality (t)=1-k quality ×|F error (t)|, η dynamic (t)=exp(-k dynamic ×|a max (t)|), where v optimal (t) represents the optimized motion velocity, in mm / s, v max The maximum permissible speed is expressed in mm / s, η. quality (t) is the quality constraint factor, η dynamic (t) is the dynamic constraint factor, F error (t) represents the coating force error, in N, a max (t) represents the maximum acceleration, in mm / s². 2 k quality ,k dynamic These are the weighting coefficients.

8. The coating testing robot system according to claim 1, characterized in that: The system collaborative optimization unit includes a multi-objective optimization algorithm, which is as follows: J total =w1×J accuracy +w2×J efficiency +w3×J stability , J accuracy =∫[0,T]|F target (t)-F actual (t)|dt,J efficiency =T total / T optimal , J stability =σ F / F mean J total For the overall performance index, w1, w2, and w3 are weighting coefficients, satisfying w1 + w2 + w3 = 1, J accuracy For accuracy indicators, the units are N·s and F. target (t) represents the target brushing force, F actual (t) represents the actual brushing force, J efficiency As an efficiency indicator, J stability As a stability indicator, T optimal The theoretically optimal time is given in seconds (s), and σ F Standard deviation of brushing force, in N, F mean This represents the average brushing force, in N. T total This indicates the total actual test time; The system collaborative optimization unit further includes an adaptive parameter adjustment algorithm, which is as follows: K new =K old +η× J(K old ), J(K)=( J / K p , J / K i , J / K d ), η(t) = η0 × exp(-α × t), Where K new ,K old These are the old and new parameter values, and η is the learning rate. J(K) is the performance metric gradient, η0 is the initial learning rate, and α is the decay coefficient, with units of 1 / s; J(K) = ( J / K p , J / K i , J / K d ), which means the partial derivative of the performance index with respect to each parameter.

Images