Screen module fault original positioning method and system

Abstract

The invention discloses a screen module fault original positioning method and system. The method comprises the following steps: spraying a quantum dot solution on key nodes of a to-be-detected screen module; collecting a vibration spectrum of the screen module by using a laser Doppler vibration meter; the acoustic impedance Za of the screen module is obtained through analysis; the electrical impedance Ze, measured by the three-dimensional impedance vector field, of the screen module is utilized, and the acoustic impedance Za and the electrical impedance Ze are combined to obtain acoustic-impedance joint confidence; the first layer is used for judging the fluorescence intensity threshold value of each key point quantum dot, the second layer is used for carrying out acoustic-impedance joint confidence evaluation on the acoustic-impedance joint confidence model, the third layer is used for outputting a maintenance strategy of reinforcement learning, a decision tree model is constructed, and fault positioning is carried out on the screen module through the decision tree model. According to the invention, the fault is positioned through the change of the quantum dot response state of each key node, and the method has the advantages of high efficiency and high accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field related to the fault location of screen modules, and specifically to a method and system for locating faulty components of a screen module. Background Art

[0002] Existing LED screens usually need to monitor the working status of the lamp points they contain in real time, generate fault information in a timely manner when a lamp point fails, and perform maintenance in a timely manner. The traditional method for detecting LED screen lamp point faults is to judge by detecting the current and voltage of each lamp point. First, a sampling resistor needs to be connected to each lamp point, and then the voltage and current of the sampling resistor are detected manually, and finally the corresponding voltage and current of the lamp point are obtained through conversion. However, the above method requires manual detection of the circuit, and the circuit cannot actively report abnormalities, and the detection process is relatively troublesome, which is not conducive to the fault maintenance of LED screens. Summary of the Invention

[0003] To solve the defects existing in the prior art, the present invention provides a method and system for locating faulty components of a screen module.

[0004] To solve the above technical problems, the present invention provides the following technical solutions:

[0005] A method for locating faulty components of a screen module according to the present invention includes the following steps:

[0006] Step 1, perform quantum marking preprocessing; spray quantum dot solution on the key nodes of the screen module to be detected;

[0007] Step 2, perform photoexcitation detection; scan the screen module to be detected with a 405nm laser, and then capture the quantum dot response of the key nodes at each driving IC / wire crossing point through a hyperspectral camera to obtain the quantum dot response state of each key node, and then obtain the quantum dot fluorescence intensity of each key node;

[0008] Step 3, perform acoustic resonance imaging again; emit variable-frequency ultrasonic waves to the screen module, and then collect the vibration spectrum of the screen module using a laser Doppler vibrometer; and analyze to obtain the acoustic impedance Za of the screen module;

[0009] Step 4, measure the grid impedance using the four-wire method, construct a three-dimensional impedance vector field, use the electrical impedance Ze of the screen module measured by the three-dimensional impedance vector field, and combine the acoustic impedance Za and the electrical impedance Ze to obtain an acoustic-impedance combined confidence model;

[0010] Step 5: Use the first layer to judge the fluorescence intensity threshold of each key point quantum dot, the second layer to perform acoustic-impedance joint confidence evaluation with a row acoustic-impedance joint confidence model, and the third layer to output the maintenance strategy of reinforcement learning, and construct a decision tree model; use the decision tree model to locate the faults of the screen module.

[0011] As a preferred technical solution of the present invention, the method of spraying quantum dot solution on the key nodes of the screen module to be detected in Step 1 is to configure quantum dot solutions with different emission wavelengths and coat them with quantum dot solutions with different emission wavelengths in different fault-sensitive areas;

[0012] Adopt a spatial encoding strategy to locate the position of each quantum dot,

[0013] where the quantum dot wavelength encoding formula: λ QD = 520 + Δλ·(x + yM x )

[0014] In the above formula: λQD represents the fluorescence wavelength of the quantum dot, x represents the horizontal coordinate (0 ≤ x < M_x), y represents the vertical coordinate (y ≥ 0), and Mx represents the maximum number of horizontal grids, which is used to prevent wavelength aliasing.

[0015] As a preferred technical solution of the present invention, the quantum dot response formula in Step 2 is

[0016] I(λ, x, y) = I0e -α(λ)d ·η QD ,

[0017] where in the formula, I0 is the initial value of the incident light intensity, a(A) is the extinction coefficient dependent on wavelength / material, and when A is the wavelength, it reflects the size effect of the quantum dot; d is the sample thickness, and nQD is the quantum dot number density.

[0018] As a preferred technical solution of the present invention, the method of acoustic-impedance joint confidence evaluation is to construct a joint confidence index through the complementary characteristics of acoustic impedance Za and electrical impedance Ze: C joint = w a ·C a + w e ·C e - λ·C a ·C e ,

[0019] where in the above formula, C a is the acoustic unimodal confidence, C e is the electrical unimodal confidence, where wa + we = 1, and λ is the modal conflict correction factor;

[0020] The impedance confidence level is:

[0021]

[0022] In the above formula: Z' k , Z” k are respectively the real part and the imaginary part of the impedance in a single measurement.

[0023] μZ', μZ” are the mean values of the real part / imaginary part of the impedance of the reference, and N refers to the number of repeated measurements;

[0024] The acoustic confidence level is:

[0025]

[0026] In the above formula, when the change of the Q value conforms to the Gaussian distribution hypothesis, it represents the matching probability between the measurement result and the normal state. ΔQ is the absolute deviation between the measured Q value and the reference value, and σ_Q is the measurement standard deviation of the Q value.

[0027] As a preferred technical solution of the present invention, the method for constructing the decision tree model is as follows: First, detect the fluorescence intensity of quantum dots, and obtain the fluorescence intensity of quantum dots through the quantum dot response formula. If the fluorescence intensity of quantum dots is abnormal, then activate the acoustic-impedance joint confidence level evaluation. If the fluorescence intensity of quantum dots is normal, then mark it as normal. At the same time, for the second-layer joint confidence level, perform reinforcement learning maintenance, manual re-inspection, mark pseudo-defects, and dynamically update the decision weights.

[0028] As a preferred technical solution of the present invention, the method for judging the fluorescence intensity threshold of quantum dots is to perform fluorescence feature extraction and separate the background noise of the fluorescence features by using a double-exponential decay model: I(t) = A1e -t / τ1 + A2e -t / τ2 + B,

[0029] where if ∣A1 / A2 - μA∣ > 3σA or τ1 > 2τ0, then trigger the second layer.

[0030] A system for implementing the above method for locating faulty components of a screen module, characterized in that it includes a quantum dot spraying robot, a laser scanner, a hyperspectral camera, a variable-frequency ultrasonic transmitter, a laser Doppler vibrometer, and an industrial computer. The dot spraying robot, the laser scanner, the hyperspectral camera, the variable-frequency ultrasonic transmitter, and the laser Doppler vibrometer are all connected to the industrial computer;

[0031] The quantum dot spraying robot is used to spray the quantum dot solution onto the key nodes of the screen module to be detected;

[0032] The 405nm laser of the laser scanner scans the screen module to be detected, and the hyperspectral camera is used to capture the quantum dot responses of each key node; the variable-frequency ultrasonic transmitter is used to emit variable-frequency ultrasonic waves to the screen module, and the laser Doppler vibrometer collects the vibration spectrum of the screen module.

[0033] The beneficial effects of the present invention are:

[0034] 1. This method for locating faulty components of the screen module performs quantum marking pretreatment; sprays quantum dot solution on the key nodes of the screen module to be detected; captures the quantum dot responses of each key node through a hyperspectral camera to obtain the quantum dot response states of each key node; emits variable-frequency ultrasonic waves on the screen module, and then uses a laser Doppler vibrometer to collect the vibration spectrum of the screen module; and analyzes to obtain the acoustic impedance Za of the screen module; measures the grid impedance by the four-wire method, constructs a three-dimensional impedance vector field, and uses the electrical impedance Ze of the screen module measured by the three-dimensional impedance vector field. Combine the acoustic impedance Za and the electrical impedance Ze to obtain the acoustic-impedance combined confidence; the first layer is the judgment of the quantum dot fluorescence intensity threshold, the second layer is the evaluation of the acoustic-impedance combined confidence, and the third layer is the output of the reinforcement learning maintenance strategy to construct a decision tree; among them, using the characteristics of quantum dots, after a key node fails, it will cause the temperature of the key node to rise, which will change the quantum dot response states of each key node. Then, scan the screen module to be detected with a 405nm laser, and then capture the quantum dot responses of each key node through a hyperspectral camera. In this way, the fault point can be determined by capturing the quantum dot responses of each key node through the hyperspectral camera. Then, combine the acoustic impedance Za and the electrical impedance Ze to obtain the acoustic-impedance combined confidence to evaluate the fault type, and then execute reinforcement learning maintenance through the decision tree, conduct manual re-inspection, mark pseudo-defects, and dynamically update the decision weights. This is convenient for locating the faults of the screen module and giving specific maintenance plans.

[0035] 2. This method for locating faulty components of the screen module combines acoustic-impedance for confidence evaluation. Its advantages are: improving measurement accuracy and reliability, and integrating complementary information: Acoustic signals (such as vibration frequency, acoustic wave attenuation) reflect dynamic behavior, and impedance parameters (such as electrical impedance or acoustic impedance) characterize the electrical or mechanical properties of the medium. The combination of the two can cover more comprehensive system state information. Cross-validation mechanism: By cross-checking acoustic and impedance data, outliers caused by environmental interference can be identified, reducing the probability of misjudgment.

[0036] 3. The double-exponential model of the method for locating faulty components in a screen module can more accurately describe the complex fluorescence decay process by introducing two decay components, fast and slow, which are usually associated with the target fluorescence characteristics, thus effectively separating noise from the true signal and enhancing the anti-noise ability. Since the single-exponential model is sensitive to mixed noise (such as instrument noise and scattered light), it is prone to parameter estimation deviation. The double-exponential model can suppress the interference of high-frequency noise on the slow-decay signal and improve the signal-to-noise ratio by distinguishing the decay behaviors on different time scales. The double-exponential model can simultaneously extract two decay time constants (τ1, τ2) and their amplitude ratio (A1 / A2), providing more dimensional information for analyzing complex systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, but do not constitute a limitation to the present invention.

[0038] In the drawings:

[0039] Figure 1 is a flowchart of a method for locating faulty components in a screen module according to the present invention;

[0040] Figure 2 is a system block diagram of a system for locating faulty components in a screen module according to the present invention.

[0041] In the figure: 1. Quantum dot spraying robot; 2. Laser scanner; 3. Hyperspectral camera; 4. Variable-frequency ultrasonic transmitter; 5. Laser Doppler vibrometer; 6. Industrial computer. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following describes the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for explaining and illustrating the present invention, and are not used to limit the present invention.

[0043] Embodiment: As Figure 1 shown, a method for locating faulty components in a screen module according to the present invention includes the following steps:

[0044] Step 1. Perform quantum labeling pretreatment; spray quantum dot solution on the key nodes of the screen module to be detected;

[0045] Step 2. Perform optoelectronic excitation detection; scan the screen module to be detected with a 405 nm laser, and then capture the quantum dot responses of the key nodes at each driving IC / trace intersection through a hyperspectral camera to obtain the quantum dot response states of each key node, and then obtain the quantum dot fluorescence intensities of each key node;

[0046] Step 3: Then perform acoustic resonance imaging; emit variable-frequency ultrasonic waves to the screen module, and then use a laser Doppler vibrometer to collect the vibration spectrum of the screen module; and analyze to obtain the acoustic impedance Za of the screen module.

[0047] Step 4: Use the four-wire method to measure the grid impedance, construct a three-dimensional impedance vector field, measure the electrical impedance Ze of the screen module using the three-dimensional impedance vector field, and combine the acoustic impedance Za and the electrical impedance Ze to obtain an acoustic-impedance joint confidence model.

[0048] Step 5: Use the first layer to judge the quantum dot fluorescence intensity threshold for each key point, the second layer to perform acoustic-impedance joint confidence evaluation using the acoustic-impedance joint confidence model, and the third layer to output the maintenance strategy of reinforcement learning to construct a decision tree; use the decision tree to locate the faults of the screen module. Perform quantum labeling preprocessing; spray quantum dot solution on the key nodes of the screen module to be detected; capture the quantum dot responses of each key node through a hyperspectral camera to obtain the quantum dot response states of each key node; emit variable-frequency ultrasonic waves to the screen module, and then use a laser Doppler vibrometer to collect the vibration spectrum of the screen module; and analyze to obtain the acoustic impedance Za of the screen module; use the four-wire method to measure the grid impedance, construct a three-dimensional impedance vector field, measure the electrical impedance Ze of the screen module using the three-dimensional impedance vector field, and combine the acoustic impedance Za and the electrical impedance Ze to obtain the acoustic-impedance joint confidence; the first layer is the quantum dot fluorescence intensity threshold judgment, the second layer is the acoustic-impedance joint confidence evaluation, and the third layer is to output the maintenance strategy of reinforcement learning to construct a decision tree; among them, using the characteristics of quantum dots, after a key node fails, it will cause the temperature of the key node to rise, which will change the quantum dot response states of each key node, and then scan the screen module to be detected with a 405nm laser, and then capture the quantum dot responses of each key node through a hyperspectral camera. In this way, the fault point can be determined by capturing the quantum dot responses of each key node through the hyperspectral camera, and then combine the acoustic impedance Za and the electrical impedance Ze to evaluate the fault type of the acoustic-impedance joint confidence, and then perform reinforcement learning maintenance through the decision tree, manual re-inspection, mark pseudo-defects, and dynamically update the decision weights, so as to facilitate the location of the screen module faults and give specific maintenance plans.

[0049] Quantum dots are semiconductor materials at the nanoscale (usually 2 - 10 nanometers), and their electronic behavior is affected by the quantum confinement effect. By adjusting the size and composition, their optical and electrical properties can be precisely controlled. For example, the emission color changes from blue light to red light as the size increases, and generally, the increase in temperature leads to a decrease in the emission intensity of quantum dots. This is because the increase in temperature intensifies the lattice vibration (phonon scattering), prompting the electron-hole pairs to release energy through non-radiative recombination pathways rather than emitting light.

[0050] Emission wavelength shift: The bandgap energy of quantum dots varies with temperature. For example, when the temperature rises, the thermal expansion effect of the material causes a change in the lattice constant, resulting in a redshift of the emission wavelength (shift towards longer wavelengths). In this way, when a failure occurs at a critical point, the resistance increases, and heat is generated after power-on, which will reduce the luminescence intensity of the quantum dots. When the temperature rises, the thermal expansion effect of the material causes a change in the lattice constant, resulting in a redshift of the emission wavelength, thus realizing the positioning function of the failure point.

[0051] In the present invention, by utilizing the characteristics of quantum dots, after a failure in a critical section, the temperature of the critical node will increase, which causes a change in the response state of the quantum dots at each critical section. Then, a 405 nm laser is used to scan the screen module to be detected, and the quantum dot responses of each critical node are captured by a hyperspectral camera. In this way, the failure point can be determined by capturing the quantum dot responses of each critical node through the hyperspectral camera. Then, the acoustic impedance Za and the electrical impedance Ze are combined to obtain the acoustic-impedance joint confidence to evaluate the failure type, and then reinforcement learning repair is performed through a decision tree, followed by manual re-inspection, marking pseudo-defects, and dynamically updating the decision.

[0052] In the present invention, impedance measurement is performed by using the four-wire method to measure the grid impedance, which has high precision and anti-interference ability, and eliminates the influence of lead resistance: The four-wire method has independent current injection and voltage detection paths, avoiding the errors of contact resistance and wire resistance, and is especially suitable for the precise measurement of low-impedance (such as <1Ω) grid cells. Suppressing environmental noise: In grid measurement, high-frequency electromagnetic interference or ground loop noise can be reduced by differential voltage measurement and shielded wiring, improving the signal-to-noise ratio. Local anomaly localization: After dividing the target area into dense grids, the four-wire method can measure the impedance distribution point by point to accurately identify material defects. Anisotropic characterization: Grid data can be used to construct a two-dimensional / three-dimensional impedance tensor model to analyze the orientation-dependent conductive behavior of composite materials (such as carbon fiber reinforced plastics). Multiphysics coupling: Combining external field parameters such as temperature and pressure, the grid impedance can be related to the thermal-electro-mechanical coupling effect. Multimodal integration: Grid impedance data can be fused with acoustic and optical signals (such as acoustic impedance joint imaging) to improve the defect classification accuracy through machine learning algorithms.

[0053] The method of spraying quantum dot solution on the critical nodes of the screen module to be detected in step 1 is to configure quantum dot solutions with different emission wavelengths and coat them with quantum dot solutions with different emission wavelengths in different failure-sensitive areas; this facilitates the positioning of different areas.

[0054] A spatial coding strategy is adopted to locate each quantum dot position.

[0055] Among them, the quantum dot wavelength coding formula: λ QD = 520 + Δλ·(x + yMx )

[0056] In the above formula: λQD represents the fluorescence wavelength of the quantum dots, x represents the horizontal coordinate (0 ≤ x < M_x), y represents the vertical coordinate (y ≥ 0), and Mx represents the maximum number of horizontal grids, which is used to prevent wavelength aliasing.

[0057] In step 2, the quantum dot response formula is

[0058] I(λ, x, y) = I0e -α(λ)d · ηQ D,

[0059] where in the formula, I0 is the initial value of the incident light intensity, a(A) is the extinction coefficient dependent on wavelength / material. When A is the wavelength, it reflects the size effect of the quantum dots; d is the sample thickness, and nQD is the quantum dot number density, so that the corresponding intensity of each quantum dot can be obtained.

[0060] The method for evaluating the combined confidence of acoustic-impedance is to construct a combined confidence index C through the complementary characteristics of the acoustic impedance Za and the electrical impedance Ze: joint = w a ·C a + w e ·C e - λ·C a ·C e ,

[0061] where in the above formula, C a is the acoustic unimodal confidence, C e is the electrical unimodal confidence, where wa + we = 1, and λ is the modal conflict correction factor; in this way, the complementary characteristics of the acoustic impedance Za and the electrical impedance Ze are used to evaluate the confidence of the fault point, thus avoiding the inaccuracy of single-factor evaluation and ensuring better confidence evaluation accuracy.

[0062] The impedance confidence is:

[0063]

[0064] In the above formula: Z' k , Z” k are the real and imaginary parts of the impedance measured once respectively,

[0065] μZ', μZ” are the mean values of the real / imaginary parts of the reference impedance, and N refers to the number of repeated measurements;

[0066] The acoustic confidence is:

[0067]

[0068] When the Q-value change in the above formula conforms to the Gaussian distribution hypothesis, it represents the matching probability between the measurement result and the normal state. ΔQ is the absolute deviation between the measured Q-value and the reference value, and σ_Q is the measurement standard deviation of the Q-value.

[0069] In the present invention, acoustic-impedance combination is used for reliability evaluation. Its advantages are as follows: improving measurement accuracy and reliability, and integrating complementary information. Acoustic signals (such as vibration frequency, acoustic wave attenuation) reflect dynamic behavior, and impedance parameters (such as electrical impedance or acoustic impedance) characterize the electrical or mechanical properties of the medium. The combination of the two can cover more comprehensive system state information. Cross-validation mechanism: By cross-checking acoustic and impedance data, outliers caused by environmental interference can be identified, and the misjudgment probability can be reduced.

[0070] Enhancing adaptability to complex scenarios. Under extreme conditions such as high temperature and high humidity, impedance data can provide redundant support to ensure the continuous evaluation ability of the system. By algorithms such as Bayesian network or Kalman filter, the uncertainty distributions of acoustic and impedance are fused, and a joint probability model can be constructed to output a more robust confidence score.

[0071] Among them, the method for constructing the decision tree model is as follows: First, detect the fluorescence intensity of quantum dots, and determine the fluorescence intensity of quantum dots by obtaining the response time through the quantum dot response formula. If the fluorescence intensity of quantum dots is abnormal, activate the acoustic-impedance joint confidence evaluation; if the fluorescence intensity of quantum dots is normal, mark it as normal. At the same time, for the second-layer joint confidence, perform reinforcement learning maintenance, manual re-inspection, mark pseudo-defects, and dynamically update the decision weights.

[0072] The construction algorithm of the decision tree model is an existing technical algorithm, and algorithms such as ID3, C4.5, and CART can be used for construction. The specific algorithm formulas are not disclosed here, and only the following functional requirements need to be met. The decision tree model should first detect the fluorescence intensity of quantum dots, and determine the fluorescence intensity of quantum dots by obtaining the response time through the quantum dot response formula. If the fluorescence intensity of quantum dots is abnormal, activate the acoustic-impedance joint confidence evaluation; if the fluorescence intensity of quantum dots is normal, mark it as normal. At the same time, for the second-layer joint confidence, perform reinforcement learning maintenance, manual re-inspection, mark pseudo-defects, and dynamically update the decision weights.

[0073] The decision tree model is a simple and easy-to-use non-parametric classifier. It does not require any prior assumptions about the data, has a relatively fast calculation speed, easy-to-interpret results, and strong robustness. In complex decision-making situations, multi-level or multi-stage decisions are often required. After a stage of decision-making is completed, there may be m new and different natural states; under each natural state, there are m new strategies to choose from, and different results are produced after selection and a new natural state is faced again, continuing to generate a series of decision-making processes. This kind of decision-making is called sequential decision-making or multi-level decision-making. At this time, if the above decision-making criteria are continued to be followed or the benefit matrix is used to analyze the problem, it is easy to make the corresponding table relationships very complex. The decision tree is an effective tool that can help decision-makers conduct sequential decision-making analysis. Its method is to represent the strategies, natural states, probabilities, revenue values, etc. in the problem in a tree-like form through lines and graphics.

[0074] Among them, the method for judging the quantum dot fluorescence intensity threshold is to perform fluorescence feature extraction and use a double-exponential decay model to separate the background noise of the fluorescence features:

[0075] I(t) = A1e -t / τ1 + A2e -t / τ2 + B,

[0076] Among them, if |A1 / A2 - μA| > 3σA or τ1 > 2τ0, the second layer is triggered;

[0077] The double-exponential model can more accurately describe the complex fluorescence decay process by introducing two decay components, fast and slow. For example: the fast decay component τ1 may correspond to background noise or short-lived interference signals; the slow decay component τ2 is usually associated with the target fluorescence features, thus effectively separating noise from the true signal. It enhances the anti-noise ability because the single-exponential model is sensitive to mixed noise (such as instrument noise, scattered light), which is prone to causing parameter estimation deviations; the double-exponential model can suppress the interference of high-frequency noise on the slow decay signal and improve the signal-to-noise ratio by distinguishing the decay behaviors on different time scales. The double-exponential model can simultaneously extract two decay time constants (τ1, τ2) and their amplitude ratio (A1

[0078] / A2), providing more dimensional information for analyzing complex systems.

[0079] A system, as Figure 2 shown, is used to implement the above method for locating faulty components of a screen module, including a quantum dot spraying robot 1, a laser scanner 2, a hyperspectral camera 3, a variable-frequency ultrasonic transmitter 4, a laser Doppler vibrometer 5, and an industrial computer 6. The quantum dot spraying robot 1, the laser scanner 2, the hyperspectral camera 3, the variable-frequency ultrasonic transmitter 4, and the laser Doppler vibrometer 5 are all connected to the industrial computer 6;

[0080] The quantum dot spraying robot 1 is used to spray quantum dot solution onto the key nodes of the screen module to be detected;

[0081] The laser scanner 2 scans the screen module to be detected with 405nm laser. The hyperspectral camera 3 is used to capture the quantum dot response of each key node. The variable-frequency ultrasonic transmitter 4 emits variable-frequency ultrasonic waves to the screen module, and the laser Doppler vibrometer 5 collects the vibration spectrum of the screen module.

[0082] Working principle: Conduct quantum labeling pretreatment; spray quantum dot solution onto the key nodes of the screen module to be detected; capture the quantum dot response of each key node through the hyperspectral camera to obtain the quantum dot response state of each key node; emit variable-frequency ultrasonic waves on the screen module, and then use the laser Doppler vibrometer to collect the vibration spectrum of the screen module; and analyze to obtain the acoustic impedance Za of the screen module; measure the grid impedance by the four-wire method, construct a three-dimensional impedance vector field, and measure the electrical impedance Ze of the screen module by using the three-dimensional impedance vector field. Combine the acoustic impedance Za and the electrical impedance Ze to obtain the acoustic-impedance combined confidence level; the first layer is the judgment of the quantum dot fluorescence intensity threshold, the second layer is the evaluation of the acoustic-impedance combined confidence level, and the third layer is to output the maintenance strategy of reinforcement learning and construct a decision tree; among them, using the characteristics of quantum dots, after a key node fails, it will cause the temperature of the key node to rise, which will change the quantum dot response state of each key node. Then scan the screen module to be detected with 405nm laser, and then capture the quantum dot response of each key node through the hyperspectral camera. In this way, the fault point can be determined by capturing the quantum dot response of each key node through the hyperspectral camera. Then combine the acoustic impedance Za and the electrical impedance Ze to evaluate the fault type of the acoustic-impedance combined confidence level, and then perform reinforcement learning maintenance through the decision tree, manual re-inspection, mark pseudo-defects, and dynamically update the decision.

[0083] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for locating a faulty component of a screen module, characterized in that, It includes the following steps: Step 1: Perform quantum labeling preprocessing; spray quantum dot solution on the key nodes of the screen module to be detected; Step 2: Perform optoelectronic excitation detection; scan the screen module to be detected with a 405nm laser, and then capture the quantum dot response of the key nodes at each driving IC / trace intersection through a hyperspectral camera to obtain the quantum dot response state of each key node, and then obtain the quantum dot fluorescence intensity of each key node; Step 3: Then perform acoustic resonance imaging; emit variable-frequency ultrasonic waves to the screen module, and then use a laser Doppler vibrometer to collect the vibration spectrum of the screen module; and analyze to obtain the acoustic impedance Za of the screen module; Step 4: Use the four-wire method to measure the grid impedance, construct a three-dimensional impedance vector field, and use the electrical impedance Ze of the screen module measured by the three-dimensional impedance vector field to combine the acoustic impedance Za and the electrical impedance Ze to obtain an acoustic-impedance joint confidence model; Step 5: Use the first layer to judge the quantum dot fluorescence intensity threshold for each key point, the second layer to perform acoustic-impedance joint confidence evaluation with the acoustic-impedance joint confidence model, and the third layer to output the maintenance strategy of reinforcement learning to construct a decision tree model; use the decision tree model to locate the faults of the screen module.

2. The method for locating a faulty component of a screen module according to claim 1, wherein The method of spraying quantum dot solution on the key nodes of the screen module to be detected in Step 1 is to configure quantum dot solutions with different emission wavelengths and coat them with quantum dot solutions with different emission wavelengths in different fault-sensitive areas; Adopt a spatial encoding strategy to locate each quantum dot position, Among them, the quantum dot wavelength encoding formula: λ QD = 520 + Δλ·(x + yM x ) In the above formula: λQD represents the fluorescence wavelength of the quantum dot, x represents the horizontal coordinate (0≤x<M_x), y represents the vertical coordinate (y≥0), and Mx represents the maximum number of horizontal grids, which is used to prevent wavelength aliasing.

3. The method for locating a faulty component of a screen module according to claim 1, characterized in that, The quantum dot response formula in Step 2 is I(λ, x, y) = I0e -α(λ)d ·η QD , Where I0 in the formula is the initial value of the incident light intensity, a(A) is the extinction coefficient dependent on wavelength / material. When A is the wavelength, it reflects the size effect of the quantum dot; d is the sample thickness, and nQD is the quantum dot number density.

4. A method for locating a faulty component of a screen module according to claim 2, characterized in that, The method of the acoustic-impedance joint confidence evaluation is to construct a joint confidence index through the complementary characteristics of the acoustic impedance Za and the electrical impedance Ze: C joint = w a · C a + w e · C e - λ · C a · C e , Among them, in the above formula, C a is the acoustic unimodal confidence, and C e is the electrical unimodal confidence, where wa + we = 1, and λ is the modal conflict correction factor; The impedance confidence is: In the above formula: Z' k , Z” k are the real part and the imaginary part of the impedance for a single measurement respectively, μZ', μZ” are the mean values of the real part / imaginary part of the reference impedance, and N refers to the number of repeated measurements; The acoustic confidence is: In the above formula, when the change of the Q value conforms to the Gaussian distribution hypothesis, it represents the matching probability between the measurement result and the normal state. ΔQ is the absolute deviation between the measured Q value and the reference value, and σ_Q is the measurement standard deviation of the Q value.

5. A method for locating a faulty component of a screen module according to claim 1, wherein, The method of constructing the decision tree model is to first detect the quantum dot fluorescence intensity, and judge the quantum dot fluorescence intensity through the quantum dot response formula. Among them, if the quantum dot fluorescence intensity is abnormal, the acoustic-impedance joint confidence evaluation is activated. If the quantum dot fluorescence intensity is normal, it is marked as normal. At the same time, for the second-layer joint confidence, perform reinforcement learning maintenance, manual re-inspection, mark pseudo-defects, and dynamically update the decision weights.

6. The method for locating a faulty component of a screen module according to claim 1, wherein The method for judging the quantum dot fluorescence intensity threshold is to perform fluorescence feature extraction and use a double-exponential decay model to separate the background noise of the fluorescence features: I(t) = A1e -t / τ1 + A2e -t / τ2 + B, If ∣A1 / A2 - μA∣ > 3σA or τ1 > 2τ0, the second layer is triggered.

7. A system for implementing a method for locating a faulty component of a screen module as described in any one of claims 1-6, characterized in that, It includes a quantum dot spraying robot (1), a laser scanner (2), a hyperspectral camera (3), a variable-frequency ultrasonic transmitter (4), a laser Doppler vibrometer (5) and an industrial computer (6). The quantum dot spraying robot (1), the laser scanner (2), the hyperspectral camera (3), the variable-frequency ultrasonic transmitter (4) and the laser Doppler vibrometer (5) are all connected to the industrial computer (6); The quantum dot spraying robot (1) is used to spray the quantum dot solution onto the key nodes of the screen module to be detected; The laser scanner (2) uses 405 nm laser to scan the screen module to be detected. The hyperspectral camera (3) is used to capture the quantum dot response of each key node. The variable-frequency ultrasonic transmitter (4) is used to emit variable-frequency ultrasonic waves to the screen module, and the laser Doppler vibrometer (5) collects the vibration spectrum of the screen module.

Images