Residual value evaluation method and equipment for second-hand electronic equipment and medium

Abstract

The invention discloses a remainder value evaluation method and device for second-hand electronic equipment and a medium, belongs to the technical field of electronic equipment value evaluation, and aims to solve the problems that in existing remainder value evaluation of the second-hand electronic equipment, manual detection evaluation is generally adopted, the subjectivity is high, the remainder value detection dimension is shallow, the labor time consumption cost is high, and the evaluation efficiency is high. And dynamic data support is also lacked. The method comprises the following steps: carrying out structural data acquisition processing related to a physical appearance mode on second-hand electronic equipment to be evaluated; performing data test processing based on a hardware performance mode under non-invasive flaw detection on the second-hand electronic equipment; using mode quantitative analysis under federal local learning is carried out on the software system using mode of the second-hand electronic equipment; carrying out equipment use mode scoring processing under the microscopic state analysis network; and carrying out residual value evaluation processing in a related market residual value evaluation interval between the comprehensive state score and the multi-dimensional value influence factors of the second-hand electronic equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic equipment value evaluation, and in particular to a residual value evaluation method for second-hand electronic equipment, an equipment and a medium. BACKGROUND

[0002] At present, the value evaluation of second-hand electronic equipment (such as smart phones, tablet computers, digital cameras, notebook computers, etc.) mainly relies on manual experience. Evaluators make rough judgments by observing the appearance, testing the basic functions, querying the device model and activation date, etc. Or access specific detection equipment, only use a single program to judge the overall residual value and performance evaluation of the current second-hand electronic equipment, and need the original electronic equipment user to perform multiple verifications and authorizations, etc. The user needs to perform complex authorization, and when the staff performs manual detection, the original electronic equipment user needs to wait for a long time.

[0003] That is, there are certain limitations in the residual value evaluation of second-hand electronic equipment: (1) Strong subjectivity and different standards: In the manual evaluation of the staff, the conclusions of different evaluators differ greatly, there is no objective and unified standard, and it is difficult to truly complete the residual value evaluation of the second-hand equipment in accordance with the actual situation.

[0004] (2) Shallow detection dimension: The detection program is relatively single, it is difficult to truly deeply detect the internal hardware state (such as battery health, screen pixel point defects, mainboard potential repair history, sensor precision attenuation, etc.), which is easy to lead to value misjudgment and transaction disputes, and it is difficult to truly evaluate the real residual mechanism of the second-hand equipment according to the market situation.

[0005] (3) Low efficiency and high cost of manual time consumption: manual detection is time-consuming and labor-intensive, and the process of manual clicking or processing is time-consuming, and the user may also need to wait for a long time, which is not conducive to user experience.

[0006] (4) Lack of dynamic data support: The value evaluation cannot be deeply combined with dynamic data such as real-time market situation and user behavior model, and it is difficult to realize the true embodiment of the residual value.

[0007] Therefore, an intelligent, comprehensive, transparent and reliable residual value evaluation method for second-hand electronic equipment is needed to enable users to clearly understand the actual condition and evaluation basis of the equipment. SUMMARY

[0008] The embodiments of the present application provide a residual value evaluation method for second-hand electronic equipment, an equipment and a medium, which are used to solve the following technical problems: in the residual value evaluation of the existing second-hand electronic equipment, manual detection evaluation is generally used, which has strong subjectivity, shallow residual value detection dimension, high cost of manual time consumption, and also lacks dynamic data support.

[0009] The embodiment of the application adopts the technical solutions as follows: In one aspect, the embodiment of the application provides a residual value evaluation method for a second-hand electronic device, comprising: performing structured data collection and processing on a physical appearance mode of a second-hand electronic device to be evaluated through an automated detection cabin to obtain appearance feature data; performing data test processing on a hardware performance mode of the second-hand electronic device based on non-intrusive flaw detection to obtain hardware performance index data; performing usage mode quantitative analysis on a software system usage mode of the second-hand electronic device under federated local learning according to user authorization permissions, and collecting and obtaining software feature data based on account state information; performing device usage mode scoring processing on the second-hand electronic device under a microscopic state analysis network according to the appearance feature data, the hardware performance index data, the software feature data and historical data labels to obtain a comprehensive state score of the second-hand electronic device; performing residual value evaluation processing on the comprehensive state score and multi-dimensional value influencing factors of the second-hand electronic device under a market residual value evaluation interval through a macroscopic value regression network to obtain final residual value evaluation information; and performing automatic classification processing on the second-hand electronic device based on the final residual value evaluation information and generating a corresponding residual value evaluation report.

[0010] The embodiment of the application greatly improves the efficiency of residual value evaluation of second-hand electronic devices through automated detection and data processing, and reduces the time and cost of manual operation. The multi-modal data collection method, including physical appearance, hardware performance and software usage mode, can more comprehensively and accurately reflect the actual state of the device. The non-intrusive flaw detection technology can avoid physical damage to the device, protect the original appearance of the device and prolong the service life of the device. It can also quantitatively analyze the software system usage mode according to user authorization permissions, and more personalized evaluation of the usage of the device. And the federated local learning technology can protect user privacy while realizing local analysis and processing of data. At the same time, through the microscopic state analysis network, the device usage mode scoring processing can more carefully analyze the usage history and potential problems of the device. Combined with the macroscopic value regression network, the market residual value of the device can be more accurately predicted, providing a more powerful reference basis for both buyers and sellers. Finally, the residual value evaluation report is automatically generated, improving the transparency and reliability of the evaluation results.

[0011] In an implementable embodiment, a structured data acquisition process is performed on a second-hand electronic device to be evaluated regarding a physical appearance modality by an automated detection cabin, to obtain appearance feature data, specifically including: performing a device model identification process on the second-hand electronic device by a multi-interface probe in the automated detection cabin, and determining detection reference data of the current second-hand electronic device based on the identified electronic device model; wherein the detection reference data is standard appearance parameter data corresponding to the current second-hand electronic device; performing a millimeter-level spectral scanning process on the current second-hand electronic device based on the standard appearance parameter data and by a hyperspectral snapshot camera in the automated detection cabin, to obtain reflectance spectral features; and performing device material self-recognition and light source self-adaptive tuning processing on the reflectance spectral features, to obtain light source self-adaptive adjustment data; wherein the device material includes device shell material, coating type, and coating color; adjusting the ultraviolet light and infrared light in the automated detection cabin according to the light source self-adaptive adjustment data to obtain optimal light source working data; wherein the optimal light source working data includes light source wavelength, light source intensity, and light source incidence angle; performing fluorescence scratch marking and coating uniformity intensity evaluation processing on the device surface coating of the current second-hand electronic device by ultraviolet light excitation equipment under the optimal light source working data, to obtain device surface coating features based on ultraviolet light; performing internal structure identification processing on the pure white display screen of the current second-hand electronic device regarding the internal structure under received transmitted light by infrared light excitation equipment under the optimal light source working data, to obtain screen internal structure images; and performing screen assembly adhesion analysis between each lamination layer in the screen assembly regarding the screen internal structure images, to obtain screen assembly adhesion features; calculating the rising rate and steady-state temperature of the corresponding area surface temperature of each screen partition point in the current second-hand electronic device under lighting, to obtain screen backlight uniformity features; and performing sub-pixel defect detection on the current second-hand electronic device, to obtain pixel point features; and obtaining the appearance feature data based on the device surface coating features, the screen assembly adhesion features, the screen backlight uniformity features, and the pixel point features.

[0012] In an embodiment, the second-hand electronic device is subjected to data testing based on non-intrusive hardware performance modalities, to obtain hardware performance index data, including: through the visual module of the automatic detection cabin, pixel feature recognition processing of the data interface image of the second-hand electronic device in an abnormal physical state is performed to obtain interface state features; wherein the abnormal physical state at least includes: foreign matter blockage, pin corrosion, and internal contact pad wear; the data interface of the second-hand electronic device is subjected to stability testing under contact impedance to obtain interface contact features; and the interface state features and the interface contact features are combined to obtain interface performance indicators; the excitation screen of the second-hand electronic device is subjected to sequence excitation and capture under pixel-level defect features to obtain screen pixel display performance indicators; and through the conductive silicone contact, screen touch response dispersion analysis based on standard grid lines is performed on the screen triggering function of the second-hand electronic device to obtain screen touch response performance indicators; and color difference value decay curve calculation is performed on each area color block of the display screen of the second-hand electronic device to obtain screen color accuracy performance indicators; based on the screen pixel display performance indicators, the screen touch response performance indicators, and the screen color accuracy performance indicators, screen performance indicators are obtained; based on a dynamic load test program simulating a real complex use scenario, a load spectrum is constructed; according to each circuit step instant of the load spectrum, the battery dynamic internal resistance of the second-hand electronic device is calculated and processed, and through a recurrent neural network model, the battery health state of the second-hand electronic device is inverted to obtain battery performance indicators; wherein the battery performance indicators include: actual maximum capacity, current SOH value, and predicted remaining cycle life interval; the sensor signal of the second-hand electronic device is subjected to reference signal comparison and dynamic excitation testing, and through a preset Kalman filtering algorithm, abnormal signal testing of the sensor signal under multi-dimensional fusion is performed to obtain sensor performance indicators; based on a standard signal source and a field intensity meter, data throughput rate curve calculation processing is performed on the radio frequency connection performance of the second-hand electronic device to obtain radio frequency performance indicators; the interface performance indicators, the screen color accuracy performance indicators, the battery performance indicators, the sensor performance indicators, and the radio frequency performance indicators are subjected to index set processing to obtain the hardware performance index data.

[0013] In a feasible implementation, before the software system usage modal of the second-hand electronic device is subjected to usage mode quantitative analysis under federated local learning according to user authorization permissions, and software feature data is collected and obtained based on account state information, the method further comprises: based on the software system usage modal, a temporary evaluation environment data package in a secure sandbox is constructed; wherein the temporary evaluation environment data package is a one-time and self-destructing data package; according to the device inherent information of the second-hand electronic device, the temporary evaluation environment data package is subjected to dynamic model loading processing related to the micro-federated learning model, to obtain a security evaluation environment framework for evaluating the software system usage modal of the second-hand electronic device; through the security evaluation environment framework, the power management system of the second-hand electronic device is subjected to data reading processing under device usage intensity, to obtain a usage intensity feature; through the security evaluation environment framework, the second-hand electronic device is subjected to storage running of a standardized storage reading and writing benchmark, and the order reading and writing speed decay rate, the occurrence frequency and type aggregation statistics of kernel-level abnormalities in a specific time period, and application non-response are recorded, to obtain a performance decay feature; through the security evaluation environment framework, the second-hand electronic device is subjected to distribution proportion of network thermal power safety types under network connection, to obtain a network environment feature; based on the usage intensity feature, the performance decay feature, and the network environment feature, a local multi-source data feature is determined.

[0014] In a feasible implementation, according to user authorization permissions, the software system usage modal of the second-hand electronic device is subjected to usage mode quantitative analysis under federated local learning, and software feature data is collected and obtained based on account state information, specifically comprising: through a lightweight self-attention network model, the local multi-source data feature is subjected to device loss feature extraction processing under attention mechanism, to output a loss feature vector; wherein the loss feature vector is a fixed-length data feature that cannot be reversely restored to any original behavior; the loss feature vector is subjected to homomorphic encryption processing, to generate encrypted summary data transmitted to the cloud; and the sandbox environment corresponding to the security evaluation environment framework is subjected to automatic destruction control; according to the account state information after zero-knowledge proof verification, the encrypted summary data is subjected to query and aggregation processing of historical data sources under multi-source heterogeneity, to obtain multi-source query data; wherein the historical data sources at least include: manufacturer official database, authorized service provider network, public market data platform, and industry alliance blockchain; each item of data in the multi-source query data is subjected to spatio-temporal alignment processing, and based on event nodes and attribute nodes in the second-hand electronic device, a time sequence knowledge graph is constructed; wherein the event nodes at least include: production, activation, repair, and transaction; the attribute nodes at least include: color and capacity; each knowledge node in the time sequence knowledge graph is subjected to credibility tracing and weighting processing, to obtain the software feature data after graph reasoning.

[0015] In one feasible implementation, based on the appearance feature data, hardware performance index data, software feature data, and historical data tags, the used electronic device is subjected to device usage modal scoring processing under a micro-state analysis network to obtain a comprehensive state score for the used electronic device. Specifically, this includes: mapping all damage features in the appearance feature data to device damage location weights to obtain an appearance modal damage index; comparing each attribute item in the hardware performance index data with hardware baseline parameters and calculating a single-item health score; assigning a risk label to each single-item health score under a decay mode, outputting a hardware modal health decay vector related to the hardware health score vector; restoring the software feature vector in the software feature data to a mode label, and based on the mode label... The system identifies the corresponding usage intensity coefficient; based on the usage intensity coefficient and the software feature vector, it performs stability scoring on the software modal usage intensity to obtain a software modal usage intensity index; it extracts historical modal information event chains related to specific event nodes of the used electronic equipment from the historical data tags; it performs cross-validation and contradiction detection among the appearance modal damage index, the hardware modal health decay vector, the software modal usage intensity index, and the historical modal information event chains to determine a consistent evidence set and the corresponding comprehensive confidence level; it performs fault derivative judgments on each attribute object in the consistent evidence set for other related components; and it performs risk value quantification calculations on each attribute object and fault derivative item related to residual value to obtain the comprehensive status score.

[0016] In one feasible implementation, a macro-value regression network is used to perform residual value assessment processing between the comprehensive state score and the multi-dimensional value influencing factors of the used electronic device within a relevant market residual value assessment range, obtaining the final residual value valuation information. Specifically, this includes: obtaining market data for the used electronic device to be assessed through a multi-source market API; wherein the market data includes at least: real-time listing price, transaction price, and transaction speed; calculating relevant market supply and demand indices based on the market data and similar configuration data of the used electronic device, and correcting the market supply and demand indices based on seasonal consumption cycles and event-driven correction factors to obtain market dynamic data; obtaining regional demand heat data based on the market dynamic data; and... Using an NLP model, a macro-level corpus is crawled and analyzed to obtain consumer preference data. The market dynamics data, regional demand data, and consumer preference data are combined to obtain multi-dimensional value influencing factors. A value assessment expert network with a gated attention mechanism is used to perform residual value estimation processing on the multi-dimensional value influencing factors of the used electronic devices, resulting in a basic value estimate. A value discount rate mapping is performed between the specific scoring defect features in the comprehensive status score and the basic value estimate. The value discount rate mapping result is then processed for range evaluation using a market residual value assessment interval to obtain the final residual value valuation information. The market residual value assessment interval is a real-time market value fluctuation range based on the multi-dimensional value influencing factors.

[0017] In one feasible implementation, based on the final residual value valuation information, the used electronic devices are automatically graded and a corresponding residual value assessment report is generated. Specifically, this includes: comparing the residual value valuations in the final residual value valuation information according to a preset valuation range to determine the residual value level of the used electronic devices; wherein the residual value level includes: premium, good, and affordable; based on the residual value level, the used electronic devices are sorted to the corresponding storage compartments of the automated testing cabin; and based on the residual value level, the final residual value valuation information, the comprehensive status score, hardware health charts, three-dimensional appearance models, and software test results, the residual value assessment report is generated.

[0018] Secondly, embodiments of this application also provide a device for assessing the residual value of used electronic devices, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method for assessing the residual value of used electronic devices as described in any of the above embodiments.

[0019] Thirdly, embodiments of this application also provide a non-volatile computer storage medium, wherein the storage medium is a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores at least one program, each program including instructions, wherein when the instructions are executed by a terminal, the terminal performs a residual value assessment method for a second-hand electronic device as described in any of the above embodiments.

[0020] This application provides a method, device, and medium for assessing the residual value of used electronic devices. Compared with the prior art, the embodiments of this application have the following beneficial technical effects: 1. Improved assessment efficiency: Through automated detection and data processing, the efficiency of assessing the residual value of used electronic equipment is greatly improved, reducing the time and cost of manual operation.

[0021] 2. Data accuracy: Employing multimodal data acquisition methods, including physical appearance, hardware performance, and software usage patterns, it can more comprehensively and accurately reflect the actual status of the equipment.

[0022] 3. Non-invasive testing: Non-invasive flaw detection technology can avoid physical damage to equipment, protect the original appearance of the equipment, and extend the service life of the equipment.

[0023] 4. Personalized assessment: Based on user-authorized permissions, quantitative analysis of software system usage modalities can be performed, enabling a more personalized assessment of equipment usage.

[0024] 5. Federated Local Learning: Federated local learning technology can protect user privacy while enabling local analysis and processing of data.

[0025] 6. Microstate analysis: By using a microstate analysis network to score the equipment usage modes, it is possible to analyze the equipment's usage history and potential problems in more detail.

[0026] 7. Market Value Forecasting: By combining macro-value regression networks, the residual market value of equipment can be predicted more accurately, providing a stronger reference for both buyers and sellers.

[0027] 8. Automated grading: Based on residual value valuation information, the equipment is automatically graded, simplifying the market circulation process.

[0028] 9. Report Generation: Automatically generates residual value assessment reports, improving the transparency and reliability of assessment results.

[0029] 10. Enhanced User Experience: The entire evaluation process is automated and intelligent, improving user satisfaction with the used electronic device evaluation service. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart illustrating a method for assessing the residual value of second-hand electronic equipment provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of a residual value assessment device for second-hand electronic equipment provided in an embodiment of this application. Detailed Implementation

[0031] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0032] This application provides a method for assessing the residual value of used electronic devices, such as... Figure 1 As shown, the method for assessing the residual value of used electronic devices specifically includes steps S101-S106: S101. Through an automated inspection chamber, structured data acquisition and processing of the physical appearance modalities of the second-hand electronic equipment to be evaluated are performed to obtain appearance feature data.

[0033] Specifically, the process begins by using multi-interface probes in an automated testing chamber to identify the model of the used electronic equipment. Based on the identified model, reference data for testing the used electronic equipment is determined. This reference data consists of standard appearance parameter data corresponding to the used electronic equipment.

[0034] Furthermore, based on standard appearance parameter data, and using a hyperspectral camera in an automated inspection chamber, millimeter-level spectral scanning is performed on the currently used electronic equipment to obtain reflectance spectral characteristics. These reflectance spectral characteristics are then processed for equipment material self-identification and adaptive light source tuning to obtain adaptive light source adjustment data. The equipment material includes: the equipment shell material, coating type, and coating color.

[0035] In one embodiment, the detection chamber houses a hyperspectral fast camera, which performs a millisecond-level broad-spectrum scan upon the device's entry. By analyzing the reflectance spectral characteristics, the AI ​​model identifies the device's casing material (e.g., aerospace aluminum alloy, polished glass, frosted plastic, ceramic, etc.), coating type (matte, glossy, pearlescent), and color in real time. Based on the identification results, the system dynamically adjusts the wavelength, intensity, and incident angle of subsequent UV and infrared light sources. For example, for a dark frosted back panel, the UV light intensity is increased to enhance scratch contrast; for a mirrored frame, grazing light at a specific angle is used to avoid interference from mirror reflection.

[0036] Furthermore, based on the adaptive adjustment data of the light source, multiple parameters of the ultraviolet and infrared light in the automated detection chamber are adjusted to obtain the optimal light source operating data. The optimal light source operating data includes: light source wavelength, light source intensity, and light source incident angle.

[0037] Furthermore, by using an ultraviolet light excitation device operating under optimal light source conditions, the surface coating of currently used electronic devices is subjected to fluorescent scratch marking and coating uniformity intensity evaluation processing to obtain the surface coating characteristics of the device under ultraviolet light.

[0038] In one embodiment, UV light of a specific wavelength is used to excite a surface coating (such as an oleophobic layer or paint) to generate characteristic fluorescence. Fluorescence images are captured using a high-sensitivity CMOS camera. In scratch quantification, the coating is damaged at the scratch location, resulting in interruption or alteration of the fluorescence signal. Through image processing, not only can the scratch location be marked, but scratches can also be classified into three levels—"micro-scratches" (damaging only the oleophobic layer), "shallow scratches" (damaging the paint), and "deep scratches" (damaging the substrate)—based on the width and depth of the fluorescence loss (calculated through fluorescence intensity gradients). In coating uniformity evaluation, the fluorescence intensity distribution map of the entire back panel can be analyzed to quantitatively assess coating aging, uneven wear, or localized repainting (the fluorescence characteristics of the repainted area will differ from the original coating).

[0039] As a feasible implementation method, multi-state identification of liquid contact indicators (LCIs) can also be performed. This involves irradiating the LCI with specific narrow-band UV light and analyzing its reflectance spectrum using a spectrometer. This not only determines whether the LCI has turned red, but also identifies the stages of its color change (such as pink, bright red, and dark red). Combined with algorithms, the possible liquid exposure time and type can be inferred to assess the corrosion risk level.

[0040] Furthermore, using an infrared excitation device operating under optimal light source conditions, the internal structure of the pure white display screen of the currently used electronic device is identified under received transmitted light, resulting in an image of the screen's internal structure. The adhesion between the various bonding layers in the screen assembly is then analyzed based on this image to obtain the screen assembly's bonding characteristics.

[0041] In one embodiment, when the screen displays a pure white, high-brightness image, a near-infrared (NIR) light source is used to illuminate the screen from the front, and an infrared camera receives the transmitted light from the side or a specific angle. Due to the different absorption / scattering of infrared light by the internal structure, an image of the screen's internal structure can be generated. Then, by analyzing the morphology and distribution of Newton's rings, dark spots, or bright spots in the image, it is possible to accurately determine whether there are localized peeling, bubbles, or uneven adhesive layers between the various bonding layers (cover glass, touch layer, display layer) in the screen assembly (especially OLED screens).

[0042] Furthermore, it is necessary to calculate the rate of temperature rise and steady-state temperature of the corresponding area under illumination for each screen partition in the current used electronic device to obtain the screen backlight uniformity characteristics. Sub-pixel defect detection is then performed on the current used electronic device to obtain pixel features.

[0043] In one embodiment, backlight uniformity quantification requires analyzing the rate of temperature rise and steady-state temperature of the corresponding area when each zone is lit. Aging or defective backlight LED areas will exhibit abnormal temperature rise curves (e.g., slow heating, low temperature, or excessively high temperature). When displaying specific intricate patterns, combining high-resolution visible light and thermal imaging can detect abnormal cold or hot spots, accurately locate bad pixels, dark spots, or persistent bright spots, and even identify defects where only a specific color (sub-pixel) fails.

[0044] Furthermore, based on the characteristics of the device surface coating, the screen assembly bonding characteristics, the screen backlight uniformity characteristics, and the pixel characteristics, the appearance feature data is finally integrated and obtained.

[0045] S102. Perform hardware performance mode data testing on the used electronic equipment based on non-invasive flaw detection to obtain hardware performance index data.

[0046] Specifically, the vision module of the automated inspection chamber first performs pixel feature recognition processing on the data interface images of the used electronic equipment to identify abnormal physical conditions, thereby obtaining interface status characteristics. Abnormal physical conditions include at least: foreign object blockage, pin corrosion, and internal contact wear.

[0047] In one embodiment, a convolutional neural network (CNN) is used to identify the interface type (USB-C, Lightning, etc.) in real time and predict the physical condition of the interface: such as whether there is foreign object blockage, pin corrosion, internal contact wear or deformation. This prediction result will determine the subsequent operation: whether to conduct normal testing or trigger an "abnormal interface handling process".

[0048] Furthermore, stability tests were conducted on the data interfaces of the used electronic devices under contact impedance to obtain interface contact characteristics. The interface state characteristics and interface contact characteristics were then combined to obtain interface performance indicators.

[0049] In one implementation, the system continuously monitors the stability of the contact impedance after docking and throughout the entire testing cycle. If the impedance fluctuation exceeds a threshold, the system determines it as "poor contact," automatically records it as a defect of the interface, and may retry docking or terminate the high-current test to ensure safety.

[0050] Furthermore, it is necessary to perform sequential excitation and capture of pixel-level defect features on the excitation screen of the used electronic device to obtain the screen pixel display performance index. Then, using conductive silicone contacts, a touch response dispersion analysis based on a standard grid line is performed on the screen triggering function of the used electronic device to obtain the screen touch response performance index. Finally, the attenuation curve of the color difference value of each color block in the display screen of the used electronic device is calculated to obtain the screen color accuracy performance index.

[0051] In one embodiment, a set of rapidly switching color and grayscale sequences (such as pure white, pure black, red, green, blue, and 0-255 grayscale levels) is injected and displayed in full screen. These sequences are captured using a synchronously triggered, high-frame-rate (≥240fps), high-resolution industrial camera. Instead of relying on a single image, the brightness-time response curve of each pixel throughout the entire color sequence is analyzed. A pixel that remains constantly bright or dark under any screen display is considered an absolute dead pixel; a pixel that only fails to respond under green screen displays is considered a damaged green subpixel. The response delay time (Rise / Fall Time) and overshoot of pixels from black to white and white to black are analyzed. Slowly responding pixel areas may indicate aging of the screen driver circuitry, ultimately yielding the screen pixel display performance indicators.

[0052] In one embodiment, a touch error heatmap of the entire screen is generated by comparing the actual coordinates of the touch point with the coordinates reported by the screen. Increased error in edge areas is a common aging phenomenon. Multiple touch points can also be used to simulate multi-finger operations to detect "ghost points" or "missing touches." Simultaneously, applying slight pressure to one side of the screen tests whether the touch signal on the other side is interfered with, thus assessing the screen's structural strength. Multiple clicks at the same location are performed, and the standard deviation of the reported coordinates is analyzed to quantify the degree of touch signal "jitter," ultimately yielding the screen's touch response performance indicators.

[0053] In one implementation, it is necessary to accurately calculate the sRGB and P3 color gamut coverage area of ​​the screen and the color difference ΔE value of each color block to generate a color accuracy report. Aging OLED screens may exhibit color shift. Then, when displaying a full white image, the brightness value is increased stepwise from the lowest to the highest brightness, and the actual light intensity at each brightness level is recorded. A measured brightness-set value curve is plotted. By comparing the deviation of this curve with the factory data for a new device (especially the "black screen glow" at low brightness or the inability to reach higher brightness levels), the decay of screen luminous efficiency can be accurately assessed, ultimately yielding the screen color accuracy performance index.

[0054] Furthermore, screen performance metrics are derived based on screen pixel display performance metrics, screen touch response performance metrics, and screen color accuracy performance metrics. Then, a load spectrum is constructed based on a dynamic load test program simulating real-world complex usage scenarios. In other words, a dynamic load test program simulating real-world complex usage scenarios can be injected into used electronic devices. This program includes: instantaneous high-current pulses (simulating taking a picture), continuous medium current (simulating video playback), and low-frequency small-current fluctuations (simulating standby), forming a load spectrum.

[0055] Furthermore, based on each circuit step instant in the load spectrum, the dynamic internal resistance of the battery in the used electronic device is calculated. Then, using a recurrent neural network model, the battery health status of the used electronic device is inverted to obtain battery performance indicators. These indicators include: actual maximum capacity, current state of equilibrium (SOH) value, and predicted remaining cycle life range.

[0056] In one embodiment, at each current step in the load spectrum, the dynamic internal resistance at that moment is calculated based on ΔV / ΔI. As the battery ages, the internal resistance increases, and the changes are more drastic at different SOCs (State of Charge). A recurrent neural network (RNN) model is then constructed. This model can collect real-time V(t) and I(t) sequences, the cycle count reported by the device, and the current temperature. The model outputs the battery's actual maximum capacity (relative to design capacity), current SOH (health status, 100% new), and predicted remaining cycle life range. Finally, the battery health status of used electronic devices is inverted to obtain battery performance indicators.

[0057] Furthermore, the sensor signals of the used electronic devices must be compared with reference signals and subjected to dynamic excitation tests. The sensor signals are then subjected to multi-dimensional fusion abnormal signal tests using a preset Kalman filter algorithm to obtain sensor performance indicators.

[0058] In one embodiment, algorithms such as extended Kalman filtering are used to fuse readings from accelerometers, gyroscopes, and magnetometers to calculate the device's attitude. When a sensor (such as a magnetometer) continuously provides signals that contradict the calculation results from other sensors, it can be determined that the sensor has malfunctioned or is severely interfered with (e.g., its internal magnetic shielding has been damaged due to a drop), thus completing the acquisition of the sensor's performance indicators.

[0059] Furthermore, based on a standard signal source and a field strength meter, the data throughput curve of the radio frequency connection performance of the used electronic equipment is calculated and processed to obtain radio frequency performance indicators.

[0060] In one embodiment, within the shielded enclosure, the connection stability and receiving sensitivity of the device's Wi-Fi and Bluetooth signals are tested using a built-in standard signal source and field strength meter. By analyzing the connection handshake success rate and data throughput curves under different signal strengths, the normal operating status of the motherboard's RF-related circuits (such as antenna contacts and RF chips) can be indirectly determined, thereby obtaining RF performance indicators.

[0061] Furthermore, the interface performance indicators, screen color accuracy performance indicators, battery performance indicators, sensor performance indicators, and radio frequency performance indicators are processed into a set of indicators to finally obtain hardware performance indicator data.

[0062] S103. Based on user authorization permissions, perform quantitative analysis of the usage patterns of the software system of second-hand electronic devices under federated local learning, and collect and obtain software feature data based on account status information.

[0063] Specifically, based on the software system usage modality, a temporary evaluation environment data package is first constructed within a secure sandbox. This temporary evaluation environment data package is a one-time, self-destructing data package.

[0064] In one embodiment, a one-time, self-destructing temporary evaluation environment data package needs to be generated by the cloud-based evaluation platform. This data package is pushed to the device under test via a secure interface (such as a device management API). This environment runs in a secure sandbox or virtual container on the device, completely isolated from the user's main environment, and cannot access any real user application data (such as photos or contacts). Before restarting, the environment requests a set of non-privacy system performance metadata permissions from the device system (such as access to CPU / memory usage history, aggregated anonymized storage I / O statistics, and system event logs), and presents a clear authorization request to the user, involving only "device performance analysis".

[0065] Furthermore, based on the inherent information of the used electronic devices, the temporary evaluation environment data package undergoes dynamic model loading processing for relevant micro-federated learning models, resulting in a secure evaluation environment framework for modal evaluation of the software system of used electronic devices. That is, the secure evaluation environment framework can dynamically load corresponding pre-trained micro-federated learning models and diagnostic script sets from the cloud based on the device model and operating system version. The micro-federated learning models and scripts are extremely simple (<5MB), optimized specifically for this device model, ensuring analysis efficiency and reducing device load.

[0066] Furthermore, using a security assessment environment framework, data on the power management system of used electronic devices under relevant usage intensity is read and processed to obtain usage intensity characteristics. Then, using the same framework, standardized storage read / write benchmarks are applied to the used electronic devices, and the frequency and type of sequential read / write speed degradation, kernel-level anomalies within specific time periods, and application unresponsiveness are aggregated and statistically analyzed to obtain performance degradation characteristics. Next, using the same framework, the distribution ratio of network thermoelectric safety types under network connectivity is analyzed to obtain network environment characteristics. Finally, based on the usage intensity characteristics, performance degradation characteristics, and network environment characteristics, local multi-source data characteristics are determined.

[0067] In one embodiment, within a security assessment environment framework, a diagnostic script runs. For usage intensity characteristics: it can read the daily average screen-on frequency and duration distribution curve (not specific time points), and the average charging cycle interval and duration from the system power management interface. For performance degradation characteristics: it can run standardized storage read / write benchmark tests, recording the sequential / random read / write speed degradation rate (compared to a new device of the same model); it can analyze the frequency and type aggregation statistics of kernel-level anomalies and application unresponsiveness (ANR) within a specific time period from the system logs (excluding any application identifiers or content). For network environment characteristics: it can anonymously statistically analyze the distribution ratio of Wi-Fi hotspot security types (such as WPA2 / 3) that the device has connected to, as an indirect indicator of the device's potential usage environment (e.g., long-term connection to an open network may suggest higher security risks or specific use cases). Finally, based on usage intensity characteristics, performance degradation characteristics, and network environment characteristics, local multi-source data characteristics are determined.

[0068] Furthermore, a lightweight self-attention network model is used to extract device loss features from the local multi-source data features under the attention mechanism, outputting a loss feature vector. This loss feature vector is a fixed-length data feature that cannot be reversed to reconstruct any original behavior.

[0069] Furthermore, the loss feature vector is homomorphically encrypted to generate encrypted digest data that is transmitted to the cloud. The sandbox environment corresponding to the security assessment framework is also automatically destroyed.

[0070] Furthermore, based on the account status information verified by zero-knowledge proofs, the encrypted digest data needs to be queried and aggregated from historical data sources across multiple heterogeneous sources to obtain multi-source query data. These historical data sources include at least: official manufacturer databases, authorized service provider networks, public market data platforms, and industry consortium blockchains.

[0071] As a feasible implementation method, in zero-knowledge proof verification, for certain situations where account information cannot be directly exposed, an interaction protocol is designed to allow the device to prove to the evaluation system that "its currently logged-in account is unique and can be logged out normally" without revealing account details. This proves that the device is not stolen and has no risk of hiding accounts, thereby obtaining the account status information after zero-knowledge proof verification.

[0072] In one embodiment, the cloud platform utilizes the device's unique identifier (such as serial number or IMEI) to initiate queries to multiple data sources, provided it is legal, compliant, and authorized by the user. These sources include: 1) Manufacturer's official database: obtaining the initial activation date, original configuration, official warranty status, and repair records. 2) Authorized service provider network: querying repair records that are not official but are within the authorized network (blockchain-based verification is required to ensure authenticity). 3) Public market data platform: crawling historical transaction price curves and market supply and demand indices for this model of device or devices in similar condition. 4) Industry consortium blockchain: querying whether the device is on an industry-shared "blacklist" (such as a list of stolen devices).

[0073] Furthermore, each data item in the multi-source query data undergoes spatiotemporal alignment processing, and a temporal knowledge graph is constructed based on event nodes and attribute nodes in the used electronic devices. Event nodes include at least: production, activation, repair, and transaction; attribute nodes include at least: color and capacity. Finally, each knowledge node in the temporal knowledge graph undergoes credibility tracing and weighting processing to obtain the software feature data after graph inference.

[0074] In one embodiment, each piece of data from multi-source query data with different sources and formats is cleaned, aligned, and associated using identifiers and timestamps. A knowledge graph is constructed centered on the device entity, containing "event" nodes (e.g., production, activation, repair A, repair B, transaction X, transaction Y) and "attribute" nodes (e.g., color, capacity). Edges represent relationships between events (e.g., "caused" or "occurred at"). The confidence level of each historical record is then weighted according to its data source (e.g., official repair records have a weight of 1.0, while third-party claims of "no disassembly / repair" have a weight of 0.4). Using data hashes stored on the blockchain, key claims (e.g., "motherboard not repaired") can be verifiable and their credibility weights significantly increased. Finally, graph inference is used to assess the impact of historical events on current value. For example, if the graph shows "an official screen repair a year ago" connected to "no subsequent fault records," then the negative impact of this repair on value may be less than a third-party motherboard repair of unknown date; thus, software feature data after graph inference is obtained.

[0075] S104. Based on appearance feature data, hardware performance index data, software feature data, and historical data tags, perform device usage modality scoring under the micro-state analysis network to obtain a comprehensive state score for the used electronic equipment.

[0076] Specifically, the appearance feature data is processed by mapping the equipment damage location weights to obtain the appearance modal damage index.

[0077] In one embodiment, all damage features (such as dents, scratches, bends, and coating peeling) in the appearance feature data are automatically identified and classified. For each damage feature, its area, depth / height, volume, and surface roughness variation are calculated. Then, based on the location of the damage, its positional weight is retrieved from a predefined "structure-function weight matrix". For example, a dent on the back of the motherboard has a high weight W_loc (potentially affecting heat dissipation or causing internal stress). Scratches on the screen display area have a higher weight than scratches on the bezel. Wear around the charging port has a higher weight than ordinary wear on the side bezel. Finally, individual damage indices are calculated and combined to form an appearance modal damage index.

[0078] Furthermore, each attribute in the hardware performance index data is compared with the hardware baseline parameters, and an individual health score is calculated. Each individual health score is then labeled with a risk level under a decay mode, and a hardware modal health decay vector is output for the overall hardware health score vector.

[0079] In one embodiment, each hardware parameter (such as maximum battery capacity, peak screen brightness, and touch sampling rate) is compared with the factory benchmark values ​​of a new device of the same model and industry-defined fault thresholds. Then, an individual health score is calculated and mapped to a 0-1 range (1 for brand new, 0 for completely failed). Next, time-series or curve analysis is used to determine whether the degradation is linear aging (such as battery cycle degradation) or non-linear abrupt changes (such as localized dead pixels on the screen), with the latter posing a higher risk. High-risk patterns such as "non-linear abrupt changes" and "abnormal performance curve fluctuations" are then marked with risk indicators. Finally, the hardware modal health score degradation vector, the set of risk indicators, and the confidence level of each individual health score (based on the stability and completeness of the test data) are output.

[0080] Furthermore, the software feature vectors in the software feature data are restored to pattern labels, and the corresponding usage intensity coefficients are determined based on the pattern labels. Based on the usage intensity coefficients and the software feature vectors, a stability score is applied to the software modal usage intensity to obtain the software modal usage intensity index.

[0081] In one embodiment, a corresponding decoder in the cloud (paired with a federated learning model) can be used to restore the feature vectors into understandable pattern labels, rather than the raw data. For example, the output might be: "High-frequency short-duration screen-on mode," "High-intensity graphics computing mode," or "Balanced daily usage mode." Then, based on the pattern label, a usage intensity coefficient U (e.g., 0.8-1.2, with 1.0 being the standard intensity) is assigned. Next, sub-features such as system anomaly frequency and storage error rate are extracted from the feature vector to generate a software stability score. Finally, the output is a comprehensive software modality usage intensity index, combining the pattern label, intensity coefficient, and stability score.

[0082] Furthermore, extract historical modal information event chains from historical data tags related to clearly defined event nodes of used electronic devices. This can be done by: extracting event nodes in the graph that have a clear impact on the current state, such as "official screen replacement (1 year ago)," "third-party motherboard repair (time unknown)," and "market circulation records (3 times)." Then, assign a credibility weight (based on the data source) and a time decay factor to each event (the older the event, the smaller the impact). Finally, determine whether the event node has a positive correction (e.g., replacing a new battery), a negative damage (e.g., water damage repair), or a risk warning (e.g., unexplained disassembly) on the overall state of the used electronic device; thus determining the historical modal information event chain.

[0083] Furthermore, cross-validation and contradiction detection are performed between the appearance modal damage index, hardware modal health decay vector, software modal usage intensity index, and historical modal information event chain to determine the consistent evidence set and the corresponding comprehensive confidence level.

[0084] In one embodiment, a lightweight verification network can be established to check the consistency between different modalities of evidence. For example, the software modality indicates "high-intensity use," but the battery health is close to that of a new device. The verification network triggers a check of historical events: if a record of "recent official battery replacement" is found, the contradiction is resolved, and the credibility of that historical record and the current battery health is increased. Another example: a minor crack is found during visual inspection, but hardware testing shows that all screen parameters are perfect. The verification network combines infrared imaging results to determine: if there are no abnormalities in the fit, it may only be damage to the outer glass cover, thus correcting the impact range of the visual damage index. Finally, the output is a consistent evidence set after cross-validation and adjustment, as well as the updated overall confidence level of each piece of evidence, which is the corresponding consistent evidence set and its corresponding overall confidence level.

[0085] Furthermore, each attribute object in the consistency evidence set is used to make fault-related judgments about other related components. Each attribute object and its fault derivatives are then subjected to a risk value quantification calculation related to residual value to obtain a comprehensive status score.

[0086] As a feasible implementation, evidence (such as "battery health H=0.75" or "existence of unofficial repair records") can be used as input to activate relevant nodes in the graph. Reasoning is then performed along the graph edges. For example, "unofficial repair" and "moderate battery performance degradation" might jointly infer "the existence of inferior replacement parts or installation process risks," further leading to "an increased potential risk of future failures in other related components." Then, for each activated current defect and derived risk node, a risk value is calculated based on its severity and probability of occurrence. Finally, the current defect set and potential risk set, along with their quantified values, are output, thus completing the failure derivation judgment for other related components. Furthermore, the risk value related to residual value is quantified for each attribute object and failure derivation item to obtain a comprehensive status score.

[0087] S105. Through the macro-value regression network, the residual value assessment is carried out between the comprehensive status score and the multi-dimensional value influencing factors of second-hand electronic devices under the relevant market residual value assessment range to obtain the final residual value valuation information.

[0088] Specifically, the process begins by accessing multi-source market APIs to obtain market data for the secondhand electronic devices to be evaluated. This market data includes at least: real-time listing prices, transaction prices, and transaction speed.

[0089] Furthermore, based on market data, the market supply and demand index is calculated for similar configurations of second-hand electronic devices. Then, the market supply and demand index is corrected based on seasonal consumption cycles and event-driven adjustment factors to obtain dynamic market data.

[0090] In one embodiment, APIs from mainstream secondhand trading platforms, auction websites, and trade-in channels can be accessed to obtain real-time listing prices, transaction prices, and transaction speed (inventory turnover days) data for devices of the same model and similar configuration (Same-Model-Similar-Spec, SMSS) as the evaluated equipment. The data is aggregated by rough classification of condition (e.g., "90% new," "80% new"). Then, based on the SMSS data, the market supply and demand index (MSI) is calculated in real-time. Finally, public calendars and industry news are integrated to automatically mark time nodes such as before and after new product launches, major shopping festivals, and seasonal consumption cycles. A basic price fluctuation coefficient is pre-set or learned for each node; that is, based on seasonal consumption cycles and event-driven correction factors, the market supply and demand index is adjusted to obtain dynamic market data.

[0091] Furthermore, based on market dynamics data, regional demand intensity data is obtained. Then, using an NLP model, a macro-level corpus is crawled and analyzed to obtain consumer preference data. Finally, the market dynamics data, regional demand intensity data, and consumer preference data are combined to derive multidimensional value influencing factors.

[0092] In one embodiment, an NLP model is used to crawl and analyze discussions about a specific model in tech forums and social media. Sentiment scores and high-frequency concerns (such as "Is the screen of this model prone to burn-in?" and "How is the battery life?") are calculated. A model feature attention vector is constructed to obtain consumer preference data.

[0093] Furthermore, by using a value assessment expert network under a gating attention mechanism, residual value is estimated among the multidimensional value-influencing factors of second-hand electronic devices to obtain a basic value estimate.

[0094] In one embodiment, the value assessment expert network under the gated attention mechanism comprises multiple value assessment expert subnetworks, each specializing in scenarios dominated by a particular type of value-influencing factor (e.g., "top-quality collector expert," "mainstream practical expert," "defective discount expert"). A lightweight neural network (gated network) is then used to calculate a set of routing weights based on the input vector. For example, if the device status score S is extremely high (>95) and the market is scarce (high MSI), the routing weights will favor the "top-quality collector expert"; if the device has obvious defects but is fully functional and is on sale, the weights will favor the "defective discount expert." Finally, each expert subnetwork independently receives all input features and outputs a basic value prediction.

[0095] Furthermore, a value discount rate mapping calculation must be performed between the specific scoring defect characteristics in the comprehensive status score and the basic value estimate. Then, using the market residual value assessment interval, the value discount rate mapping result is processed for interval range assessment to obtain the final residual value valuation information. The market residual value assessment interval is the real-time market value fluctuation interval based on multi-dimensional value influencing factors.

[0096] As a feasible implementation method, a continuously learning mapping rule library needs to be built first, linking the specific defects diagnosed by MSP-Net with market-validated value depreciation rates. For example: "For every percentage point below 90% in battery health SOH, a depreciation of XX yuan is incurred on mainstream models." "A visible dead pixel on the screen depreciates the device's current base value by Y%." "Unofficial repair records exist, but the device functions normally, resulting in a depreciation of Z%." (The Z value varies depending on the repaired part). Then, the system automatically queries the mapping library for the corresponding depreciation rule for each item in the diagnostic report D_vec, calculating the total depreciation value. Considering market fluctuations and assessment uncertainties, a value assessment range is output. Then, through the market residual value assessment range associated with the total depreciation value, the value depreciation rate mapping result is processed for range range assessment to obtain the final residual value valuation information. That is, the true final residual value of the current used electronic device is derived by combining real-time dynamic market conditions.

[0097] S106. Based on the final residual value valuation information, the second-hand electronic devices are automatically graded and a corresponding residual value assessment report is generated.

[0098] Specifically, based on a preset valuation range, the remaining value valuation in the final remaining value valuation information is compared within that range to determine the remaining value grade of the used electronic devices. These remaining value grades include: Excellent, Good, and Affordable.

[0099] Furthermore, based on the residual value level, the used electronic devices are sorted to the corresponding storage compartments in the automated inspection cabin. Finally, based on the residual value level, final residual value valuation information, comprehensive condition score, hardware health chart, 3D appearance model, and software test results, a residual value assessment report is generated.

[0100] In addition, embodiments of this application also provide a device for assessing the residual value of second-hand electronic equipment, such as... Figure 2 As shown, the equipment used for residual value assessment of second-hand electronic equipment specifically includes: At least one processor 201; and a memory 202 communicatively connected to the at least one processor 201; wherein the memory 202 stores instructions executable by the at least one processor 201 to enable the at least one processor 201 to execute: The automated inspection chamber is used to collect and process structured data on the physical appearance modalities of the second-hand electronic equipment to be evaluated, and to obtain appearance feature data. The hardware performance modal data of the used electronic equipment is processed by non-invasive flaw detection to obtain hardware performance index data. Based on user authorization permissions, the usage patterns of the software system of second-hand electronic devices are quantitatively analyzed under federated local learning, and software feature data is collected and obtained based on account status information. Based on appearance feature data, hardware performance index data, software feature data, and historical data tags, the device usage modality scoring under the micro-state analysis network is performed on the used electronic devices to obtain a comprehensive state score for the used electronic devices. By using the macro value regression network, the residual value assessment is performed on the comprehensive status score and the multi-dimensional value influencing factors of second-hand electronic devices under the relevant market residual value assessment range to obtain the final residual value valuation information. Based on the final residual value valuation information, the used electronic devices are automatically graded and a corresponding residual value assessment report is generated.

[0101] This application's embodiments significantly improve the efficiency of residual value assessment for used electronic equipment through automated detection and data processing, reducing the time and cost of manual operations. Employing a multimodal data acquisition method, including physical appearance, hardware performance, and software usage patterns, it can more comprehensively and accurately reflect the actual condition of the equipment. Utilizing non-invasive flaw detection technology avoids physical damage to the equipment, preserving its original appearance and extending its lifespan. Furthermore, it can quantitatively analyze software system usage patterns based on user authorization permissions, enabling more personalized assessments of the equipment's usage status. Federated local learning technology protects user privacy while enabling local data analysis and processing. Simultaneously, through a micro-state analysis network, it scores equipment usage patterns, allowing for a more detailed analysis of the equipment's usage history and potential problems. Combined with a macro-value regression network, it can more accurately predict the equipment's market residual value, providing a stronger reference for both buyers and sellers. Finally, it automatically generates a residual value assessment report, improving the transparency and reliability of the assessment results.

[0102] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0103] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0104] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0106] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0109] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0110] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0111] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0112] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of this specification.

Claims

1. A method of assessing the residual value of a second-hand electronic device, characterized in that, The method includes: The automated inspection chamber is used to collect and process structured data on the physical appearance modalities of the second-hand electronic equipment to be evaluated, and to obtain appearance feature data. The used electronic equipment is subjected to data testing and processing based on hardware performance modalities under non-invasive flaw detection to obtain hardware performance index data. Based on user authorization permissions, the software system usage modalities of the second-hand electronic devices are subjected to quantitative analysis of usage patterns under federated local learning, and software feature data is collected and obtained based on account status information. Based on the appearance feature data, hardware performance index data, software feature data, and historical data tags, the used electronic device is subjected to device usage modality scoring under a micro-state analysis network to obtain a comprehensive state score for the used electronic device. By using a macro-value regression network, the comprehensive status score is compared with the multi-dimensional value influencing factors of the second-hand electronic equipment to perform residual value assessment processing under the relevant market residual value assessment range, so as to obtain the final residual value valuation information. Based on the final residual value valuation information, the second-hand electronic devices are automatically graded, and a corresponding residual value assessment report is generated.

2. A method of assessing the residual value of a used electronic device according to claim 1, wherein, The automated inspection chamber performs structured data acquisition and processing on the physical appearance modalities of the used electronic equipment to be evaluated, resulting in appearance feature data, specifically including: The second-hand electronic equipment is identified by the multi-interface probes in the automated testing chamber, and the testing reference data is determined based on the identified electronic equipment model. The testing reference data is the standard appearance parameter data corresponding to the second-hand electronic equipment. Based on the standard appearance parameter data, and using a hyperspectral camera in the automated inspection chamber, the current used electronic equipment is subjected to millimeter-level spectral scanning to obtain reflectance spectral characteristics; and the reflectance spectral characteristics are then subjected to equipment material self-identification and light source adaptive tuning to obtain light source adaptive adjustment data; wherein, the equipment material includes: equipment shell material, coating type, and coating color; Based on the adaptive adjustment data of the light source, multiple parameters of the ultraviolet and infrared light in the automated detection chamber are adjusted to obtain the optimal light source operating data; wherein, the optimal light source operating data includes: light source wavelength, light source intensity, and light source incident angle; Using an ultraviolet light excitation device operating under the optimal light source data, the surface coating of the currently used electronic device is subjected to fluorescent scratch marking and coating uniformity intensity evaluation processing to obtain the surface coating characteristics of the device under ultraviolet light. Using an infrared light excitation device operating under the optimal light source data, the internal structure of the pure white display screen of the current second-hand electronic device under received transmitted light is identified to obtain an image of the screen's internal structure; and the adhesion between the various bonding layers in the screen assembly is analyzed to obtain the screen assembly's bonding characteristics. obtain a screen backlight uniformity feature by calculating the rising rate of the surface temperature of the corresponding area and the steady-state temperature of each screen partition point in the current second-hand electronic device; obtain the appearance feature data based on the device surface coating feature, the screen assembly fitting feature, the screen backlight uniformity feature, and the pixel point feature.

3. The method of claim 1, wherein, perform data testing and processing on the second-hand electronic device based on the hardware performance modal of non-intrusive flaw detection to obtain hardware performance index data, specifically including: perform pixel feature recognition processing on the data interface image of the second-hand electronic device in an abnormal physical state through the visual module of the automatic detection cabin to obtain an interface state feature; wherein the abnormal physical state at least includes foreign matter blockage, pin corrosion, and internal contact pad wear; perform a stability test on the data interface of the second-hand electronic device under contact impedance to obtain an interface contact feature; and combine the interface state feature and the interface contact feature to obtain an interface performance index; perform sequence excitation and capture on the second-hand electronic device in a pixel-level defect feature to obtain a screen pixel display performance index; perform touch response dispersion analysis on the screen trigger function of the second-hand electronic device based on a standard grid line through a conductive silicone contact to obtain a screen touch response performance index; and perform color difference value decay curve calculation on each area color block of the display screen of the second-hand electronic device to obtain a screen color accuracy performance index; obtain a screen performance index based on the screen pixel display performance index, the screen touch response performance index, and the screen color accuracy performance index; construct a load spectrum based on a dynamic load test program simulating a real complex use scenario; perform calculation processing on the dynamic internal resistance of the battery of the second-hand electronic device at each circuit step instant of the load spectrum, and perform inversion processing on the battery health status of the second-hand electronic device through a recurrent neural network model to obtain a battery performance index; wherein the battery performance index includes actual maximum capacity, current SOH value, and predicted remaining cycle life interval; perform benchmark signal comparison and dynamic excitation testing on the sensor signal of the second-hand electronic device, and perform abnormal signal testing on the sensor signal through a pre-set Kalman filtering algorithm in a multi-dimensional fusion to obtain a sensor performance index; perform data throughput rate curve calculation processing on the radio frequency connection performance of the second-hand electronic device based on a standard signal source and a field strength meter to obtain a radio frequency performance index; perform index set processing on the interface performance index, the screen color accuracy performance index, the battery performance index, the sensor performance index, and the radio frequency performance index to obtain the hardware performance index data.

4. The method of claim 1, wherein, Before performing usage mode quantization analysis on the software system usage modal of the second-hand electronic device based on user authorization permissions in federated local learning, and collecting and obtaining software feature data based on account state information, the method further includes: Based on the software system usage mode, a temporary evaluation environment data package in a secure sandbox is constructed; wherein the temporary evaluation environment data package is a one-time and self-destructing data package; According to the device inherent information of the second-hand electronic device, the temporary evaluation environment data package is subjected to dynamic model loading processing related to the micro-federal learning model, and a security evaluation environment framework for the second-hand electronic device software system usage mode evaluation is obtained; Through the security evaluation environment framework, the power management system of the second-hand electronic device is subjected to data reading processing related to device usage intensity, and usage intensity characteristics are obtained; Through the security evaluation environment framework, the second-hand electronic device is subjected to storage operation of standardized storage read-write benchmark, and the order read-write speed decay rate, the occurrence frequency and type aggregation statistics of kernel-level abnormalities and application non-response in a specific time period are recorded, and performance decay characteristics are obtained; Through the security evaluation environment framework, the second-hand electronic device is subjected to network connection under the distribution proportion of network thermal safety type, and network environment characteristics are obtained; Based on the usage intensity characteristics, the performance decay characteristics and the network environment characteristics, local multi-source data characteristics are determined.

5. A method of assessing the residual value of a used electronic device according to claim 4, wherein, According to the user authorization permission, the software system usage mode of the second-hand electronic device is subjected to usage mode quantitative analysis under federal local learning, and software feature data is collected and obtained based on account state information, specifically including: Through a lightweight self-attention network model, the local multi-source data characteristics are subjected to device wear-out feature extraction processing related to attention mechanism, and a wear-out feature vector is output; wherein the wear-out feature vector is a fixed length data feature that cannot be reversely restored to any original behavior; The wear-out feature vector is subjected to homomorphic encryption processing to generate encrypted summary data transmitted to the cloud; and the sandbox environment corresponding to the security evaluation environment framework is automatically destroyed and controlled; According to the account state information verified by zero-knowledge proof, the encrypted summary data is subjected to query and aggregation processing of historical data sources under multi-source heterogeneity, and multi-source query data is obtained; wherein the historical data sources at least include: manufacturer official database, authorized service provider network, public market data platform and industry alliance blockchain; Each item of data in the multi-source query data is subjected to spatio-temporal alignment processing, and a time sequence knowledge graph is constructed based on event nodes and attribute nodes in the second-hand electronic device; wherein the event nodes at least include: production, activation, maintenance and transaction; the attribute nodes at least include: color and capacity; Each knowledge node in the time sequence knowledge graph is subjected to credibility tracing and weighting processing, and the software feature data after graph reasoning is obtained.

6. The method of claim 1, wherein, According to the appearance feature data, the hardware performance index data, the software feature data and the historical data label, the second-hand electronic device is subjected to device usage mode scoring processing under the microscopic state analysis network, and a comprehensive state score of the second-hand electronic device is obtained, specifically including: Mapping processing of all damage features in the appearance feature data is performed by a device damage position weight to obtain an appearance modal damage index; Each attribute item in the hardware performance index data is compared with a hardware benchmark parameter, and a single-item health degree is calculated. Each single-item health degree is marked with a risk in a decay mode to output a hardware modal health degree decay vector of a hardware health degree vector; The software feature vector in the software feature data is restored to a mode label, and a corresponding use intensity coefficient is determined based on the mode label. According to the use intensity coefficient and the software feature vector, a stability score processing of software modal use intensity is performed to obtain a software modal use intensity index; A historical modal signal event chain related to the explicit event node of the second-hand electronic equipment in the historical data label is extracted; The appearance modal damage index, the hardware modal health degree decay vector, the software modal use intensity index, and the historical modal signal event chain are mutually cross-verified and contradiction detected to determine a consistency evidence set and a corresponding comprehensive confidence; Each attribute object in the consistency evidence set is subjected to a failure derivation judgment related to other associated components, and each attribute object and failure derivation item is subjected to a risk value quantitative calculation related to residual value to obtain the comprehensive state score.

7. The method of claim 1, wherein, Through a macro value regression network, the comprehensive state score and the multi-dimensional value influencing factors of the second-hand electronic equipment are subjected to residual value evaluation processing related to a market residual value evaluation interval to obtain final residual value evaluation information, specifically including: Market quotation data of the second-hand electronic equipment to be evaluated is obtained by accessing a multi-source market API. The market quotation data at least includes real-time listing price, transaction price, and transaction speed. According to the market quotation data, market supply and demand index calculation related to similar configuration data of the second-hand electronic equipment is performed, and the market supply and demand index is corrected based on seasonal consumption cycle and event correction factor to obtain market dynamic data. Based on the market dynamic data, regional demand heat data is obtained. Through an NLP model, macro corpus is crawled and analyzed to obtain consumer preference data. The market dynamic data, the regional demand heat data, and the consumer preference data are combined to obtain multi-dimensional value influencing factors. Through a value evaluation expert network under a gate attention mechanism, residual value estimation processing between the multi-dimensional value influencing factors of the second-hand electronic equipment is performed to obtain a basic value estimate. Value conversion rate mapping calculation between specific score defect features in the comprehensive state score and the basic value estimate is performed, and value conversion rate mapping results are subjected to interval range evaluation processing through a market residual value evaluation interval to obtain the final residual value evaluation information. The market residual value evaluation interval is a real-time market value fluctuation interval based on multi-dimensional value influencing factors.

8. The method of claim 1, wherein, Based on the final residual value estimation information, the second-hand electronic equipment is automatically classified and processed, and a corresponding residual value evaluation report is generated, specifically including: According to the preset estimation interval, the residual value estimation in the final residual value estimation information is compared in intervals to determine the residual value grade of the second-hand electronic equipment; wherein the residual value grade includes: excellent, good and affordable; Based on the residual value grade, the second-hand electronic equipment is sorted to the corresponding storage compartment of the automatic detection cabin; Based on the residual value grade, the final residual value estimation information, the comprehensive state score, the hardware health chart, the three-dimensional appearance model and the software test result, the residual value evaluation report is generated.

9. A residual value assessment device for a second-hand electronic device, characterized by, The device comprises: At least one processor; and The memory is in communication connection with the at least one processor; wherein The memory stores instructions executable by the at least one processor, so that the at least one processor can execute the residual value evaluation method of the second-hand electronic equipment according to any one of claims 1-8.

10. A non-transitory computer storage medium, comprising, The storage medium is a non-volatile computer readable storage medium, and the non-volatile computer readable storage medium stores at least one program, and each program includes instructions, which make the terminal execute the residual value evaluation method of the second-hand electronic equipment according to any one of claims 1-8 when executed by the terminal.

Images