Building insulation board recycling and tracing method based on internet of things communication and blockchain technology

Abstract

The application discloses a building insulation board recycling traceability method based on Internet of Things communication and blockchain technology, and relates to the technical field of building environmental protection, and comprises the following steps: S1, at multiple transfer nodes of an insulation board recycling chain, a plurality of physical quantities of the insulation board having inherent physical conversion relationships are independently collected by Internet of Things collection equipment, at least including wet weight, external size and moisture content; S2, the self-consistency of the declaration parameters is verified through at least two independent derivation paths by using the physical quantities.The building insulation board recycling traceability method provided by the application independently collects physical quantities such as wet weight, volume and moisture content by using Internet of Things collection equipment, cross-verification is performed through the inherent physical conversion relationships between the quantities, high-precision detection equipment is not needed, ordinary industrial-grade sensors can meet the verification requirements, and online real-time verification can be realized in scenes such as a demolition site and a transfer station.

Description

Technical Field

[0001] This invention relates to the field of building environmental protection technology, specifically to a method for recycling and tracing building insulation boards based on Internet of Things communication and blockchain technology. Background Technology

[0002] Building insulation boards, mainly including EPS polystyrene boards and XPS extruded polystyrene boards, are the primary materials for building exterior wall insulation systems, generating a large amount of construction waste during demolition. Some insulation boards contain hexabromocyclododecane (HBCD) as a flame retardant during production. This substance is a persistent organic pollutant, and insulation boards containing HBCD should be managed and disposed of as hazardous waste, with disposal costs far exceeding those of ordinary construction waste. This cost difference leads to issues such as falsifying parameters during the recycling process, such as falsely reporting density to obtain subsidies and concealing hazardous waste attributes to evade disposal fees.

[0003] Currently, the traceability of building insulation board recycling mainly uses electronic manifests to record information on each stage of waste generation, transportation, transfer, and disposal. However, this primarily involves macro-level single-point verification based on weight and trajectory comparison, lacking a mechanism for cross-verification of the physical conversion relationships between the material's intrinsic properties, and failing to identify inherent contradictions between declared parameters. While laboratory testing technology can accurately measure individual physical parameters of insulation boards, the equipment is expensive and the testing cycle is long, making it difficult to deploy at demolition sites, transfer stations, and other handover nodes for online real-time verification. Furthermore, existing traceability systems focus on recording the flow path, lacking a mechanism for the reliable quantitative assessment of material property verification results and for their step-by-step transmission along the chain. Once classification information is given, it is transmitted downstream as definitive information, and downstream recipients cannot judge its reliability. This may lead to insulation boards containing HBCD being misclassified as ordinary construction waste, resulting in environmental risks and economic disputes. To address this, we propose a building insulation board recycling traceability method based on IoT communication and blockchain technology. Summary of the Invention

[0004] To address the problems raised in the background technology, this invention provides a method for recycling and tracing building insulation boards based on Internet of Things (IoT) communication and blockchain technology, specifically including the following steps: S1. At multiple junctions of the insulation board recycling chain, multiple physical quantities of the insulation board with inherent physical conversion relationships are independently collected by IoT data collection devices, including at least wet weight, dimensions and moisture content. S2. Using the physical quantity, perform self-consistency verification on the declared parameters through at least two independent derivation paths; based on the deviation comparison results of at least two paths, generate the parameter self-consistency verification result for this node. S3. Through blockchain smart contracts, combining the historical traceability credibility transmitted by upstream nodes with the parameter self-consistency verification results of this node, automatically calculate the updated current traceability credibility, and put the verification data of this node and the updated traceability credibility on the blockchain for evidence storage. S4. The downstream processing terminal reads the current traceability credibility on the chain and automatically matches the corresponding insulation board processing procedure according to the preset credibility range.

[0005] Preferably, the IoT data acquisition device includes: a dynamic weighing sensor, a three-dimensional vision scanning device or a laser ranging device, a microwave moisture sensor or an infrared moisture sensor, and an industrial camera; Each physical quantity is acquired by its own independent sensor. The wet weight is collected by the dynamic weighing sensor. The external dimensions are collected by the three-dimensional vision scanning device or laser rangefinder and the measured volume is obtained by geometric calculation. The moisture content is collected by the microwave moisture sensor or infrared moisture sensor. The inherent physical conversion relationships include: dry weight = wet weight / (1 + moisture content), dry density = dry weight / measured volume; the correction factor for moisture content is set according to the type of insulation board material, with EPS board corresponding to the first correction factor and XPS board corresponding to the second correction factor.

[0006] Preferably, the two independent derivation paths in S2 are as follows: The first approach: The dry weight is derived from the measured wet weight and the corrected measured moisture content. The dry weight is then divided by the measured volume to obtain the estimated dry density. The estimated dry density is then compared with the declared dry density. The second approach involves deriving the theoretical dry weight from the declared dry density and the measured volume, then using the theoretical dry weight combined with the corrected measured moisture content to deduce the theoretical wet weight, and finally comparing the deviation between the theoretical wet weight and the measured wet weight.

[0007] Preferably, step S2 further includes a third derivation path, specifically: The third derivation path is to derive the theoretical volume by dividing the dry weight by the declared dry density, and then compare the deviation between the theoretical volume and the measured volume. The deviation thresholds are set according to the type of insulation board material. When the deviations of the three derivation paths are all within the corresponding threshold range, the declaration parameters are deemed to be self-consistent. When the deviation of any path exceeds the threshold, the declaration parameters are deemed to be contradictory.

[0008] Preferably, when the material being transferred is bulk broken insulation board, the volume of the stack obtained from the three-dimensional scan is converted into an estimated solid volume by using the pre-calibrated relationship between the degree of breakage and the stacking coefficient, and then substituted into the derivation path to participate in the self-consistency verification.

[0009] Preferably, the appearance image of the insulation board is captured by an industrial camera, and the EPS and XPS material types are distinguished by image recognition; when the image recognition result does not match the declared material, an abnormal alarm is triggered directly and the traceability credibility is reduced.

[0010] Preferably, the update rule for the traceability credibility in S3 is as follows: When the cross-validation results of all core parameters are self-consistent, the source traceability confidence score is increased, and the increase is positively correlated with the number of self-consistent derivation paths. When the cross-validation results of any core parameter are contradictory, the source traceability credibility score is reduced, and the reduction is positively correlated with the degree of deviation. When non-core parameters are abnormal or data is missing, the source tracing credibility score remains unchanged, and it is marked as an incomplete information state. The traceability credibility is set with an upper and lower limit. The upper limit is less than 100%, and the lower limit is the mandatory testing threshold. When the traceability credibility is lower than the lower limit, the laboratory full-item testing process is automatically triggered.

[0011] Preferably, the source traceability credibility of the current node is obtained by weighted fusion based on the credibility adjustment value corresponding to the self-consistency verification result of the parameters of this node and the historical source traceability credibility transmitted by the upstream node. The weighting coefficients are determined by the device accuracy level and the sufficiency of sampling at the current node.

[0012] Preferably, at the first handover node of the insulation board recycling chain, an initial value of traceability credibility is set based on the initial classification results at the demolition site, wherein the traceability credibility adopts a continuous quantitative score from 0 to 100. In the initial classification, batches with clear material type, complete source identification, and classification conditions that meet the preset requirements are assigned an initial credibility score range corresponding to the first score range. For batches whose materials are bulk crushed materials, whose source identification is missing, or whose classification conditions do not meet the preset requirements in the initial classification, the initial confidence level corresponds to the second score range. The initial value is stored on the blockchain as the starting data for the traceability chain, and the traceability credibility updates of each subsequent node are all based on the credibility value of the preceding node and are passed down level by level.

[0013] Preferably, the content stored on the blockchain in S3 includes: the original IoT data, the derivation results and deviation values ​​of each derivation path, the parameter self-consistency verification conclusion, the values ​​before and after the traceability credibility update and the basis for the update, and the timestamp; the blockchain smart contract automatically triggers the credibility update and storage operation after the node completes the IoT data upload. In S4, the downstream processing terminal automatically matches the corresponding processing flow according to the range of traceability credibility: materials in the high credibility range directly enter the corresponding material recycling production line, materials in the medium credibility range are allocated to the sampling production line and the testing frequency and items are increased, and materials in the low credibility range are forcibly sent to the laboratory for full testing including HBCD content, and are classified and processed according to the actual results. The matching rules for the disposal process are pre-fixed by a blockchain smart contract. The disposal end can only read the matching results and has no right to modify the matching rules. The actual disposal results are sent back to the blockchain for on-chain storage and evidence storage, forming a complete traceability closed loop.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: The proposed method for tracing the recycling of building insulation boards utilizes IoT devices to independently collect physical quantities such as wet weight, volume, and moisture content. Cross-verification is performed through the inherent physical conversion relationships between these quantities. This eliminates the need for high-precision testing equipment; ordinary industrial-grade sensors suffice for verification, reducing deployment costs at various handover nodes and facilitating real-time online verification at demolition sites and transfer stations. Cross-verification among multiple physical quantities reveals inconsistencies in the physical relationships of other parameters when a single parameter is falsified. The method improves the ability to identify parameter falsification by comparing the deviations of at least two derivation paths. Addressing the specific characteristics of insulation board recycling scenarios, mechanisms such as moisture content correction based on material type and bulk crushed material stacking coefficient correction are implemented to adapt to different working conditions and reduce the probability of misjudgments due to environmental factors or material morphology differences. Automatic trust updates and evidence storage are achieved through blockchain smart contracts. Original collected data and the derivation results of each path are stored on the blockchain, reducing the possibility of human intervention in the traceability results and ensuring the objectivity and traceability of the traceability process. Attached Figure Description

[0015] Figure 1 This is a flowchart of the building insulation board recycling and traceability method of the present invention. Detailed Implementation

[0016] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0017] Reference Figure 1 As shown, the method for recycling and tracing building insulation boards based on IoT communication and blockchain technology includes the following steps: S1. At multiple junctions of the insulation board recycling chain, multiple physical quantities of the insulation board with inherent physical conversion relationships are independently collected by IoT data collection devices, including at least wet weight, dimensions and moisture content. The IoT data acquisition devices include dynamic weighing sensors, 3D vision scanning devices or laser ranging devices, microwave moisture sensors or infrared moisture sensors, and industrial cameras; wet weight is acquired by dynamic weighing sensors, external dimensions are acquired by 3D vision scanning devices or laser ranging devices and the measured volume is obtained through geometric calculations, and moisture content is acquired by microwave moisture sensors or infrared moisture sensors. Each physical quantity is acquired by its own independent sensor, and there is no situation where the measurement result of one parameter is used to deduce another parameter; The measurement accuracy of the microwave moisture sensor in the low moisture content measurement range is characterized by the absolute moisture content deviation, that is, the absolute value of the difference between the measured value and the true value does not exceed a set upper limit. The basis for determining this upper limit is as follows: Moisture content is used in the dry weight conversion during cross-validation (dry weight = wet weight / (1 + moisture content)). The error in moisture content measurement, after error propagation, should not exceed the set proportion of the deviation threshold in the calculation of dry weight and dry density, so as to ensure that the moisture content measurement error does not constitute the main source of deviation exceeding the tolerance. Taking EPS board as an example, when the typical moisture content is 2%, if the absolute deviation of moisture content measurement is ±0.2%, the relative error transmitted to dry weight is approximately 0.2% / (1+0.02)≈0.2%, which is far below the deviation threshold of ±10% and will not have a substantial impact on the effectiveness of cross-validation. For example, a microwave moisture sensor with an absolute deviation of no more than ±0.2% in the range of 0-5% absolute moisture content is selected. This level of accuracy is sufficient to support the requirements of subsequent cross-validation for moisture content data, as determined by error propagation analysis.

[0018] The accuracy level of each sensor is selected according to the data reliability requirements of each node in the recycling chain. The accuracy of the weighing sensor should ensure that the weighing error does not exceed the set proportion of the measured value within the typical weight range of the insulation board. This proportion is determined by back-calculation based on the deviation tolerance of each derivation path in the subsequent cross-validation, so that the weighing error itself does not constitute the main source of deviation exceeding the tolerance. The dimensional measurement accuracy of the 3D vision scanning equipment or laser rangefinder is determined based on the typical external dimensions of the insulation board and the error propagation relationship of volume calculation, so that the relative error of volume estimation does not exceed the set proportion; the moisture content measurement accuracy of the microwave moisture sensor or infrared moisture sensor is determined based on the normal operating moisture content range of the insulation board and its influence on dry weight estimation. The specific process of obtaining the measured volume using a 3D vision scanning device is as follows: The structured light scanner scans the surface of the insulation board to obtain surface point cloud data. The point cloud data is preprocessed, including statistical filtering to remove outlier noise points and RANSAC algorithm-based removal of ground and background plane points. For the whole insulation board, the orthogonal projection method is used to project the filtered point cloud onto three orthogonal planes respectively. The convex hull boundary is taken on each projection plane to obtain the three dimensions of length, width and height. The three dimensions are multiplied to obtain the measured volume. For irregularly shaped damaged insulation boards, a voxelization method is used to divide the point cloud space into voxel meshes of equal size. The total number of voxels is counted, and the sum of the voxel volumes is taken as the measured volume. The method for determining the side length of the voxel mesh is as follows: The voxel side length is taken as the smaller of the following two constraint values. The first constraint value is determined by the minimum identifiable size of the damaged insulation board. The voxel side length is not greater than the set fraction of the minimum identifiable size to ensure that each identifiable feature is covered by at least a set number of voxels. The calibration method of this fraction is as follows: using a standard size block with a known volume as the calibration object, calculate the voxelized volume estimate under different fraction values, and take the minimum fraction value that makes the relative error of the volume estimate not exceed the set upper limit, so as to minimize the number of voxels and reduce the amount of calculation while meeting the accuracy requirements. The second constraint value is determined by the point cloud accuracy of the scanning device. The voxel side length should not be less than a set multiple of the scanning accuracy to avoid the point cloud noise being amplified into volume estimation error due to excessively fine voxel division. The calibration method for this multiple is as follows: scan the same target object multiple times under the same scanning conditions, count the standard deviation of the point cloud coordinates at each voxel position, and take the smallest multiple that makes the standard deviation of the point cloud coordinates within the voxel lower than the set ratio of the nominal scanning accuracy to ensure that the voxelization process does not amplify noise. When the first constraint value is less than the second constraint value, it indicates that the current scanning equipment accuracy cannot support the volume estimation of crushed materials at this level. A higher-precision scanning equipment should be used, or the voxel edge length should be increased, accepting the corresponding loss of accuracy. The formula for calculating the voxel edge length is: Voxel edge length = min(minimum recognizable size / k1, scanning accuracy × k2), where k1 and k2 are empirical coefficients. For example, for level 4, a single piece volume less than 100 cm³... 3 The smallest recognizable particle size is about 20mm. When the scanning accuracy is 3mm, k1=3, k2=5, the voxel side length is 6.7mm, which is rounded to 7mm.

[0019] The matching relationship between the minimum feature size of the insulation board and the scanning accuracy should be determined so that the voxel side length is not greater than the set ratio of the minimum identifiable size of the damaged insulation board, so as to ensure the geometric resolution of volume estimation, and at the same time, not greater than this value to avoid excessive calculation due to too many voxels. When laser ranging equipment is used as an alternative, uniformly distributed sampling points are selected on the surface of the insulation board. The volume is calculated by combining multi-point ranging with the geometric contour model of the board. The number of sampling points is determined by the geometric contour complexity index of the board, specifically: The complexity index is defined as the ratio of the actual perimeter of the board outline to the perimeter of the smallest rectangle surrounding the outline. The closer the ratio is to 1, the more regular the shape of the board. The larger the ratio is, the more irregular the shape and the more surface undulations there are. The basic sampling points cover nine locations: the four corners, the midpoints of the four sides, and the center of the board. When the complexity index exceeds the set threshold, an additional set of sampling points is added for each increment exceeding the set threshold. Each set of sampling points consists of four points, which are evenly distributed in each quadrant of the board. The maximum spacing between sampling points should not exceed a set proportion of the longest side length of the board material, so as to ensure that the fitting error of linear interpolation between adjacent sampling points to the surface undulations does not exceed a set proportion of the total allowable error of volume calculation. For example, the complexity index threshold is 1.2, and for every 0.1, an additional set of 4 sampling points is added, with the maximum spacing not exceeding 1 / 4 of the longest side length of the board material, and the linear interpolation fitting error not exceeding 1 / 3 of the total allowable error of volume calculation.

[0020] The microwave moisture sensor utilizes the correlation between the amplitude attenuation and phase shift of microwaves propagating in a water-containing medium and the moisture content to collect the surface moisture content of the insulation board. The industrial camera acquires cross-sectional images of the insulation board under fixed light source conditions. The shooting distance and shooting angle are determined according to the clarity requirements of the cross-sectional features. The shooting distance should ensure that the resolution of the cross-sectional texture in the image is sufficient to support the subsequent material classification and recognition by the convolutional neural network. The shooting angle is perpendicular to the cross-section to minimize perspective distortion. The inherent physical conversion relationships include: dry weight = wet weight / (1 + moisture content), where the moisture content is expressed in decimal form, and for example, a moisture content of 3% is taken as 0.03 when substituted into the calculation; dry density = dry weight / measured volume; the correction factor for moisture content is set according to the water absorption characteristics of the insulation board material type. The physical meaning of the correction factor is the systematic deviation correction factor between the measured moisture content on site and the moisture content under standard testing conditions. Its value is determined based on the moisture content deviation characteristics of each material under typical on-site conditions. EPS board has a closed-cell structure with low water absorption rate, and the volumetric water absorption rate is usually less than 2%. The systematic deviation between the on-site non-equilibrium moisture content and the moisture content under standard testing conditions is negligible. The corresponding first correction coefficient is, for example, 1.0, which means that the on-site measured moisture content can be directly used for calculation without additional correction. XPS boards absorb almost no water, with a volumetric water absorption rate typically less than 1%. However, due to their higher closed-cell ratio, there is a slight systematic difference between the on-site non-equilibrium moisture content and the moisture content under standard testing conditions. A second correction factor, exemplarily set to 0.95, is used to correct this systematic deviation. The specific value of the correction factor can be further refined through calibration experiments. The correction factor was determined through the following calibration experiment: Select no less than a set number of insulation board samples. This number is determined based on the statistical stability of the correction coefficient estimate and the significance test requirements for the differences between materials. For example, select no less than 30 samples and cover no less than 3 different batches. After the moisture content of each sample was collected at the recycling site using a microwave moisture sensor, it was immediately sealed and sent to the laboratory. The laboratory conditioned the samples according to the standard curing conditions and then determined the standard moisture content. The standard curing conditions refer to the standard environment for conditioning plastic samples as specified in GB / T 2918-2018, namely, temperature 23±2℃, relative humidity 50±5%, and conditioning time of not less than 48h. For each sample, the correction factor calibration value is obtained by dividing the standard moisture content by the field moisture content. The physical meaning of the correction factor is that the field non-equilibrium moisture content is usually higher than the moisture content under standard curing conditions. The correction factor is less than 1 and is used to correct the field measured moisture content down to the standard testing condition level. Corrected moisture content = measured moisture content × correction factor. When the correction factor is taken as standard moisture content / situ moisture content, the corrected moisture content converges to the standard moisture content level; after removing outliers using the Grubbs criterion at a set significance level, the arithmetic mean of the remaining samples is taken as the calibration value of the correction factor for the material. The calibration test cycle is determined based on equipment replacement and changes in environmental conditions. When the data acquisition equipment is replaced, the sensor is recalibrated, or the climatic conditions of the environment change significantly, the calibration test should be repeated. The system automatically determines the material type by removing the material information or image recognition results entered on-site and calls the corresponding correction coefficient. The system simultaneously captures images of the insulation board's appearance using an industrial camera and employs a pre-trained convolutional neural network for material classification and recognition. The convolutional neural network is based on ResNet-18 and includes one input convolutional layer, four residual block groups, one global average pooling layer, and one fully connected output layer. The input layer receives RGB three-channel images uniformly scaled to 224×224 pixels, and the output layer is a binary softmax layer, corresponding to the two material categories EPS and XPS respectively. Before training, the input images are subjected to mean normalization and random horizontal flipping data augmentation preprocessing. During training, the training set and validation set are divided according to a set ratio, for example, an 8:2 ratio; the loss function is cross-entropy loss, the optimization algorithm is Adam algorithm, the initial learning rate is set to a set value, for example, 0.001, and the batch size is set to a set value, for example, 32; a transfer learning strategy is adopted, the network parameters are initialized with ImageNet pre-trained weights, the last two fully connected layers are fine-tuned, and the parameters of the preceding convolutional layers are updated with a lower learning rate; The number of training epochs should be no less than a set value. The convergence criterion is that the validation set accuracy is continuously no less than a set value, and the change in the number of epochs does not exceed a set percentage. For example, the number of training epochs should be no less than 50, and the convergence criterion is that the change in the validation set accuracy does not exceed 0.5% for 5 consecutive epochs. The number of training samples for the model is determined based on the recognition accuracy required for the material classification task and the statistical separability of the feature differences between the two types of materials. The model's recognition accuracy on the independent validation set should not be lower than a set percentage, which is set based on the following: Material identification results directly trigger confidence adjustment. If the identification is incorrect, it will lead to errors in subsequent correction coefficient calls and deviations in cross-validation conclusions. Therefore, the identification accuracy should be significantly higher than the upper limit of the confidence adjustment's tolerance for material identification errors. The model training samples include two classes: EPS and XPS. The number of samples in each class is determined by cross-validation according to the accuracy requirements mentioned above. The typical feature of EPS is white bead-like cross-sectional texture, and the typical feature of XPS is colored continuous bubble structure. When the image recognition result does not match the declared material, the system directly triggers an anomaly alarm and lowers the traceability credibility without waiting for subsequent parameter cross-validation results; For the handover of bulk broken insulation boards, the 3D vision scanning equipment obtains the stacked volume rather than the physical volume of the material, so the stacking coefficient needs to be corrected. The boundaries for classifying the degree of crushing are determined based on the statistical characteristics of the actual particle size distribution of bulk insulation boards at the recycling site, specifically as follows: Particle size statistics were performed on bulk insulation boards collected from a large number of recycling sites. The particle size distribution usually exhibits a multi-peak characteristic, with obvious distribution inflection points near the set particle size value. These inflection points are used as grade boundaries to ensure that the materials in each grade have similar stacking behavior. For example, the degree of fragmentation is divided into four levels according to the range of the volume of the individual fragments: Level 1 is large pieces with a volume greater than 5000 cm³, Level 2 is medium pieces with a volume between 1000 and 5000 cm³, Level 3 is small pieces with a volume between 100 and 1000 cm³, and Level 4 is granular material with a volume less than 100 cm³. The packing coefficient is based on a standard metal container, and the container size is determined according to the maximum external dimensions of the insulation board. The inner diameter of each dimension of the container should not be less than a set multiple of the corresponding dimension of the largest single insulation board to ensure that the influence of the container wall effect on the packing coefficient is negligible. A sample of bulk insulation board of a specific crushing grade is dropped naturally into a container from a set height. This height is determined based on the typical drop of bulk materials at the recycling site and should simulate the free fall state under actual loading and unloading conditions. The surface is leveled so that the top surface of the material is flush with the container opening. At this time, the accumulated volume of the material in the container is equal to the container volume. The dry weight of the material is weighed. The formula for calculating the packing factor under the corresponding grade is: Packing factor = Dry weight of material in container / (Container volume × Standard density of the material). The physical meaning of this formula is: The dry weight of the material in the container divided by the standard density of the material gives the solid volume of the material in the container. The solid volume divided by the container volume gives the packing factor, which is the proportion of the volume actually occupied by the solid material in the accumulated volume.

[0021] Each level undergoes a set number of parallel trials, the number of which is determined based on the statistical stability of the packing coefficient estimate. When the change in the arithmetic mean caused by a new trial is less than a set proportion, it is considered that the number of trials is sufficient. For example, each level undergoes 5 parallel trials, each using a new sample, and the arithmetic mean of the 5 results is taken as the final packing coefficient calibration value for that level. The calibration results are stored in the database of the edge computing unit in the form of a graded correspondence table; the exemplary calibration data are as follows: EPS material grade 1 corresponds to a packing factor of 0.82, grade 2 corresponds to 0.71, grade 3 corresponds to 0.63, and grade 4 corresponds to 0.55; XPS material grade 1 corresponds to a packing factor of 0.85, grade 2 corresponds to 0.74, grade 3 corresponds to 0.66, and grade 4 corresponds to 0.58. When determining the degree of breakage on-site, the material image can be obtained through 3D vision scanning, and the image analysis module can automatically identify the volume of each piece of material and calculate the average volume to determine the grade. Alternatively, the operator can manually observe and enter the grade. The packing coefficient of the corresponding material and grade is called to convert the packing volume into the estimated solid volume, that is: estimated solid volume = packing volume × packing coefficient, which is then substituted into the subsequent derivation path to participate in the self-consistency verification. In the scenario of bulk crushed materials, the deviation threshold of each derivation path is multiplied by a relaxation factor based on the overall plate threshold. This relaxation factor is determined based on the repeatability error of the packing factor calibration experiment, that is, the ratio of the standard deviation to the mean of the parallel test results of each level. The relaxed threshold should be able to cover the additional volume error introduced by the packing factor conversion. For example, the relaxation factor is 1.5. S2. Using independently collected physical quantities, verify the self-consistency of the declared parameters through at least two independent derivation paths. The first path density dimension verification specifically involves deriving the dry weight from the measured wet weight and the corrected measured moisture content: The estimated dry density is obtained by dividing the dry weight by the measured volume. The calculated dry density was compared with the declared dry density, and the deviation was: In the formula, To calculate dry weight, For actual wet weight, This is the corrected measured moisture content. This is the estimated value for dry density. For the actual measured volume, For the relative deviation of the density dimension, For the purpose of declaring dry density; The second path involves verifying the weight dimension, specifically by deriving the theoretical dry weight from the declared dry density and the measured volume. The theoretical wet weight is derived by combining the theoretical dry weight with the corrected measured moisture content: The deviation between the theoretical wet weight and the measured wet weight is compared, and the deviation is: In the formula, For theoretical purposes, Theoretical wet weight, This refers to the relative deviation in the weight dimension; The added third derivation path for volume dimension verification is derived by dividing the dry weight by the declared dry density to derive the theoretical volume: The deviation between the theoretical volume and the measured volume is compared, and the deviation is: In the formula, This is due to the relative deviation in volume dimension. Theoretical volume; The three derivation paths together form a triangular interlocking verification loop. Its core principle is: a forger may modify a single declared parameter, but cannot make all the derivation results of three different dimensions pass the verification at the same time, because wet weight, moisture content and volume are all objective physical quantities collected by independent sensors. Forgers cannot predict these measured values, and therefore cannot reverse the process to construct a false declaration that is completely self-consistent in terms of physical relationships. The deviation threshold is set according to the type of insulation board material. The setting is determined by adding the allowable deviation of product performance specified in the national standard for each material to the comprehensive measurement error of the on-site IoT testing equipment, so that the deviation threshold can distinguish between normal test fluctuations and substantial parameter tampering. Specifically, the lower limit of the deviation threshold is determined by the comprehensive measurement error of the field equipment, and should not be less than the total relative error after the errors of each independent sensor are synthesized by the error transmission formula, so as to avoid normal detection fluctuations being misjudged as parameter contradictions; the upper limit of the deviation threshold is determined by the national standard allowable deviation, and should not be greater than the sum of the national standard allowable deviation and the comprehensive measurement error of the equipment, so as to ensure that counterfeit and substandard products exceeding the national standard allowable range can be effectively identified. The density, wet weight, and volume deviation thresholds for EPS boards are determined based on the current GB / T 10801.1-2025 standard, combined with the industry-standard nominal density internal control fluctuation range for EPS boards, and the comprehensive measurement accuracy error of on-site IoT testing equipment. For example, the unified deviation threshold for the three channels in the internal control verification is ±10%. The deviation thresholds for XPS boards are determined based on the current GB / T10801.2-2025 standard, combined with the industry-standard nominal density internal control fluctuation range for XPS boards, and the comprehensive accuracy error of on-site testing equipment. For example, the unified deviation threshold for the three channels in the internal control verification is ±8%.

[0022] The volume definitions of bulk crushed materials and shaped plates are different. It is necessary to first calculate the repeatability error through the packing coefficient calibration test. The above-mentioned calibration threshold for shaped plates is then multiplied by the relaxation factor of the matching repeatability error to serve as the special calibration threshold for bulk crushed materials.

[0023] When the deviations of the three derivation paths are all within the corresponding threshold range, the declaration parameters are determined to be self-consistent; when the deviation of any path exceeds the threshold, the declaration parameters are determined to be contradictory, and an alarm is triggered. S3. Through blockchain smart contracts, combining the historical traceability credibility transmitted by upstream nodes with the parameter self-consistency verification results of this node, automatically calculate the updated current traceability credibility, and put the verification data of this node and the updated traceability credibility on the blockchain for evidence storage. Traceability reliability uses a continuous quantitative score to represent the degree of reliability of the attribute information of the current batch of materials in the traceability chain; the upper limit of the reliability score is determined based on the objective limitations of attribute confirmation under on-site conditions. Compared with standard laboratory testing, on-site IoT testing has inherent differences in equipment accuracy, environmental interference, and operating procedures. These differences make it impossible to fully confirm the authenticity of material properties under on-site conditions. Therefore, the upper limit should be lower than the full score to reflect the lower limit of the inherent uncertainty of on-site testing. For example, the upper limit is set at 95 points, and no further upward adjustment will be made after reaching 95 points. The lower limit of the credibility score is determined based on the trigger logic of mandatory testing. When the credibility score drops to this value or below, it indicates that the current on-chain information is insufficient to support attribute determination and must be confirmed by full laboratory testing. This value should make the historical attribute determination accuracy of batches below this value unacceptable and require laboratory testing intervention. For example, the lower limit is set to 30 points, which is the threshold for mandatory testing. When the traceability credibility score drops to 30 points or below, the system automatically generates a mandatory testing instruction and pushes it to the disposal terminal, requiring the batch of materials to be sent to the laboratory for full testing, including HBCD content. The core parameters are the physical quantities that directly participate in the above cross-derivation verification, including four categories: dry density, wet weight, moisture content, and volume. The non-core parameters are auxiliary parameters that do not participate in the self-consistency calculation, including the degree of appearance damage, the degree of surface contamination, and local thickness deviation. When the core parameters are abnormal, the credibility score is lowered. When the non-core parameters are abnormal or the data is missing, the credibility score remains unchanged, but it is marked as an incomplete information state in the traceability record, prompting downstream nodes to pay attention. At the first handover point in the insulation board recycling chain, namely the demolition site, an initial value for traceability credibility is set based on the initial classification results. The initial value is set based on the historical accuracy statistics of the initial classification at the site, and the historical accuracy is mapped to the credibility score range so that the initial credibility reflects the prior estimate of the reliability of information under the classification conditions. The mapping rule is that the higher the initial classification accuracy, the closer the initial credibility is to the upper half of the credibility range. For example, the initial classification accuracy of a batch with complete flasks and intact source identification is approximately 92%, corresponding to an initial confidence level of 80 points, which is in the upper-middle level of the confidence level range; the initial classification accuracy of a batch with bulk crushed material or missing source identification is approximately 60%, corresponding to an initial confidence level of 50 points, which is in the lower-middle level of the confidence level range; and the initial confidence level of a batch with complete flasks but missing identification is set at 65 points, which is in the middle of the above two ranges. Operators input basic batch information on-site via terminal devices. The system automatically matches the corresponding initial credibility score based on the input information, and stores it on the blockchain along with the batch number and timestamp as the starting data for the traceability chain. The subsequent credibility update rules for each handover node are as follows: When the cross-validation results of all core parameters are self-consistent, the source tracing credibility score is increased, and the increase is positively correlated with the number of self-consistent derivation paths; the increase formula is: In the formula, To increase credibility, the score was adjusted. To ensure a self-consistent and effective verification of the number of paths, To increase the credibility, the base coefficient was adjusted upwards. This is the average of the ratios of each path deviation value to its corresponding threshold, i.e.: In the formula, The corresponding dimension deviation threshold; The determination method is as follows: taking the actual deployed recycling chain as the calibration object, the total number of handover nodes in the chain, the typical equipment accuracy level and corresponding typical weight coefficient of each node, and the typical deviation level of each node when all three paths are self-consistent; define the following two quantitative evaluation indicators. The first indicator is the margin gap, which is the difference between the upper limit value and the final credibility after all nodes are verified. The margin gap should not be less than the set value to ensure that the final credibility leaves enough space from the upper limit to maintain the distinguishability of subsequent anomalies. For example, the margin gap should not be less than 5 points. The second indicator is the cumulative increase, which is the difference between the final credibility score and the initial value. The cumulative increase should not be less than a set value to ensure that the cumulative trust effect of continuous compliance can be effectively observed. For example, the cumulative increase should not be less than 8 points. When calibrating, it is based on... The initial value is incremented by a set step size for each... The forward simulation is run with the following input parameters: the total number of chain nodes, the equipment accuracy level and corresponding weight coefficient of each node, and the typical deviation level of each node when all three paths are self-consistent. The simulation output is the final confidence value. Calculate each The margin interval and cumulative increase corresponding to the selected values ​​are chosen as the minimum that simultaneously satisfies both requirements. The value is used as a calibration result to avoid oversensitivity while meeting design requirements; When the average number of nodes in the recycling chain changes or the accuracy level of the node equipment is adjusted, it should be recalibrated. For example, positive simulation results show that in a typical recycling chain containing approximately four handover nodes, with an average weight coefficient of approximately 0.4 for each node, and with each node being self-consistent across all three paths and exhibiting low deviation levels, Setting it to 2.0 allows the final confidence score to accumulate from the initial value of 80 to the range of approximately 88-90, leaving a margin of about 5-7 points from the upper limit of 95, thus meeting the above design requirements.

[0024] When the cross-validation results of any core parameter are contradictory, the source tracing credibility score is lowered, and the magnitude of the reduction is positively correlated with the degree of deviation; the reduction formula is: In the formula, The score was lowered for a single path. To lower the coefficient, This is the value obtained by subtracting 1 from the ratio of the deviation value of the out-of-tolerance path to the corresponding threshold, i.e.: In the formula, This is the single-path error margin coefficient; a value greater than 0 indicates that the path exceeds the allowable threshold. This represents the relative deviation of the measured values ​​for a single path. If multiple paths exceed the tolerance simultaneously, the maximum downward adjustment value among all paths is taken; the square relationship is used so that the downward adjustment accelerates as the deviation increases, reflecting the increasing punishment for serious fraud. The method for determining it is as follows: The penalty gradient is determined based on the desired penalty gradient; the design objective is that when the deviation just exceeds the threshold, the reduction magnitude should be small to provide a tolerance within a reasonable error range; twice the threshold value is taken as the reference calibration point, and the expected reduction magnitude at this point is determined according to the design objective. A slight penalty should be applied when the deviation just exceeds the threshold to preserve tolerance; when the deviation reaches twice the threshold, a sufficiently significant downsampling should be applied to quickly move the confidence level out of the high confidence range. This can be determined by inversely calculating the expected downsampling magnitude from the reference calibration point. Then, it is verified whether the downward adjustment under a smaller deviation is within an acceptable tolerance range; for example, take The expected value was lowered by about 8 points, which yielded the following result. ;verify: The score has dropped by about 2 points. The score was reduced by approximately 0.72 points. This gradient distribution conforms to the design principle of not imposing heavy penalties for small deviations and accelerating penalties for large deviations.

[0025] After completing its own verification and credibility adjustment, each handover node needs to perform a weighted fusion of its credibility adjustment result with the historical credibility transmitted upstream to calculate the final traceability credibility of the current node; the weighted fusion formula is as follows: In the formula, This determines the final credibility of this node. The current node's fusion weight, Historical credibility transmitted upstream The algebraic sum of the confidence adjustment value of this node, i.e.: The credibility of historical tracing transmitted to upstream nodes; weighting coefficients The reliability of the verification conclusion of this node is determined by the accuracy level of the equipment and the sufficiency of sampling at the current node. The higher the equipment accuracy and the more sufficiency of sampling, the higher the reliability of the verification conclusion of this node and should be given greater weight. The accuracy level of the equipment is classified according to the proportion of the sensor's overall measurement error to the deviation threshold: the smaller the proportion of the overall measurement error to the deviation threshold, the higher the accuracy level. For example, the accuracy level of the equipment is divided into three levels: Level A is high-precision equipment, where the proportion of the sensor's overall measurement error to the deviation threshold does not exceed the set lower limit; Level B is standard-precision equipment, where the proportion is between the set lower limit and the set upper limit; and Level C is low-precision equipment, where the proportion exceeds the set upper limit. The adequacy of sampling is measured by the proportion of the number of samples in this batch to the total number of batches. The boundary of this proportion is determined based on the inflection point of the marginal contribution of sample size to estimation accuracy in the sampling statistics. The specific values ​​of the weighting coefficients are determined by the following methods: Using equipment accuracy level and sampling adequacy as two dimensions, based on the consistency rate between the verification conclusions and laboratory re-inspection conclusions under different combinations of accuracy level and sampling level in historical data, the combination with the highest consistency rate is assigned the largest weight, the combination with the lowest consistency rate is assigned the smallest weight, and the intermediate combinations are linearly interpolated according to the relative order of consistency rate; for example, w is 0.5 when it is a Class A equipment and the sampling ratio is greater than or equal to 30%, w is 0.45 when it is a Class A equipment and the sampling ratio is less than 30%, w is 0.4 when it is a Class B equipment and the sampling ratio is greater than or equal to 30%, w is 0.35 when it is a Class B equipment and the sampling ratio is less than 30%, and w is 0.3 regardless of the sampling ratio for Class C equipment; The traceability credibility interval is divided into three levels. The historical accuracy data of material attribute determination is traced back at different credibility levels. The set boundary value of historical accuracy is used as the basis for interval division, so that the materials in each interval have similar reliability of attribute determination and correspond to different handling procedures. The boundary values ​​of the confidence interval are determined by the following method: Collect the accuracy distribution of batch material attribute determination under different confidence scores in historical operation data, and use the set boundary value of accuracy as the basis for interval division, so that the materials in each interval have similar reliability of attribute determination. The lower limit of the high confidence interval is the lowest confidence score that ensures the historical accuracy rate is not lower than the set upper limit; the lower limit of the medium confidence interval is the lowest confidence score that ensures the historical accuracy rate is not lower than the set lower limit; and the lower limit of the low confidence interval is the same as the confidence lower limit. After the boundary values ​​are calibrated, they need to be verified through positive simulation to confirm that the processing procedures corresponding to each interval can effectively distinguish batches with different reliability levels. For example, the high reliability interval is 80 to 95 points, and the historical judgment accuracy rate of batches in this interval is above 95%, so they can directly enter the recycling line; the medium reliability interval is 50 to 80 points, and the historical judgment accuracy rate of batches in this interval is between 75% and 95%, so additional sampling is required to confirm the attributes; the low reliability interval is 30 to 50 points, and the historical judgment accuracy rate of batches in this interval is below 75%, so the attributes must be confirmed through full-item laboratory testing; the interval boundaries are recalibrated based on actual operating data after the system has accumulated a set running time.

[0026] The content stored on the blockchain includes the original IoT data, the derivation results and deviation values ​​of each derivation path, the parameter self-consistency verification conclusion, the values ​​before and after the traceability credibility update and the basis for the update, and the timestamp; the blockchain smart contract automatically triggers the credibility update and storage operation after the node completes the IoT data upload; this embodiment adopts a consortium blockchain deployment, with independent nodes running at the site, transfer station, recycling plant and regulator, and each node synchronizes data through a consensus mechanism; Smart contracts contain three core functions: a data upload and verification function, a credibility update function, and a disposal instruction generation function.

[0027] The data upload verification function takes batch number, node identifier, array of original collected data, and array of verification results as input parameters. The processing logic of each core function is as follows: The data format verification rules of the uploadData function include: verifying whether the batch number conforms to the preset coding rules, whether the node identifier is a valid identifier in the registered node list, whether each physical quantity field in the original collected data array is complete and the value type is correct, and whether the value of each physical quantity is within a reasonable physical range, such as moisture content not less than 0, volume not less than 0, and wet weight not less than 0. The integrity check rules include: verifying whether the length of the verification result array is consistent with the number of derivation paths, and whether each deviation value can be independently recalculated and verified with the original collected data; when the data format verification or integrity check fails, the function refuses to write and returns an error code and the specific reason for failure; the processing logic of the confidence update function updateConfidence is as follows: Read the upstream trustworthiness value and device accuracy level stored on the chain. Based on the device accuracy level and sampling ratio, look up the weight coefficient correspondence table to obtain the current node's weight w. Calculate the trustworthiness adjustment value for this node according to the trustworthiness adjustment formula (either upward or downward). Calculate the temporary credibility after single-node adjustment, and then calculate the final credibility according to the weighted fusion formula. Write to on-chain storage and return the updated trust value; When the calculation result exceeds the upper and lower limits of confidence, it is automatically truncated to the corresponding boundary value; The processing logic of the `generateDisposalOrder` function is as follows: read the current confidence value, match the corresponding disposal process identifier according to the interval division rules, output the disposal process identifier "normal disposal" when the current confidence is in the high confidence interval, output the sampling inspection and verification when it is in the medium confidence interval, and output the mandatory inspection when it is in the low confidence interval. The execution order of the smart contract is as follows: After the node's IoT data is uploaded, uploadData is automatically triggered to perform data format verification and integrity checks. After the verification is successful, updateConfidence is automatically triggered to perform a confidence update. After the confidence update is completed, generateDisposalOrder is automatically triggered to generate a disposal instruction. The three functions are executed serially in this order. If any function fails, subsequent function calls are terminated and an error log is recorded.

[0028] The permissions for each participating node are divided as follows: The demolition site nodes and transfer station nodes have data upload permissions and credibility update permissions; the recycling plant nodes have data read permissions and disposal result feedback permissions; the monitoring nodes have full read and write permissions; the on-chain data is stored in structured JSON format.

[0029] S4. The downstream processing terminal reads the current traceability credibility on the chain and automatically matches the corresponding insulation board processing procedure according to the preset credibility range. Materials in the high-confidence range are directly sent to the corresponding material recycling production line for routine processing. Materials in the medium confidence range are allocated to the sampling inspection production line for sampling inspection and verification: for batches that pass the sampling inspection, the traceability confidence score is corrected and the on-chain data is updated, and then they are transferred to the corresponding material recycling production line; for batches that fail the sampling inspection, the confidence score is recalculated and the range is reclassified based on the latest measured data. If the confidence score falls into the low confidence range, the mandatory laboratory testing process is initiated.

[0030] The sampling frequency and sample size are determined according to GB / T 2828.1-2012 Sampling Inspection Procedures for Attributes, using General Inspection Level II. The selection is based on a comprehensive consideration of the typical weight range of the insulation board recycling batch and on-site testing cost constraints, ensuring that the sampled materials have batch spatial representativeness and that the density retest results have statistical stability. For example, one sample set is taken for every 5 tons of material, with each set containing no fewer than 3 boards; for a single batch of 10 tons of material, two sets can be taken, totaling no fewer than 6 boards, covering sampling of materials from different locations within the batch. The credibility adjustment for batches that pass the medium-credibility sampling inspection follows the principle of appropriate trust rewards and no skipping of levels. Passing the sampling inspection means that the on-chain traceability data has been independently sampled and verified to be basically credible, and a small credibility adjustment is given. After the adjustment, the score will not exceed the upper limit of the medium-credibility range, indicating that the sampling verification is credible, but the trust level has not undergone full laboratory testing. For example, the adjustment range is 3 points. Materials in the low confidence range are required to be sent to the laboratory for full testing, including indicators such as density, thermal conductivity, combustion performance, and HBCD content. If the test results are qualified, the materials will be reclassified and disposed of according to the actual test results. If the test results are unqualified, the materials will be disposed of in accordance with the requirements for hazardous waste management. Materials in the low confidence range are prohibited from skipping the testing and being directly processed or disposed of. The system will automatically lock the mandatory testing instruction.

[0031] All tiered handling matching rules are pre-fixed and tamper-proof by blockchain smart contracts. The handling terminal can only read and execute them, and has no right to modify the verification and tiering logic. All actual handling results, sampling results, and re-inspection results are all transmitted back to the blockchain for evidence storage.

[0032] When a single handover node experiences partial data loss, the following rules apply: If the missing parameter is a core parameter, such as a moisture content sensor malfunction causing missing moisture content data, the traceability credibility score will be lowered and the system will mark it as incomplete. At the same time, the system will automatically switch the verification mode and only use the remaining calculable derivation paths for verification. For example, if the moisture content is missing, the first and second paths cannot be calculated because they depend on the moisture content. Only the third path, which derives the theoretical volume from the dry weight and declared density, will be retained. In this case, the dry weight cannot be derived from the wet weight and moisture content. It must be derived from the declared density and measured volume to derive the theoretical dry weight. If the verification of a single path passes, the credibility score will remain lowered and will not be lowered further. If the verification of a single path also fails, the score will be lowered according to the normal downgrade rules. The basis for determining the confidence score reduction when core parameters are missing is as follows: the absence of one core parameter causes the triangular interlock verification to be reduced to single-path verification, and the sufficiency of the verification is significantly reduced. The score reduction should reflect this loss of sufficiency. The value is determined based on the historical error rate increment of single-path verification compared to three-path verification, so that the confidence level after the reduction matches the actual reliability of single-path verification. For example, the score reduction is 5 points. If the missing parameter is a non-core parameter, such as the absence of appearance images due to industrial camera failure, the confidence remains unchanged, but the marked image data is missing. When two core parameters are missing at the same time, since all three derivation paths require wet weight, moisture content and measured volume as necessary input parameters, no derivation path can complete the full calculation after the absence of any two measured parameters. The traceability confidence is directly reduced to the lower limit value, and the system automatically generates a mandatory inspection instruction. When one core parameter and one non-core parameter are both missing, the system will handle it according to the rules for missing a single core parameter. Missing non-core parameters will only mark the information as incomplete, without any additional reduction in credibility. When all three core parameters are missing, cross-validation will be completely impossible, the credibility of traceability will be directly reduced to the lower limit, and the system will automatically generate a mandatory inspection instruction.

[0033] When material information is completely missing, i.e. no material information was recorded at the demolition site and the industrial camera cannot complete the identification due to equipment failure or insufficient image quality, the system cannot determine the moisture content correction coefficient and deviation threshold to be called. At this time, the cross-validation lacks the necessary material-related parameter configuration, and the validation cannot be effectively performed. The processing rules are as follows: The system first attempts to perform a complete three-path cross-validation once each for the correction coefficients and deviation thresholds of EPS and XPS materials. If both material assumptions are deemed self-consistent, the lower of the confidence scores calculated under the two material assumptions is taken as the confidence update result of this node, ensuring that false trust is not generated due to material misjudgment based on the principle of conservatism. If only one material assumption is self-consistent while the other is out of tolerance, the validation result under the self-consistent material assumption is executed, and the material is marked as pending confirmation, prompting downstream nodes to pay attention. If both material assumptions are out of tolerance, the judgment parameters are contradictory, and the traceability confidence is lowered according to the normal downward adjustment rules. If the lower limit is reached after the adjustment, mandatory inspection is triggered.

[0034] When the verification results of two adjacent handover nodes fluctuate significantly, the review mechanism is triggered. The so-called large fluctuations refer to the fact that, based on the credibility update data during normal operation, the upward adjustment of a single node verification under self-consistency is usually within the set range after weighted fusion, and the downward adjustment when there is a slight deviation is also within the set range. In normal operation, the absolute value of the credibility adjustment value has a statistical upper limit. Setting the review trigger threshold to a multiple of the upper limit of the normal adjustment range can effectively distinguish between normal fluctuations and abnormal events. For example, the review trigger threshold is 10 points, which is about twice the upper limit of the normal adjustment range. After the review is triggered, the system sends a review request to the regulatory node, which arranges a re-inspection of the batch of materials. The re-inspection result is used as the final judgment value to cover the current node's credibility adjustment value, and the re-inspection data is stored on the blockchain for evidence. When materials from different traceability credibility batches need to be mixed during the circulation process, a mixed-packing conversion is performed; The traceability reliability of the new batch after mixing is calculated based on the weight-weighted average of the original batches, that is: in, To ensure batch reliability after mixing, and These represent the traceability credibility and weight of the i-th original batch, respectively. If the mixed-packing calculation result crosses the interval boundary, the disposal process is executed according to the lower interval to ensure disposal safety with a conservative principle. The mixed-packing event and the conversion process are recorded on the blockchain as new nodes for evidence storage. When weight data for some original batches is missing during the mixing process due to weighing equipment malfunction or communication interruption, the following rules shall apply: The batches with known weights and the batches with unknown weights are grouped separately. The batches with known weights are calculated using the normal weighting formula. The weight of the batches with unknown weights is based on the weight data recorded in the nearest upstream node. If there is no record in the upstream node, the declared weight of the batch is used as a substitute value in the weighting calculation. At the same time, the source of the weight data is marked as a substitute value in the mixed loading record so that downstream nodes can identify the difference in data reliability. The credibility after mixing is still calculated using the weight-weighted average formula, but the source of the weight data of each batch involved in the calculation must be noted in the on-chain record to distinguish between the measured weight and the substitute value, so that downstream nodes can assess the reliability of the mixed credibility accordingly.

[0035] When IoT data upload times out, if a single node's IoT acquisition device completes data acquisition but fails to upload the data to the blockchain node within the preset time window due to network interruption or communication failure, the system will initiate a timeout handling process. Under normal network conditions, the latency from data collection to successful upload for IoT data is typically in the second range. The time window should be significantly longer than the normal latency to eliminate misjudgments caused by normal network fluctuations, but it should not be too long to avoid delaying decision-making. For example, the time window is set to 30 minutes. After the first timeout, the edge computing unit automatically caches the data and retryes the upload. The retry interval and the number of retries are determined based on the typical duration and recovery characteristics of network faults. The retry interval should allow sufficient time between each retrieval to distinguish between transient jitter and persistent faults, and the number of retries should ensure that the cumulative retry time covers the recovery time of most recoverable network faults. For example, the retry interval is 5 minutes, with a maximum of 3 retries. If the upload still fails after retrying, the system will record the timeout event in the local log and mark the credibility of this batch as pending confirmation. The credibility update and processing instructions will not be generated for the time being. The data will be re-uploaded and the verification process will be executed normally after the network is restored. When uploading supplementary data to the blockchain, the original acquisition timestamp and the supplementary transmission timestamp must be recorded separately for downstream nodes to identify the timeliness of the data. If the timeout period exceeds the preset maximum tolerance time limit, the data of that node is considered unusable and is handled according to the rules for missing core parameters. The maximum tolerance time limit is determined based on the typical dwell time of the insulation board at a single handover node. After this time, the material has already flowed to the next stage, and the timeliness and relevance of the supplementary data are significantly reduced. For example, the maximum tolerance time limit is set to 24 hours. When a smart contract execution times out, each function has an execution timeout threshold set. This threshold is determined based on the statistical distribution of the function's execution time under normal load. It should be significantly greater than the normal execution time to exclude normal computational fluctuations, but it should not be too long to avoid prolonged suspension of blockchain transactions, which could affect subsequent processing. For example, the execution timeout threshold for a single function is set to 60 seconds. If a function execution times out, the blockchain node records the timeout log and triggers an automatic retry. The number of retries is determined based on the typical type of timeout cause. For timeouts caused by instantaneous load fluctuations, one retry is usually sufficient to recover. If the timeout still occurs after the retry, the transaction is marked as pending and manually investigated by the supervisory node. Before the timeout transaction is resolved, subsequent nodes in the same batch continue to perform verification using the last successfully updated on-chain credibility value as the upstream historical credibility, to avoid the entire chain from stalling due to an anomaly in a single node contract. When data synchronization between blockchain nodes times out, the nodes in the consortium blockchain synchronize data through a consensus mechanism. Under normal circumstances, the data synchronization delay has a statistical upper limit based on the type of consensus mechanism and network bandwidth. If a node fails to complete data synchronization with a majority of nodes after a set time has elapsed, the node is deemed to have a synchronization anomaly. The duration is determined by adding a network jitter margin to the set multiple of the normal synchronization delay. For example, the synchronization anomaly judgment duration is set to 5 minutes. During the synchronization anomaly, the node suspends its data upload permission and retains only read permission until the data catch-up is completed and the block height is restored to the same level as the majority of nodes, at which point the upload permission is automatically restored. When multiple abnormal events are triggered simultaneously on the same node, the following stacking rules apply: When material identification discrepancies and parameter cross-validation contradictions occur simultaneously, the credibility reduction is based on the larger of the two values ​​rather than a simple summation. This design is based on the fact that material identification discrepancies themselves suggest a fundamental problem with the declared information. In this case, parameter verification contradictions may be different manifestations of the same fraudulent behavior, and simply summing them would lead to duplicate penalties for the same act. The fixed reduction value for material identification discrepancies is determined as follows: Material identification errors will lead to errors in subsequent correction coefficient calls and deviations in cross-validation conclusions. The downward adjustment value should ensure that the resulting artificially high credibility does not exceed the typical value of the normal upward adjustment range of a single node, so as to avoid the false trust effect caused by incorrect identification exceeding the compliance accumulation of a normal node. By statistically analyzing the distribution of the increase in credibility of each node under normal compliance conditions, the median percentage is taken as the reference benchmark for the decrease value. The decrease value is verified through positive simulation to effectively offset the false increase in trust caused by material identification errors. For example, the fixed decrease value for material identification mismatch is 5 points, and the decrease value for parameter verification contradiction is calculated according to the normal formula. The maximum of the two is taken as the final decrease value.

[0036] When core parameter missing and parameter verification contradiction occur simultaneously, the adjustment is first made according to the data missing rule, and then the adjustment value of the out-of-tolerance part is calculated according to the normal adjustment rule. The two are superimposed. The design basis is that data missing reduces the sufficiency of verification, while the out-of-tolerance of the residual path further increases the degree of suspicion. The two types of anomalies are different in nature and should be superimposed. When non-core parameter anomalies occur simultaneously with any other anomalies, non-core parameter anomalies only indicate incomplete information and do not result in a decrease in confidence level. The return of results and on-chain information updates following mandatory testing shall be executed according to the following closed-loop rules: After batches with low credibility or whose credibility has dropped to the lower limit or below are forcibly sent to the laboratory, the laboratory will complete all tests and then send the test results back to the blockchain through the authorization interface of the regulatory node. The sent data will be stored on the blockchain as a new node record. The record includes the laboratory node identifier, test timestamp, measured values ​​of various test indicators, test conclusion, and test method standard number. If the test results are qualified and all indicators are consistent with the declared parameters, the traceability credibility of the batch will be reset to the set score. The determination of this score is based on the following: the laboratory's full inspection qualification indicates that the material properties have been confirmed with the highest level of reliability. However, considering that subsequent circulation still needs to go through handover nodes, the reset score should enable the batch to continue to accumulate credibility at a higher starting point in subsequent nodes, but it should not be directly reset to the upper limit value to maintain sensitivity to subsequent anomalies. This reset score is a configurable system parameter and can be periodically calibrated based on the credibility convergence data of subsequent nodes of the batches that are subject to mandatory testing in actual operation. For example, if the reset score is 75 points, the mandatory testing status is lifted, and the subsequent handling process is carried out according to the medium credibility range. If the test result is qualified but some indicators deviate from the declared parameters, the laboratory measured value is used instead of the declared value and substituted into the three derivation paths for re-verification, and the credibility is calculated according to the normal rules; if the test result is unqualified, the credibility of the batch is permanently locked at 0 points and marked as unqualified: hazardous waste disposal status. This record cannot be overwritten or modified by subsequent nodes to ensure that the mandatory testing results are final.

[0037] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for recycling and tracing building insulation boards based on Internet of Things (IoT) communication and blockchain technology, characterized in that: Includes the following steps: S1. At multiple junctions of the insulation board recycling chain, multiple physical quantities of the insulation board with inherent physical conversion relationships are independently collected by IoT data collection devices, including at least wet weight, dimensions and moisture content. S2. Using the physical quantity, perform self-consistency verification on the declared parameters through at least two independent derivation paths; based on the deviation comparison results of at least two paths, generate the parameter self-consistency verification result for this node. S3. Through blockchain smart contracts, combining the historical traceability credibility transmitted by upstream nodes with the parameter self-consistency verification results of this node, automatically calculate the updated current traceability credibility, and put the verification data of this node and the updated traceability credibility on the blockchain for evidence storage. S4. The downstream processing terminal reads the current traceability credibility on the chain and automatically matches the corresponding insulation board processing procedure according to the preset credibility range.

2. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 1, characterized in that, In S1: The IoT data acquisition devices include: dynamic weighing sensors, three-dimensional vision scanning devices or laser rangefinders, microwave moisture sensors or infrared moisture sensors, and industrial cameras. Each physical quantity is acquired by its own independent sensor. The wet weight is collected by the dynamic weighing sensor. The external dimensions are collected by the three-dimensional vision scanning device or laser rangefinder and the measured volume is obtained by geometric calculation. The moisture content is collected by the microwave moisture sensor or infrared moisture sensor. The inherent physical conversion relationships include: dry weight = wet weight / (1 + moisture content), dry density = dry weight / measured volume; the correction factor for moisture content is set according to the type of insulation board material, with EPS board corresponding to the first correction factor and XPS board corresponding to the second correction factor.

3. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 2, characterized in that, The two independent derivation paths in S2 are as follows: The first approach: The dry weight is derived from the measured wet weight and the corrected measured moisture content. The dry weight is then divided by the measured volume to obtain the estimated dry density. The estimated dry density is then compared with the declared dry density. The second approach involves deriving the theoretical dry weight from the declared dry density and the measured volume, then using the theoretical dry weight combined with the corrected measured moisture content to deduce the theoretical wet weight, and finally comparing the deviation between the theoretical wet weight and the measured wet weight.

4. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 3, characterized in that, The S2 also includes a third derivation path, specifically: The third derivation path is to derive the theoretical volume by dividing the dry weight by the declared dry density, and then compare the deviation between the theoretical volume and the measured volume. The deviation thresholds are set according to the type of insulation board material. When the deviations of the three derivation paths are all within the corresponding threshold range, the declaration parameters are deemed to be self-consistent. When the deviation of any path exceeds the threshold, the declaration parameters are deemed to be contradictory.

5. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 1, characterized in that, When the material being transferred is bulk broken insulation board, the volume of the stack obtained from the three-dimensional scan is converted into the estimated solid volume by using the pre-calibrated relationship between the degree of breakage and the stacking coefficient, and then substituted into the derivation path to participate in the self-consistency verification.

6. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 1, characterized in that, Industrial cameras are used to capture images of the insulation board's appearance, and image recognition is used to distinguish between EPS and XPS material types. When the image recognition result does not match the declared material, an anomaly alarm is triggered directly, and the traceability credibility is lowered.

7. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 1, characterized in that, The update rule for the source tracing credibility in S3 is as follows: When the cross-validation results of all core parameters are self-consistent, the source traceability confidence score is increased, and the increase is positively correlated with the number of self-consistent derivation paths. When the cross-validation results of any core parameter are contradictory, the source traceability credibility score is reduced, and the reduction is positively correlated with the degree of deviation. When non-core parameters are abnormal or data is missing, the source tracing credibility score remains unchanged, and it is marked as an incomplete information state. The traceability credibility is set with an upper and lower limit. The upper limit is less than 100%, and the lower limit is the mandatory testing threshold. When the traceability credibility is lower than the lower limit, the laboratory full-item testing process is automatically triggered.

8. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 7, characterized in that, In S3, the credibility adjustment value corresponding to the self-consistency verification result of the current node parameter is used as the basis, and the historical source tracing credibility transmitted by the upstream node is weighted and fused to obtain the source tracing credibility of the current node. The weighting coefficients are determined by the device accuracy level and the sufficiency of sampling at the current node.

9. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 1, characterized in that, At the first handover node of the insulation board recycling chain, an initial value for traceability credibility is set based on the initial classification results at the demolition site. The traceability credibility adopts a continuous quantitative score from 0 to 100. In the initial classification, batches with clear material type, complete source identification, and classification conditions that meet the preset requirements are assigned an initial credibility score range corresponding to the first score range. For batches whose materials are bulk crushed materials, whose source identification is missing, or whose classification conditions do not meet the preset requirements in the initial classification, the initial confidence level corresponds to the second score range. The initial value is stored on the blockchain as the starting data for the traceability chain, and the traceability credibility updates of each subsequent node are all based on the credibility value of the preceding node and are passed down level by level.

10. The method for recycling and tracing building insulation boards based on IoT communication and blockchain technology according to claim 1, characterized in that: The content stored on the blockchain in S3 includes: original IoT data, derivation results and deviation values ​​of each derivation path, parameter self-consistency verification conclusions, values ​​before and after the traceability credibility update and the basis for the update, and timestamps; the blockchain smart contract automatically triggers credibility update and storage operations after the node completes the IoT data upload. In S4, the downstream processing terminal automatically matches the corresponding processing flow according to the range of traceability credibility: materials in the high credibility range directly enter the corresponding material recycling production line, materials in the medium credibility range are allocated to the sampling production line and the testing frequency and items are increased, and materials in the low credibility range are forcibly sent to the laboratory for full testing including HBCD content, and are classified and processed according to the actual results. The matching rules for the disposal process are pre-fixed by a blockchain smart contract. The disposal end can only read the matching results and has no right to modify the matching rules. The actual disposal results are sent back to the blockchain for on-chain storage and evidence storage, forming a complete traceability closed loop.

Images