Product production quality inspection management method and system based on quantum sensing
By using quantum sensing arrays and intelligent algorithms, the problems of detection accuracy and control delay in microscopic defects in industrial product quality inspection have been solved, achieving efficient and real-time quality inspection management.
Patent Information
- Application Number
- PCT/CN2026/078532
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-06
- Filing Date
- 2026-02-11
- Publication Date
- 2026-11-12
AI Technical Summary
Existing technologies for industrial product quality inspection suffer from problems such as insufficient accuracy in detecting micro-defects, severe environmental noise interference, limited visualization of test results, and delayed control feedback, making it difficult to achieve efficient and real-time quality inspection management.
Multidimensional field data is collected using a quantum sensing array. Spatiotemporal correlation feature matrix is extracted through quantum state decoupling processing. Combined with multi-scale convolutional neural network and quantum annealing algorithm, a visual heat map is generated and feedback control commands are triggered to form a closed loop of the entire process.
It enables precise capture of microscopic defects inside materials, reduces the risk of false positives and false negatives, improves quality inspection efficiency and process control precision, and forms a quality inspection closed loop with a response time of seconds.
Smart Images

Figure CN2026078532_12112026_PF_FP_ABST
Abstract
Description
A product manufacturing quality inspection management method and system based on quantum sensing Technical Field
[0001] This invention relates to the fields of industrial automation and quantum detection technology, and in particular to a product manufacturing quality inspection management method and system based on quantum sensing. Background Technology
[0002] In the industrial product manufacturing process, quality inspection is a core link in ensuring product consistency and reliability. Traditional quality inspection technologies mainly rely on optical imaging, ultrasonic testing, or electromagnetic sensing, but these methods have significant limitations in detecting microscopic defects. Optical testing is limited by wavelength resolution, making it difficult to capture atomic-level lattice distortions or subsurface microcracks within materials; ultrasonic technology lacks accuracy for complex structures and is easily affected by environmental noise; electromagnetic sensing cannot effectively identify the microscopic characteristics of non-metallic materials. Furthermore, existing inspection systems often use fixed thresholds to determine defects, failing to adapt to dynamic fluctuations in parameters such as temperature and pressure during production, resulting in high false positive and false negative rates. As product precision increases, traditional methods are insufficient in processing multi-physics coupled data, making it difficult to establish an accurate mapping relationship between high-dimensional features and quality status.
[0003] At the data processing and decision-making level, existing technologies rely on traditional machine learning models or manual experience rules, lacking efficient means to analyze the high-dimensional heterogeneous data generated by quantum sensing. Most systems employ single-scale feature extraction strategies, failing to simultaneously capture the spatial correlation between microscopic defects and macroscopic anomalies, and the feature fusion process carries the risk of information loss. Furthermore, the visualization of quality inspection results is limited, primarily consisting of two-dimensional grayscale images or numerical reports, making it difficult for operators to quickly locate defects and assess their severity, resulting in low efficiency for manual re-inspection. In the control feedback loop, traditional systems lack real-time closed-loop mechanisms, leading to significant delays between inspection results and production line equipment adjustments. Process parameter calibration relies on manual experience, hindering precise adaptive control.
[0004] While quantum sensing has demonstrated high-precision detection potential in laboratory environments, its industrial applications still face multiple challenges. Quantum signals are susceptible to environmental electromagnetic interference, resulting in low signal-to-noise ratios. Insufficient timing synchronization accuracy during multi-sensor collaborative acquisition affects the extraction of spatiotemporal features. Furthermore, the integration of quantum data with traditional control systems faces technological barriers, making it difficult to construct a complete closed loop from detection to execution. These issues severely restrict the large-scale application of quantum sensing technology in industrial production, highlighting the urgent need for a comprehensive system solution integrating high-precision detection, intelligent decision-making, and real-time control. Summary of the Invention
[0005] To address the technical problems in the prior art, this invention provides a product manufacturing quality inspection management method and system based on quantum sensing.
[0006] The technical solution provided by this invention is as follows:
[0007] First aspect:
[0008] This invention provides a product manufacturing quality inspection management method based on quantum sensing, comprising:
[0009] S1. Real-time acquisition of multi-dimensional field data of the target product through a quantum sensing array, wherein the multi-dimensional field data includes quantum phase difference spectrum, spin relaxation time distribution and energy density waveform;
[0010] S2. Perform quantum state decoupling processing on the multidimensional field data and extract the spatiotemporal correlation feature matrix. The spatiotemporal correlation feature matrix contains two core parameters: quantum hue consistency coefficient (QCHC) and structural anomaly index (SAI).
[0011] S3. Construct a dynamic quality assessment model by inputting the spatiotemporal correlation feature matrix into a pre-trained multi-scale convolutional neural network MSCNN and outputting a three-dimensional quality feature tensor.
[0012] S4. Optimize the quality decision boundary based on the quantum annealing algorithm, and dynamically adjust the defect judgment threshold according to real-time production parameters;
[0013] S5. Generate a visual quality heat map and trigger feedback control instructions, which include process parameter adjustment amounts and equipment calibration vectors.
[0014] The second aspect:
[0015] This invention provides a product manufacturing quality inspection management system based on quantum sensing, comprising:
[0016] A quantum sensing module with a multi-physics field synchronous acquisition device; an edge computing unit integrating quantum noise suppression and feature extraction accelerators; and a dynamic decision engine based on a quantum-inspired optimization algorithm library.
[0017] The beneficial effects of the technical solution provided by this invention include at least the following:
[0018] (1) In this invention, multidimensional quantum field data is collected by a quantum sensing array, and combined with the high sensitivity of the diamond NV color center sensor, the physical limits of traditional detection technology are broken through, and microscopic defects such as atomic-level lattice distortion and microcracks inside the material are accurately captured. The quantum state decoupling processing technology effectively separates environmental noise from the target signal, extracts the quantum hue consistency coefficient and structural anomaly index, significantly improves the spatial resolution and reliability of defect identification, and solves the problem of insufficient detection capability of traditional methods for microscopic features;
[0019] (2) In this invention, a dynamic quality decision boundary is constructed based on the quantum annealing algorithm, and the production process parameters are encoded as the barrier function in the quantum optimization model to analyze the energy state of the quality feature vector in real time. When the fluctuation of the production environment causes feature shift, the system automatically adjusts the defect judgment threshold to achieve intelligent matching between the detection standard and the real-time working condition, overcomes the adaptability defects of the fixed threshold system in dynamic production scenarios, and greatly reduces the risk of false detection and missed detection;
[0020] (3) In this invention, a multi-dimensional interactive visualization heatmap of hue, brightness, and saturation is generated by mapping the three-dimensional quality feature tensor to the HSV color space, which intuitively presents the distribution and severity of defects. Combined with the augmented reality interface and the real-time feedback control module, the response time from the detection result to the equipment calibration command is reduced to the second level, forming a closed loop of the entire process of "perception-analysis-decision-execution", which significantly improves the efficiency of quality inspection and the accuracy of process control, and solves the technical bottlenecks of lagging human-computer interaction and isolated control in traditional systems. Attached Figure Description
[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 is a flowchart illustrating a product manufacturing quality inspection management method based on quantum sensing provided in an embodiment of the present invention;
[0023] Figure 2 is a schematic diagram of the quantum state decoupling process of a product manufacturing quality inspection management method based on quantum sensing provided in an embodiment of the present invention;
[0024] Figure 3 is a schematic diagram of the multi-scale convolutional neural network implementation process of a product production quality inspection management method based on quantum sensing provided in an embodiment of the present invention.
[0025] Figure 4 is a schematic diagram of the barrier function construction and dynamic threshold control process of a product production quality inspection management method based on quantum sensing provided in an embodiment of the present invention.
[0026] Figure 5 is a schematic diagram of the relevant process of a visualized quality heat map of a product production quality inspection management method based on quantum sensing provided in an embodiment of the present invention;
[0027] Figure 6 is a schematic diagram of the visualization quality heat map generation and feedback control command triggering process of a product production quality inspection management method based on quantum sensing provided in an embodiment of the present invention. Detailed Implementation
[0028] The technical solution of the present invention will now be described with reference to the accompanying drawings.
[0029] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.
[0030] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.
[0031] In embodiments of the present invention, sometimes the subscript is as follows: It may be mistakenly written as a non-subscript form such as W1. When the distinction is not emphasized, the meaning they express is the same.
[0032] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.
[0033] Referring to Figure 1 in the specification, a flowchart illustrating a product manufacturing quality inspection management method based on quantum sensing provided by an embodiment of the present invention is shown.
[0034] This invention provides a product manufacturing quality inspection management method based on quantum sensing. This method can be implemented by a product manufacturing quality inspection management device based on quantum sensing, which can be a terminal or a server. The processing flow of this quantum sensing-based product manufacturing quality inspection management method may include the following steps:
[0035] S1. Real-time acquisition of multi-dimensional field data of the target product through a quantum sensing array, wherein the multi-dimensional field data includes quantum phase difference spectrum, spin relaxation time distribution and energy density waveform.
[0036] Understandably, the core reason for using quantum sensing arrays to acquire multidimensional field data (quantum phase difference spectrum, spin relaxation time distribution, energy density waveform) in real time is that traditional sensors cannot capture microscopic quantum state characteristics, while quantum sensing technology can break through the limits of classical physics, achieving non-destructive detection of deep-seated characteristics such as the internal lattice structure and energy distribution of materials. This step provides a high-dimensional raw data foundation for quality inspection: the quantum phase difference spectrum reflects the consistency of atomic arrangement, the spin relaxation time reveals the lifetime characteristics of defect states, and the energy density waveform characterizes microscopic energy dissipation modes. Logically, this step constitutes the input end of the technology chain, providing the physical basis for subsequent feature extraction and decision analysis.
[0037] S2. Perform quantum state decoupling processing on the multidimensional field data and extract the spatiotemporal correlation feature matrix. The spatiotemporal correlation feature matrix contains two core parameters: quantum hue consistency coefficient (QCHC) and structural anomaly index (SAI).
[0038] Understandably, the necessity of implementing quantum state decoupling to extract the spatiotemporal correlation feature matrix stems from the strong coupling characteristics of quantum sensing data—environmental noise and multi-physics interference cause the original signal to become mixed. After separating the eigenstate signals through the quantum entanglement decoupling equation, spatiotemporal correlation analysis is used to extract two parameters: QCHC (quantum surface uniformity) and SAI (characterizing the degree of structural distortion). This transforms the raw quantum data into computable engineering features. QCHC detects microscopic color differences through quantum phase consistency, while SAI, combined with spatiotemporal gradients, identifies latent defects. Logically, this step is the pivotal point for data-to-information transformation, providing feature input for the quality assessment model.
[0039] Furthermore, as shown in Figure 2, the quantum state decoupling process is as follows:
[0040] S201, Signal Preprocessing:
[0041] S2011. The original signal is decomposed into 5 levels using the Daubechies 9 wavelet basis. The decomposition formula is as follows:
[0042] ;
[0043] in, Denotes the approximation coefficients of the k-th layer. Represents the detail coefficients of the k-th layer;
[0044] S2012. Calculate the time offset between multiple sensors using a cross-correlation algorithm:
[0045] ;
[0046] in, and represents the raw signals from the two sensors, and represents the time delay variable.
[0047] S202, Quantum Channel Separation:
[0048] S2021. Constructing the total Hamiltonian model:
[0049] ;
[0050] in, Denotes the intrinsic Hamiltonian of the target system. The environmental noise interference term is represented by the eigenstates separated by the variable quantum characteristic solver (VQE). The number of iterations is set to 200, and the convergence tolerance is 1e-6.
[0051] S2022. Perform a windowed Fourier transform on the decoupled signal:
[0052] ;
[0053] in, This is a window function with a window length of 10ms and a Hanning window type.
[0054] S203, Spatiotemporal Feature Fusion:
[0055] S2031. Calculate the quantum coherence of adjacent sensor nodes:
[0056] ;
[0057] in, Represents the quantum state of the i-th sensor. It is the inner product of two quantum states;
[0058] S2032. Construct the Markov state transition matrix:
[0059] ;
[0060] in, This represents the number of times a state transitions from state m to state n.
[0061] S3. Construct a dynamic quality assessment model by inputting the spatiotemporal correlation feature matrix into a pre-trained multi-scale convolutional neural network MSCNN, and outputting a three-dimensional quality feature tensor.
[0062] Understandably, a dynamic quality assessment model is constructed using a multi-scale convolutional neural network (MSCNN) to process the spatiotemporally correlated feature matrix. Traditional single-scale models struggle to simultaneously capture microscopic defects (such as lattice dislocations) and macroscopic anomalies (such as surface cracks). MSCNN, however, extracts atomic-level features through 1×1 quantum convolution kernels and captures millimeter-level defect distributions through 3×3 spatiotemporal convolution kernels. It then fuses multi-scale information via an attention mechanism, ultimately outputting a three-dimensional quality feature tensor. Its purpose is to establish a mapping relationship between quantum features and quality levels; the three-dimensional tensor (spatial coordinates × defect type × confidence level) comprehensively characterizes the product's state. Logically, this step is the core decision-making unit of the technical solution, enabling intelligent mapping from features to quality criteria.
[0063] S4. Optimize the quality decision boundary based on the quantum annealing algorithm and dynamically adjust the defect judgment threshold according to real-time production parameters.
[0064] Understandably, step S4 optimizes the quality decision boundary based on the quantum annealing algorithm because traditional static thresholds cannot adapt to the dynamic fluctuations of production parameters (such as temperature and pressure). By encoding quality characteristics as qubit spin states and embedding process constraints into the Hamiltonian, the quantum tunneling effect is used to cross local optima and dynamically adjust the defect judgment threshold. Its function is to achieve adaptive optimization of defect detection; when the production environment changes, the decision boundary automatically drifts with the energy ground state. Logically, this step constitutes the dynamic adjustment center of quality control, ensuring that the detection standards strictly match the real-time operating conditions.
[0065] S5. Generate a visual quality heat map and trigger feedback control instructions, which include process parameter adjustment amounts and equipment calibration vectors.
[0066] Understandably, the design motivation for generating a visual quality heatmap and triggering feedback control commands is to address the problems of poor interpretability and high control latency in traditional quality inspection results. By mapping the three-dimensional quality tensor to the HSV color space (H maps uniformity deviation, S maps defect probability, and V maps energy density), the heatmap can intuitively locate the defect location and severity. Simultaneously, based on the response relationship between the quality feature tensor and the process parameter space, equipment calibration vectors (such as robotic arm pose correction) are calculated. This forms a closed loop of "inspection-display-control," with the heatmap guiding manual re-inspection and control commands driving the production line's self-adjustment. Logically, this step is the output terminal of the technical solution, completing the final link in quality control.
[0067] In one possible implementation, S2 specifically includes:
[0068] The quantum hue consistency coefficient QCHC is calculated using the following formula:
[0069] ;
[0070] in, Let be the mean of the quantum phase shift of color channel c. This represents the maximum standard deviation of the three channels.
[0071] In one possible implementation, S2 specifically includes:
[0072] The structural anomaly index (SAI) is calculated as follows:
[0073] ;
[0074] in, The second derivative norm of the quantum gradient field is represented. This refers to the energy fluctuation over time. , These represent the standard deviations of the spatial gradient and the temporal fluctuation, respectively.
[0075] In one possible implementation, S3 specifically includes:
[0076] The multi-scale convolutional neural network includes:
[0077] A 1×1 quantum convolution kernel used to extract microscopic defect features;
[0078] 3×3 spatiotemporal convolution kernels used to capture macroscopic anomalies;
[0079] Multi-scale features are dynamically fused through an attention mechanism.
[0080] Furthermore, as shown in Figure 3, the implementation details of the multi-scale convolutional neural network are as follows:
[0081] S301. Network Architecture Implementation: A three-level feature processing architecture is adopted to achieve full-scale detection from microscopic defect capture to macroscopic anomaly recognition.
[0082] S3011. Intelligent compression and recombination of feature channels are implemented using 1×1 convolutional kernels. The input layer receives a 64-channel feature map from the quantum decoupling module, and the number of channels is compressed to 16 dimensions through a trainable weight matrix, retaining key quantum features while removing redundant information. A channel shuffling mechanism is introduced, and after each convolution operation, the output channels are grouped and permuted to enhance the information interaction capability between different feature channels.
[0083] S3012. Dynamically evaluate the signal quality of each channel and randomly deactivate low-quality channels with a signal-to-noise ratio below 20dB. Specifically, during the training phase, the output of this channel is masked with a 50% probability, forcing the network to learn robust feature representations.
[0084] S302, Macroscopic Feature Extraction Layer:
[0085] The S3021 employs a 3×3 convolution kernel combined with periodic boundary condition processing to effectively capture spatial correlation features across sensor nodes. In the temporal dimension, a spatiotemporal cube is constructed by stitching together cross-frame features to identify anomalous propagation patterns of energy fluctuations. Dilated convolution technology is deployed with a dilation rate of 2, expanding the receptive field to three times that of traditional convolution without increasing computational load, ensuring sensitive detection of large-area defects.
[0086] S3022. Implement a grouped convolution strategy, dividing the input channel into 4 independent processing groups. Each group performs convolution operations in parallel before feature fusion, reducing the computational load to 25% of traditional methods. Employ dynamic precision quantization technology, using 8-bit fixed-point numbers to represent low-frequency components in the feature map, while retaining 16-bit floating-point precision for high-frequency components, reducing memory usage by 40% while maintaining accuracy.
[0087] S303, Multi-scale Feature Fusion:
[0088] S3031. A dual-path attention mechanism is designed. Path one generates channel attention weights using global average pooling, while path two generates a spatial attention mask using spatial pyramid pooling. The two are superimposed to form a dynamic feature selection matrix, guiding the weighted fusion of micro and macro features. Residual connections are introduced during the fusion process, directly injecting 30% of the original input features into the final output to avoid deep feature degradation.
[0089] In one possible implementation, S4 optimizes the decision boundary through the quantum tunneling effect, specifically including:
[0090] Production parameter constraints are encoded as barrier functions, triggering threshold adjustments when the quality feature vector crosses the barrier region.
[0091] Furthermore, as shown in Figure 4, the process of constructing the barrier function and controlling the dynamic threshold is as follows:
[0092] S401. Automatically extract key process parameters from the production line control system, including variables that directly affect product quality such as temperature, pressure, and flow rate. The screening criteria are that the parameter fluctuation exceeds the historical normal fluctuation range and has a strong correlation with past quality defect records. For example, when the injection molding machine pressure value continuously exceeds the statistical range of qualified samples, this parameter will be marked as an object requiring monitoring.
[0093] S402. The selected process parameters' allowable operating range is transformed into a constraint boundary in the quality feature space. In a high-dimensional space composed of quantum features, a spherical safety region is delineated centered on the feature vector distribution of historical qualified products. The radius of the region is dynamically adjusted according to the dispersion of qualified samples to ensure coverage of normal production fluctuations.
[0094] S403. An energy barrier mechanism is set up outside the safe zone. When the mass feature vector detected in real time approaches or exceeds the constraint boundary, the system automatically increases the energy level of that area. This energy barrier forms a "peak" structure similar to a topographic map in the feature space, and the energy value increases more dramatically the closer it is to the danger zone.
[0095] S404. The current position of the quality feature vector on the energy topography map is calculated in real time, and the overall energy level is assessed in conjunction with the structural anomaly index. When the energy value exceeds the dynamically calculated warning threshold for multiple consecutive detection cycles, it is determined to be a valid barrier crossing event. The warning threshold is automatically updated based on the recent energy mean and standard deviation to avoid environmental noise interference.
[0096] S405. Generate threshold adjustment instructions based on the direction and distance of the feature vector deviating from the safe zone. If the vector continues to shift towards the high-temperature sensitive side, the defect judgment threshold in the high-temperature direction is correspondingly relaxed; if the vector suddenly jumps into the high-pressure danger zone, the detection sensitivity in that area is temporarily increased. The adjustment range is logarithmically related to the offset distance, ensuring fine-tuning for small fluctuations and rapid response for large deviations.
[0097] S406. A smoothing filter technique is employed to adjust the threshold, avoiding misjudgments caused by sudden changes in detection standards. By setting a transition time constant, the new threshold gradually approaches the target value within a preset time window, while retaining a priority channel for manual adjustment. The adjusted threshold is synchronized to all detection nodes in real time, and the execution effect is verified through closed-loop feedback.
[0098] In one possible implementation, S5 includes:
[0099] The visualized quality heatmap shows:
[0100] Hue component mapping surface uniformity deviation value;
[0101] The luminance component is positively correlated with the energy density distribution;
[0102] The saturation component reflects the probability intensity of defects.
[0103] Furthermore, as shown in Figure 5, the relevant process for visualizing the quality heatmap is as follows:
[0104] S501. Establish a mapping rule between surface uniformity deviation value and hue component. In the HSV color model, set the reference hue of qualified products to green (120 degrees). As the deviation value increases, the hue gradually changes in a clockwise direction. Areas with severe deviations are displayed as red (0 degrees), and areas with moderate deviations transition to yellow (60 degrees), forming an intuitive hue warning system.
[0105] S502: Normalize the energy density data collected by the quantum sensing array, and nonlinearly map the energy value to the brightness channel through gamma correction. High energy density areas are displayed in bright colors, while low energy areas remain dark. At the same time, dynamic range compression is implemented to prevent local overexposure.
[0106] S503. Based on the confidence level of the defect probability output by the multi-scale convolutional neural network, control the intensity of the saturation component. When the defect probability is below 5%, use a low-saturation, soft display. Within the probability range of 5%-95%, the saturation increases linearly. When it exceeds 95%, switch to a high-saturation, flashing warning mode.
[0107] S504 integrates hue, brightness, and saturation three-dimensional data to generate a pseudo-color heatmap with superimposed spatial coordinate information. Subpixel anti-aliasing is implemented in the rendering engine to ensure clear color boundaries at different resolutions, and flickering interference caused by instantaneous noise is eliminated through a time integration algorithm.
[0108] S505: Deploys a dynamic visual optimization mechanism that automatically enhances the color contrast of a specific area and overlays contour lines when the operator focuses on that area. The system continuously monitors the position of the user's gaze point and performs real-time resolution enhancement processing on key observation areas.
[0109] S506. The final generated heat map is overlaid with the 3D model of the production equipment, achieving a seamless display of virtual and real elements in the augmented reality interface. Defect areas are automatically associated with process parameter adjustment suggestions; clicking on a color block on the heat map retrieves raw sensor data such as the quantum phase difference spectrum for that location.
[0110] In one possible implementation, a defect evolution prediction model is established by performing source analysis on historical quality data through a time-reversal algorithm.
[0111] Furthermore, as shown in Figure 6, the process for generating the visualized quality heatmap and triggering feedback control commands is as follows:
[0112] S601: Receive the three-dimensional quality feature tensor input and establish a mapping relationship between its spatial dimension and the HSV color space. Normalize the surface uniformity deviation value of each detection point, and form a hue component reference layer by linearly varying the hue value from green (120°) to red (0°) according to the degree of deviation from low to high.
[0113] S602. Analyze the energy density distribution data and convert the energy values at each point into luminance components according to a logarithmic relationship. Enable dynamic luminance suppression in areas with energy density higher than 200% of the average level to prevent local overexposure; enhance shadow details in areas with energy density lower than 50% to ensure visibility in low-energy areas.
[0114] S603. Calculate the saturation component based on the defect probability intensity. When the probability is less than 5%, use a base saturation of 20%. For every 1% increase in probability, the saturation increases by 2%. When the probability exceeds 95%, enable the high saturation pulse flashing mode. The flashing frequency is positively correlated with the severity of the defect.
[0115] The S604 system integrates hue, brightness, and saturation three-dimensional data to generate an initial heat map. It then uses a spatial interpolation algorithm to eliminate gaps between sensor points and employs edge sharpening technology to enhance the display of defect boundaries. Finally, it outputs a pseudo-color image with a resolution of no less than 4K.
[0116] S605. Based on the defect distribution characteristics in the heat map, calculate the adjustment amount of process parameters. For continuous large-area defect areas, generate instructions to synchronously reduce equipment pressure and temperature; for discrete point defects, trigger the calculation of the robotic arm pose calibration vector with an accuracy requirement of ±0.1mm.
[0117] S606. Establish a feedback control instruction priority mechanism, setting critical defect instructions that directly affect product safety as the highest priority (response delay <50ms), and appearance defect instructions as medium priority (response delay <200ms). Automatically attach diagnostic log records for historically recurring defects.
[0118] In one possible implementation, the quantum sensing array employs a diamond NV color center sensor with a sensitivity that satisfies:
[0119] ;
[0120] in, For the accuracy of magnetic field detection, is Planck's constant. It is the gyromagnetic ratio. For coherent time, This represents the number of effective sensing units.
[0121] In one possible implementation, the feedback control command generation includes:
[0122] The quality feature tensor is reduced in dimension and mapped to the process parameter space to calculate the parameter adjustment vector:
[0123] ;
[0124] in, For the pre-trained process response matrix, For real-time quality vectors, This is the standard mass vector.
[0125] Referring to Figure 2 in the specification, a schematic diagram of a product production quality inspection management system based on quantum sensing provided in an embodiment of the present invention is shown.
[0126] This invention also provides a product manufacturing quality inspection management system based on quantum sensing, applied to the aforementioned product manufacturing quality inspection management method based on quantum sensing, comprising:
[0127] A quantum sensing module with a multi-physics field synchronous acquisition device; an edge computing unit integrating quantum noise suppression and feature extraction accelerators; and a dynamic decision engine based on a quantum-inspired optimization algorithm library.
[0128] The system comprises a quantum sensing module, an edge computing unit, a dynamic decision engine, a human-machine interface terminal, and a feedback control module. The quantum sensing module, equipped with a diamond NV color center sensor and a superconducting quantum interference device, is deployed at the production line's inspection nodes. It acquires the quantum phase difference spectrum, spin relaxation time distribution, and energy density waveform of the target product in real time via a multi-physics synchronous acquisition device. The raw data is transmitted to the edge computing unit via a fiber optic network. The edge computing unit integrates a quantum noise suppression unit and a feature extraction accelerator to perform noise reduction and spatiotemporal feature extraction on the sensing data, generating a spatiotemporal correlation feature matrix containing quantum hue consistency coefficients and structural anomaly indices. The processed data is then transmitted to the dynamic decision engine via an industrial communication protocol.
[0129] The dynamic decision engine deploys a multi-core heterogeneous computing architecture and incorporates a quantum-inspired optimization algorithm library. After receiving the spatiotemporal correlation feature matrix, it calculates the deviation between the quality feature vector and the standard vector using a pre-trained process response matrix. It then optimizes the decision boundary using a quantum annealing algorithm and generates a dynamic defect judgment threshold. Simultaneously, it outputs process parameter adjustments and equipment calibration vectors to the feedback control module via a real-time operating system. The feedback control module includes an industrial communication protocol stack, directly connecting to the production line control equipment. It converts calibration commands into executable signals to drive robotic arm pose correction or process parameter adjustment, forming a closed-loop detection-control chain.
[0130] The human-machine interface terminal supports augmented reality display and multi-terminal collaborative operation. After receiving the 3D quality feature tensor output by the dynamic decision engine, it maps it to the HSV color space to generate a visualized quality heatmap. The hue component reflects the surface uniformity deviation, the brightness component is related to the energy density distribution, and the saturation component indicates the probability intensity of defects. Operators can view defect location information in real time through the interactive interface and manually intervene in higher-priority control commands. Data synchronization between system modules is achieved through a distributed storage array. The quantum sensing module and the edge computing unit use a hard real-time communication link, and the dynamic decision engine and the feedback control module are connected through a low-latency industrial network to ensure end-to-end collaboration from data acquisition to control execution.
[0131] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:
[0132] (1) In this invention, multidimensional quantum field data is collected by a quantum sensing array, and combined with the high sensitivity of the diamond NV color center sensor, the physical limits of traditional detection technology are broken through, and microscopic defects such as atomic-level lattice distortion and microcracks inside the material are accurately captured. The quantum state decoupling processing technology effectively separates environmental noise from the target signal, extracts the quantum hue consistency coefficient and structural anomaly index, significantly improves the spatial resolution and reliability of defect identification, and solves the problem of insufficient detection capability of traditional methods for microscopic features;
[0133] (2) In this invention, a dynamic quality decision boundary is constructed based on the quantum annealing algorithm, and the production process parameters are encoded as the barrier function in the quantum optimization model to analyze the energy state of the quality feature vector in real time. When the fluctuation of the production environment causes feature shift, the system automatically adjusts the defect judgment threshold to achieve intelligent matching between the detection standard and the real-time working condition, overcomes the adaptability defects of the fixed threshold system in dynamic production scenarios, and greatly reduces the risk of false detection and missed detection;
[0134] (3) In this invention, a multi-dimensional interactive visualization heatmap of hue, brightness, and saturation is generated by mapping the three-dimensional quality feature tensor to the HSV color space, which intuitively presents the distribution and severity of defects. Combined with the augmented reality interface and the real-time feedback control module, the response time from the detection result to the equipment calibration command is reduced to the second level, forming a closed loop of the entire process of "perception-analysis-decision-execution", which significantly improves the efficiency of quality inspection and the accuracy of process control, and solves the technical bottlenecks of lagging human-computer interaction and isolated control in traditional systems.
[0135] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
[0136] The following points need to be explained:
[0137] (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.
[0138] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the invention, i.e., these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element or there may be intermediate elements.
[0139] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.
[0140] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A product manufacturing quality inspection management method based on quantum sensing, characterized in that, include: S1. Real-time acquisition of multi-dimensional field data of the target product through a quantum sensing array, wherein the multi-dimensional field data includes quantum phase difference spectrum, spin relaxation time distribution and energy density waveform; S2. Perform quantum state decoupling processing on the multidimensional field data and extract the spatiotemporal correlation feature matrix. The spatiotemporal correlation feature matrix contains two core parameters: quantum hue consistency coefficient (QCHC) and structural anomaly index (SAI). S3. Construct a dynamic quality assessment model by inputting the spatiotemporal correlation feature matrix into a pre-trained multi-scale convolutional neural network MSCNN and outputting a three-dimensional quality feature tensor. S4. Optimize the quality decision boundary based on the quantum annealing algorithm, and dynamically adjust the defect judgment threshold according to real-time production parameters; S5. Generate a visual quality heat map and trigger feedback control instructions, which include process parameter adjustment amounts and equipment calibration vectors.
2. The product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, S2 specifically includes: The quantum hue consistency coefficient QCHC is calculated using the following formula: ; in, Let be the mean of the quantum phase shift of color channel c. This represents the maximum standard deviation of the three channels.
3. The product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, S2 specifically includes: The structural anomaly index (SAI) is calculated as follows: ; in, The second derivative norm of the quantum gradient field is represented. This refers to the energy fluctuation over time. 、 These represent the standard deviations of the spatial gradient and the temporal fluctuation, respectively.
4. The product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, S3 specifically includes: The multi-scale convolutional neural network includes: A 1×1 quantum convolution kernel used to extract microscopic defect features; 3×3 spatiotemporal convolution kernels used to capture macroscopic anomalies; Multi-scale features are dynamically fused through an attention mechanism.
5. The product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, The S4 optimizes the decision boundary through the quantum tunneling effect, specifically including: Production parameter constraints are encoded as barrier functions, triggering threshold adjustments when the quality feature vector crosses the barrier region.
6. The product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, S5 includes: The visualized quality heatmap shows: Hue component mapping surface uniformity deviation value; The luminance component is positively correlated with the energy density distribution; The saturation component reflects the probability intensity of defects.
7. The product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, include: By using a time-reversal algorithm to trace the source of historical quality data, a defect evolution prediction model is established.
8. A product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, include: The quantum sensing array employs a diamond NV color center sensor, whose sensitivity satisfies: ; in, For the accuracy of magnetic field detection, is Planck's constant. It is the gyromagnetic ratio. For coherent time, This represents the number of effective sensing units.
9. A product manufacturing quality inspection management method based on quantum sensing according to claim 1, characterized in that, The feedback control command generation includes: The quality feature tensor is reduced in dimension and mapped to the process parameter space to calculate the parameter adjustment vector: ; in, For the pre-trained process response matrix, For real-time quality vectors, This is the standard mass vector.
10. A product manufacturing quality inspection management system based on quantum sensing, characterized in that, include: A quantum sensing module configured with a multi-physics field synchronous acquisition device; An edge computing unit integrating quantum noise suppression and feature extraction accelerators; Deploy a dynamic decision engine based on a quantum-inspired optimization algorithm library.
