Solid waste space-time positioning and attribute association method based on body element knowledge unit

Abstract

The application discloses a solid waste space-time positioning and attribute association method based on a body element knowledge unit, and relates to the technical field of environmental monitoring and pollution tracing. Through acquisition of multi-modal environmental data such as Internet of Things sensors and unmanned aerial vehicle remote sensing, a three-dimensional abnormal body element with a space-time range is identified; a body element knowledge unit library integrating space-time labels, waste attribute labels, environmental attribute labels and source labels is constructed; the abnormal body element is associated with a VKU through space-time matching and causal association of waste attributes and environmental abnormal indicators; the VKU is matched with enterprise waste production / dumping records based on the principle that the three conditions of time, space and attribute are consistent; and finally, through multi-source cross verification such as laboratory detection, trajectory verification and witness inquiry, the accuracy of the tracing result is ensured. The space-time accuracy and causal association reliability of pollution tracing are improved, a scientific basis is provided for pollution liability identification, and the problems of low space-time accuracy, poor multi-source data integration and insufficient result reliability of existing methods are solved.

Description

Technical Field

[0001] This invention relates to the field of environmental management and resource reuse technology for solid waste landfills, specifically to a method for spatiotemporal location and attribute association of solid waste based on Voxel Knowledge Unit (VKU). Background Technology

[0002] Hazardous waste (hereinafter referred to as "hazardous waste") contains toxic and harmful substances such as heavy metals (e.g., copper, lead, cadmium) and organic pollutants (e.g., dioxins), making its landfill disposal a key focus of environmental management. Accurate recording of the spatial and temporal extent and property characteristics of underground solid waste is a crucial prerequisite for subsequent reuse (e.g., resource recycling) and leakage risk prevention (e.g., leachate pollution). Currently, the management of solid waste in hazardous waste landfills faces the following core issues:

[0003] First, underground solid waste is "invisible and difficult to locate": after being landfilled, solid waste enters underground space, making it impossible to accurately record its three-dimensional spatial coordinates (such as depth). Planar position ) and time range (e.g., pouring time) This leads to the need for blind excavation during future reuse, resulting in low efficiency and high cost (for example, when a landfill was retrieving copper waste, the amount of excavation was three times the actual requirement because the depth could not be determined).

[0004] Second, "separation of spatiotemporal data from attribute data": waste generation records (such as heavy metal content) ), transfer manifests (such as dumping time) ,coordinate Landfill monitoring (such as leachate concentration) Data such as these are scattered across different systems, making it impossible to link them into a unified "spatiotemporal-attribute" knowledge carrier (e.g., "dumped on October 1, 2025"). "HW17 waste, copper content 5%", when leakage occurs, cannot quickly match the location of contamination (such as anomalous elements). The corresponding solid waste properties (such as copper content and toxicity) lead to a lag in the response to leakage risks (for example, if the copper content in the leachate of a landfill exceeds the standard, it may take 3 days to find the corresponding solid waste source).

[0005] Third, the volume accuracy is low: existing methods cannot accurately calculate the volume of solid waste (e.g., 0.1 cubic meters), which affects the quantity control and cost estimation during reuse (e.g., a company needs to retrieve 10 cubic meters of copper waste, but due to the low volume accuracy, it actually excavates 12 cubic meters, increasing the cost by 20%).

[0006] These problems severely restrict the intelligent management of hazardous waste landfills, failing to meet the requirements of "full-process traceability" (such as recording every step from enterprise waste generation to landfill), "rapid response to leakage risks" (such as locating the source of pollution within hours), and "precise volume positioning." To address these issues, there is an urgent need for a spatiotemporal-attribute association method that integrates multimodal data and is based on three-dimensional voxels to achieve accurate recording and rapid association of underground solid waste, providing an effective technical means for environmental management of hazardous waste landfills and the location of recyclable resources in the later stages. Summary of the Invention

[0007] This invention discloses a method for spatiotemporal location and attribute association of solid waste based on Voxel Knowledge Unit (VKU), which aims to solve the core problem of "accurate association between the spatiotemporal range and attribute characteristics of underground solid waste" in hazardous waste landfills.

[0008] The technical solution of this invention: A method for spatiotemporal localization and attribute association of solid waste based on volumetric knowledge units, comprising the following steps:

[0009] Step (1) Obtain multimodal environmental data of the landfill area, including IoT sensor data and UAV remote sensing images; identify anomalous voxels through the multimodal environmental data, wherein the anomalous voxels are three-dimensional voxels with a spatiotemporal range, the spatiotemporal range including three-dimensional spatial coordinates. Abnormal time range ,in For planar coordinates, For depth coordinates, This is the abnormal start time. This is the abnormal end time;

[0010] Step (2) Constructing a volumetric knowledge unit, namely the VKU library, wherein the VKU is a three-dimensional volumetric data structure integrating multimodal tags, the multimodal tags including spatiotemporal tags, waste attribute tags, environmental attribute tags, and source tags, wherein the spatiotemporal tags include dumping time. Tilting coordinates The waste attribute tags include VKU's waste categories. Toxicity level T Collection of heavy metal contents , This represents the number of heavy metal species. For the first The mass fraction of heavy metals, the environmental attribute label including soil permeability Groundwater flow velocity The source label includes the company name. Transfer document number ;

[0011] Step (3) associate the abnormal entity with the VKU in the VKU library. The association rule is: the abnormal time range of the abnormal entity. The pouring time including the VKU ,Right now And the three-dimensional spatial coordinates of the abnormal volume element The tilt coordinates of the VKU The Euclidean distance Δs in the spatiotemporal overlap spatial error threshold Within, simultaneously, the environmental anomaly indicators of the abnormal entity. There is a causal mapping relationship between the waste attribute tags of the VKU and the VKU, and the causal mapping relationship is calculated by establishing a waste heavy metal leaching model to calculate the set of heavy metal contents of the VKU. Corresponding theoretical environmental anomaly indicators If the environmental anomaly indicators of the abnormal element and relative error , If the preset threshold is used, it is determined that the VKU is successfully associated with the abnormal entity.

[0012] Step (4) Match the associated VKU with the enterprise's waste generation / dumping records for solid waste source tracing. The matching rule is based on the principle of consistency of three conditions: time, space, and attribute. The dumping time of the VKU is also considered. With the time of enterprise waste generation Dumping time of transfer manifest Time difference Δt p , Δt l all Preset time threshold ,Right now , The tilt coordinates of the VKU GPS track coordinates of transport vehicles Landfill unloading record coordinates Spatial difference Δs g Δs u all Spatial matching spatial error threshold The waste category of the VKU Collection of heavy metal contents Waste categories in the company's waste generation records Collection of heavy metal contents Completely identical, that is and , The first in the enterprise's waste generation record Mass fraction of heavy metals;

[0013] Step (5) Perform multi-source cross-validation on the matched spatiotemporal and attribute association results of solid waste. The validation methods include: laboratory validation, collecting waste samples of the anomalous voxels, and detecting the heavy metal content of the waste samples by ICP-MS. , The first in the waste sample The mass fraction of each heavy metal is used to determine the error between the set of heavy metal contents in the waste sample and the set of heavy metal contents in the VKU. , The content error threshold is set; trajectory verification involves retrieving the GPS trajectory and video surveillance data of the transport vehicle to confirm its arrival time. With the pouring time of the VKU Time difference Δt a The trajectory verification time error threshold T1, i.e. and reach the coordinates The tilt coordinates of the VKU Spatial difference Δs a Trajectory verification spatial error threshold Witness verification involved questioning landfill staff to confirm the company name on the origin label of the VKU. The names of the waste-generating companies as remembered by the staff Consistent, and the pouring time of the VKU. The time of dumping stated by the staff Time difference , The threshold for time error in witness verification;

[0014] Multi-source cross-validation: If laboratory validation, trajectory validation, and witness validation all meet the preset conditions, the spatiotemporal positioning and attribute association results of solid waste are determined to be accurate; if any of the validations does not meet the preset conditions, repeat steps (3) and (4) to obtain VKU and abnormal body element association data and solid waste source matching data again, and then perform multi-source cross-validation of laboratory validation, trajectory validation, and witness validation.

[0015] Furthermore, the method for identifying anomalous voxels through the multimodal environmental data in step (1) is as follows: using the isolated forest algorithm or the local anomaly factor algorithm to detect anomalies in the environmental monitoring parameters in the IoT sensor data, and combining the waste pile location identified by UAV remote sensing image to determine the three-dimensional voxels with anomalous environmental monitoring parameters as anomalous voxels.

[0016] Furthermore, the tilting time of the spatiotemporal tag mentioned in step (2) The dumping coordinates of the VKU are the actual time of waste dumping recorded in the transfer manifest. The unloading location plane coordinates recorded in the GPS trajectory of the transport vehicle and the unloading depth recorded by the landfill depth sensor; the waste category of the waste attribute label. The toxicity level is specified as a hazardous waste category in the National Hazardous Waste List. T The heavy metal content of the VKU is determined according to the toxicity level specified in the "Standard for Identification of Hazardous Waste". The laboratory test results in the company's waste generation records; the soil permeability of the environmental attribute label. Groundwater flow velocity Historical monitoring data from the landfill environmental monitoring station; the company name on the source label. The name of the waste-generating enterprise in the transfer manifest, and the transfer manifest number. It serves as the unique identifier for the transfer manifest.

[0017] Furthermore, the specific implementation method of the association rule in step (3) is as follows: First, the three-dimensional spatial coordinates of the abnormal volume element are quickly filtered out by spatial index. Error in the spatiotemporal overlap space error threshold The VKU candidate set is then processed by time filtering to retain the pouring time. Located within the abnormal time range The VKUs within the time filter are then used for attribute matching.

[0018] Furthermore, the specific matching process for the three conditions in step (4) is as follows: First, the associated VKU is matched with the enterprise's waste production records by time to filter out the enterprise's waste production time. With the pouring time of the VKU Time difference ≤ The candidate set of enterprises is then used for spatial matching to filter out the GPS trajectory coordinates of transport vehicles. The tilt coordinates of the VKU Spatial difference ≤ The enterprises were then spatially matched; finally, attribute matching was performed on the enterprises after spatial matching to filter out the waste categories in their waste production records. Collection of heavy metal contents Enterprises whose waste attribute tags are completely consistent with those of the VKU are considered as the main sources of solid waste, and their spatiotemporal and attribute information in their waste generation / dumping records is associated with them.

[0019] Furthermore, the method for constructing the VKU library in step (2) is as follows: collect historical dumping event data from landfills, including transfer manifest data, enterprise waste generation records, transport vehicle GPS data, landfill unloading records, and environmental monitoring data; and extract spatiotemporal tags for each historical dumping event. Waste attribute tags Environmental attribute tags Source tags Generate the corresponding VKU; store all VKUs in a relational database or a non-relational database to form a VKU library, where the VKU indexing methods include spatial index, time index, and attribute index.

[0020] Furthermore, the three-dimensional spatial coordinates of the abnormal volume element mentioned in step (1) The method for determining this is as follows: the planar coordinates of the waste pile are identified through multispectral analysis of UAV remote sensing images. The depth of the waste pile is obtained through depth sensors in the landfill. Abnormal time range The determination method is as follows: through time series analysis of IoT sensor data, the time when the environmental monitoring parameter first exceeds the threshold is identified as... The time it takes for environmental monitoring parameters to recover to within the threshold is used as... .

[0021] Furthermore, the causal mapping relationship judgment method in step (3) is as follows: a machine learning model is used to establish a mapping relationship between the heavy metal content of waste and environmental anomaly indicators, and the input of the machine learning model is the set of heavy metal content of VKU. Soil permeability Groundwater flow velocity The output is a theoretical environmental anomaly index. ; optimize model parameters through cross-validation to reduce the model's prediction error , It is the total number of samples used for validation. It is the first Measured environmental anomaly indicators of individual abnormal elements. It is determined by the machine learning model based on the first The theoretical environmental anomaly index is predicted from the waste attribute labels of each VKU. This is the maximum RMSE threshold allowed by the model; if the environmental anomaly index of the abnormal entity... With the model output correlation coefficient , If a preset correlation threshold is set, a causal mapping relationship is determined to exist. It is all The arithmetic mean of the measured values, i.e. , It is all The arithmetic mean of the predicted values, i.e. , It is the Pearson correlation coefficient, which measures the degree of linear correlation between measured and predicted values, and its value ranges from [value range missing]. .

[0022] Preferably, the allowable error range for the laboratory verification in step (5) is: for each element in the heavy metal content set. , The value is 0.1%; the allowable range for the trajectory verification time error threshold is: T 1 ≤ 30 minutes, the allowable range of spatial matching spatial error threshold is: ≤0.5 meters; the allowable range for the spatial error threshold of trajectory verification is: ≤0.5 meters; the allowable range for witness verification time error is: T 2 ≤ 60 minutes.

[0023] The beneficial effects of this invention are: it addresses the core problems of "spatiotemporal invisibility and difficulty in attribute correlation" after solid waste is landfilled. This method integrates multimodal data from IoT sensors, UAV remote sensing, enterprise waste generation records, transfer manifests, GPS trajectories, and environmental monitoring to construct a three-dimensional volumetric knowledge carrier (VKU) that combines "spatiotemporal tags + attribute tags." This accurately records the spatiotemporal range (time, location, depth) and attribute characteristics (waste category, heavy metal content, toxicity level) of solid waste, enabling full-process traceability of solid waste from "enterprise waste generation - transportation - landfill," and precise matching of "pollution location - solid waste attributes." This forms a closed-loop process management of "abnormal volumetric identification - VKU association - solid waste source tracing and matching - result verification," providing technical support for precise excavation for future solid waste reuse (such as location and quantity control during resource recycling), pollutant source tracing during leakage (such as source attribute identification of leachate pollution), and intelligent operation of landfills. Detailed Implementation

[0024] I. Abnormal Voxel Identification

[0025] Acquire multimodal environmental data of the landfill area, including IoT sensor data (such as leachate heavy metal concentration and groundwater COD) and UAV remote sensing images, where the sensor data is time-series data. , It is the first The timestamp of the monitoring session; It is the first The first monitoring Environmental parameter values ​​(such as leachate heavy metal concentration, groundwater COD, etc.); It is the number of types of environmental parameters (i.e., the parameter dimensions monitored each time). This refers to the number of monitoring sessions (the length of the time series). UAV remote sensing images are used to identify the planar location of waste piles. The Isolation Forest algorithm is used to detect anomalies in high-dimensional environmental data, and the anomaly score s(x) for each sample is calculated (formula is given). ,in ,in It is a single sample for which the anomaly score is to be calculated (in the landfill scenario, it is a vector of sensor readings at a certain moment). It is a sample In the isolated forest, the path length, i.e. The number of random partitions required to isolate the virus. It is the expected path length after averaging multiple isolated trees. It is the total number of samples used to construct the current isolated forest (i.e., the length of the time series). It is the Euler-Macheronny constant. , yes The average path length normalization factor for each sample is used to... Mapped to a comparable interval of 0–1. When the calculated... ( When the preset anomaly threshold (usually set to 0.7-0.9) is used, the environmental monitoring time / location corresponding to the sample is determined to be abnormal, and an anomaly volume element is generated.

[0026] Three-dimensional spatial coordinates based on anomaly samples (planar coordinates identified through UAV remote sensing images) Depth coordinates are obtained through landfill depth sensors. ) and abnormal time range (identifying the time when environmental monitoring parameters first exceed a threshold as the anomaly start time through time series analysis of IoT sensor data). The time it takes for environmental monitoring parameters to recover to within the threshold is taken as the anomaly end time. ), determine the spatiotemporal range of the anomalous volume element as ,in Three-dimensional spatial coordinates ( For planar coordinates, (for depth coordinates) For the abnormal time range, the environmental anomaly indicators of the abnormal entity are: , For the first Monitoring values ​​of various environmental parameters, such as copper concentration in leachate. The volume of abnormal volumes is calculated using the volumetric precision (e.g., 0.1 cubic meters / volume element, volume). ,in m, m, with a volume accuracy of 0.1 cubic meters.

[0027] The method for identifying anomalous voxels in multimodal environmental data is as follows: an isolated forest algorithm or local anomaly factor algorithm is used to detect anomalies in the environmental monitoring parameters in the IoT sensor data, and the location of the waste pile is identified by UAV remote sensing imagery to determine the three-dimensional voxels with anomalous environmental monitoring parameters as anomalous voxels.

[0028] II. VKU Library Construction and Association

[0029] VKU is a three-dimensional volumetric metadata structure that integrates multimodal tags to associate "solid waste spatiotemporal information" with "attribute features," denoted as... .in For spatiotemporal tags, including the time of pouring. (Actual time of waste dumping recorded in the transfer manifest), dumping coordinates (The plane coordinates of the unloading location recorded in the GPS track of the transport vehicle and the unloading depth recorded by the landfill depth sensor); Waste attribute tags, including VKU's waste categories. (e.g., HW17 heavy metal waste, which conforms to the "National Hazardous Waste List") Toxicity level T (Determined according to the "Hazardous Waste Identification Standard", such as hazardous level), heavy metal content collection (Laboratory test results in the company's waste generation records) This represents the number of heavy metal species. For the first The mass fraction of heavy metals, such as 5% copper and 2% lead. Environmental attribute labels, including soil permeability (Historical monitoring data from landfill environmental monitoring stations, unit: m / d), groundwater flow velocity (Unit: m / d); Source label, including company name (Name of the waste-generating enterprise in the transfer manifest), Transfer manifest number (The unique identifier of the transfer document, such as LD20251001002).

[0030] The VKU database is constructed as follows: Historical dumping event data from landfills is collected, including transfer manifest data, enterprise waste generation records, transport vehicle GPS data, landfill unloading records, and environmental monitoring data; for each historical dumping event, spatiotemporal tags are extracted. Waste attribute tags Environmental attribute tags Source tags Generate the corresponding VKU; store all VKUs in a relational database or a non-relational database to form a VKU library, where the VKU indexing methods include spatial index, time index, and attribute index.

[0031] The association between anomaly elements and VKUs must satisfy the conditions of "spatiotemporal overlap + attribute matching":

[0032] 1. Spatiotemporal overlap: The anomalous temporal range of anomalous voxels. Pouring time including VKU (Right now ), and the three-dimensional spatial coordinates of the anomalous element tilt coordinates with VKU Euclidean distance ( The spatial error threshold for spatiotemporal overlap is typically set to 0.5-1m to ensure volumetric accuracy. The three-dimensional spatial coordinates of the anomalous volume element can be quickly filtered using a spatial index (such as an R-tree). Error in the spatiotemporal overlap space error threshold The VKU candidate set is then processed by time filtering to retain the pouring time. Located within the abnormal time range VKU within.

[0033] 2. Attribute Matching: The set of heavy metal contents in VKU is calculated by establishing a waste heavy metal leaching model (such as the PHREEQC model). Corresponding theoretical environmental anomaly indicators (e.g., the theoretical value of copper concentration in leachate), if the environmental abnormality indicators of abnormal elements and relative error ( If a preset threshold value is used (usually 5%), then the VKU is determined to be successfully associated with the abnormal entity, thus realizing the association between the state indicator (leachate) and the performance indicator (waste attribute).

[0034] The method for determining the causal mapping relationship through attribute matching is as follows: a machine learning model is used to establish a mapping relationship between the heavy metal content in waste and environmental anomaly indicators. The input of the machine learning model is the set of heavy metal content in waste. Soil permeability Groundwater flow velocity The output is a theoretical environmental anomaly index. ; optimize model parameters through cross-validation to reduce the model's prediction error , It is the total number of samples used for validation (number of folds in cross-validation × number of samples per fold). It is the first Measured environmental anomaly indicators of an abnormal element (such as the concentration of a certain heavy metal in leachate). It is determined by the machine learning model based on the first The theoretical environmental anomaly indicators predicted by the waste attribute tags (heavy metal content, soil permeability, groundwater flow velocity, etc.) of each VKU It is the maximum RMSE threshold allowed by the model, for example, the value it can take. (Unit and environmental abnormal indicators) The corresponding parameters are the same, such as mg / L). The environmental anomaly indicators of the abnormal volume element are... With the model output correlation coefficient , A preset correlation threshold, which can be set to 0.9, is used to determine if a causal mapping relationship exists. It is all The arithmetic mean of the measured values, i.e. , It is all The arithmetic mean of the predicted values, i.e. , It is the Pearson correlation coefficient, which measures the degree of linear correlation between measured and predicted values, and its value ranges from [value range missing]. .

[0035] III. Solid Waste Source Tracing and Matching

[0036] The associated VKU is matched with the enterprise's waste generation / dumping records for solid waste source tracing. The matching must follow the principle of consistency in three conditions: time, space, and attributes. The aim is to associate the "spatiotemporal information" of solid waste with the "enterprise's waste generation records."

[0037] 1. Time Matching: Filter by the time when a company generates waste. Pouring time with VKU Time difference Dumping time with transfer manifest Time difference ( The candidate set of enterprises is set to a preset time threshold, typically 1 hour.

[0038] 2. Spatial matching: Filter out the GPS trajectory coordinates of transport vehicles. Landfill unloading record coordinates tilt coordinates with VKU Spatial difference Δs g Δs u :

[0039] ;

[0040] ;

[0041] For spatial matching, the spatial error threshold is typically set to 0.5 meters for enterprises.

[0042] 3. Attribute Matching: Filter waste categories from the company's waste generation records. Collection of heavy metal content in enterprise waste generation records Completely consistent with VKU's waste attribute tag (i.e. and , The first in the enterprise's waste generation record For enterprises with heavy metal mass fractions, link the "spatiotemporal and attribute information" (such as the mass fraction of heavy metals) in their waste production / dumping records. , , As the main source of solid waste, it achieves "full traceability from enterprise waste generation to landfill".

[0043] IV. Result Verification

[0044] The results verification employs a multi-source cross-validation approach, combining laboratory verification, trajectory verification, and witness verification, to ensure the accuracy and speed of solid waste spatiotemporal location and attribute association results.

[0045] 1. Laboratory Validation: Waste samples of abnormal volumes were collected, and their heavy metal content was detected by ICP-MS. , The first in the waste sample The mass fraction of each heavy metal, if combined with the heavy metal content of VKU In The relative error, ,in The content error threshold is usually set to 0.1%, which is sufficient for the attribute consistency verification to pass.

[0046] 2. Track Verification: Retrieve the GPS track and video surveillance data of the transport vehicle to confirm its arrival time. Pouring time with VKU Time difference ( The trajectory verification time error threshold is typically set at 30 minutes, and the arrival coordinates are... tilt coordinates with VKU Spatial difference, ,in The spatial error threshold for trajectory verification is typically set to 0.5 meters.

[0047] 3. Witness verification: Question landfill staff to confirm the company name. The names of the waste-generating companies as remembered by the staff Consistent, and the pouring time of VKU Time of statement to staff Time difference ( The time error threshold for witness verification is usually set at 60 minutes.

[0048] Through the above process, this invention achieves a precise correlation between the "spatial and temporal range and attribute characteristics of solid waste" in hazardous waste landfills, providing technical support for precise excavation for future reuse (such as positioning and quantity control during resource recycling) and matching of pollutant attributes during leakage (such as source attribute identification of leachate pollution).

[0049] The core advantage of this method is:

[0050] 1. Precise Spatiotemporal Positioning: Identifying the three-dimensional spatial coordinates of anomalous volumes using "UAV remote sensing + depth sensor". And through the precision of voxel division (e.g. m、 m、 m) to achieve accurate volume calculation (volume) The volume of the waste is 1 cubic meter, which meets the volume accuracy requirement of "0.1 cubic meter level" in the task book. It solves the problem of "underground solid waste being invisible and difficult to locate" (for example, when copper waste was retrieved from a landfill, the volume error of the location by this method was less than 5%, and the excavation efficiency was increased by 2 times).

[0051] 2. Attribute Feature Correlation: Calculate theoretical environmental anomaly indicators using a "waste heavy metal leaching model" (such as PHREEQC). and environmental anomaly indicators Comparison (relative error) This method enables the correlation between "state indicators (leachate) and performance indicators (waste attributes)," solving the problem of "separation of spatiotemporal data and attribute data." For example, if copper exceeds the standard in the leachate of a landfill, this method can be used to find the corresponding solid waste attribute—HW17 waste with a copper content of 5%—within 1 hour, providing a targeted basis for leakage repair.

[0052] 3. Full process traceability: Through the "three conditions consistency principle" (time difference) Hours, spatial differences (With identical meter and attributes) the VKU is linked to the enterprise's waste generation records, enabling full traceability from "enterprise waste generation - transportation - landfill" (e.g., for HW17 waste in an enterprise's waste generation records, this method can trace its dumping time to...). Tilting coordinates of VKU This provides precise spatiotemporal information for future reuse.

[0053] 4. Rapid response to leakage risks: through "multi-source cross-validation" (laboratory validation) Error and trajectory verification Minute time difference, witness verification (Minute time difference) ensures that the corresponding solid waste properties can be found within hours when leakage occurs (e.g., after copper exceeds the standard in the leachate of a landfill, this method can complete the verification within 2 hours, which buys time for pollution spread control).

[0054] This invention integrates multimodal data such as "waste generation, transportation, landfill, and monitoring" through an integrated volumetric knowledge unit (VKU) of "spatiotemporal tags + attribute tags," thereby realizing intelligent management of hazardous waste landfills and meeting the core requirements of the task book of "full-process traceability" and "volume accuracy at the 0.1 cubic meter level."

Claims

1. A method for solid waste spatiotemporal positioning and attribute association based on body element knowledge unit, characterized in that, Includes the following steps: Step (1) Obtain multimodal environmental data of the landfill area, including IoT sensor data and UAV remote sensing images; Anomaly voxels are identified using the multimodal environmental data. These anomaly voxels are three-dimensional voxels with a spatiotemporal range, which includes three-dimensional spatial coordinates. Abnormal time range ,in For planar coordinates, For depth coordinates, This is the abnormal start time. This is the abnormal end time; Step (2) Constructing a volumetric knowledge unit, namely the VKU library, wherein the VKU is a three-dimensional volumetric data structure integrating multimodal tags, the multimodal tags including spatiotemporal tags, waste attribute tags, environmental attribute tags, and source tags, wherein the spatiotemporal tags include dumping time. Tilting coordinates The waste attribute tags include VKU's waste categories. Toxicity level Collection of heavy metal contents , This represents the number of heavy metal species. For the first The mass fraction of heavy metals, the environmental attribute label including soil permeability Groundwater flow velocity The source label includes the company name. Transfer document number ; Step (3) associate the abnormal entity with the VKU in the VKU library. The association rule is: the abnormal time range of the abnormal entity. The pouring time including the VKU ,Right now And the three-dimensional spatial coordinates of the abnormal volume element The tilt coordinates of the VKU Euclidean distance Δ s Spatial error threshold of spatiotemporal overlap Within, simultaneously, the environmental anomaly indicators of the abnormal entity. There is a causal mapping relationship between the waste attribute tags of the VKU and the VKU, and the causal mapping relationship is calculated by establishing a waste heavy metal leaching model to calculate the set of heavy metal contents of the VKU. Corresponding theoretical environmental anomaly indicators If the environmental anomaly indicators of the abnormal element and relative error , If the preset threshold is used, it is determined that the VKU is successfully associated with the abnormal entity. Step (4) Match the associated VKU with the enterprise's waste generation / dumping records for solid waste source tracing. The matching rule is based on the principle of consistency of three conditions: time, space, and attribute. The dumping time of the VKU is also considered. With the time of enterprise waste generation Dumping time of transfer manifest Time difference Δ t p Δ t l all Preset time threshold ,Right now , The tilt coordinates of the VKU GPS track coordinates of transport vehicles Landfill unloading record coordinates Spatial difference Δ s g Δ s u all Spatial matching spatial error threshold The waste category of the VKU Collection of heavy metal contents Waste categories in the company's waste generation records Collection of heavy metal contents Completely identical, that is and , The first in the enterprise's waste generation record Mass fraction of heavy metals; Step (5) Perform multi-source cross-validation on the matched spatiotemporal and attribute association results of solid waste. The validation methods include: laboratory validation, collecting waste samples of the anomalous voxels, and detecting the heavy metal content of the waste samples by ICP-MS. , The first in the waste sample The mass fraction of each heavy metal is used to determine the error between the set of heavy metal contents in the waste sample and the set of heavy metal contents in the VKU. , The content error threshold is set; trajectory verification involves retrieving the GPS trajectory and video surveillance data of the transport vehicle to confirm its arrival time. With the pouring time of the VKU Time difference Δ t a Trajectory verification time error threshold T 1, that is and reach the coordinates The tilt coordinates of the VKU Spatial difference Δ s a Trajectory verification spatial error threshold Witness verification involved questioning landfill staff to confirm the company name on the origin label of the VKU. The names of the waste-generating companies as remembered by the staff Consistent, and the pouring time of the VKU. The time of dumping stated by the staff Time difference , The threshold for time error in witness verification; Multi-source cross-validation: If laboratory validation, trajectory validation, and witness validation all meet the preset conditions, the spatiotemporal positioning and attribute association results of solid waste are determined to be accurate; if any of the validations does not meet the preset conditions, repeat steps (3) and (4) to obtain VKU and abnormal body element association data and solid waste source matching data again, and then perform multi-source cross-validation of laboratory validation, trajectory validation, and witness validation.

2. The method for spatiotemporal localization and attribute association of solid waste based on volumetric knowledge units according to claim 1, characterized in that, The method for identifying anomalous voxels in step (1) using the multimodal environmental data is as follows: anomaly detection is performed on the environmental monitoring parameters in the IoT sensor data using the isolated forest algorithm or the local anomaly factor algorithm, and the location of the waste pile identified by the UAV remote sensing image is combined to determine the three-dimensional voxels with anomalous environmental monitoring parameters as anomalous voxels.

3. The knowledge cell-based spatiotemporal positioning and attribute association method for solid waste according to claim 1, characterized in that, The tilting time of the spatiotemporal tag mentioned in step (2) The dumping coordinates of the VKU are the actual time of waste dumping recorded in the transfer manifest. The unloading location plane coordinates recorded in the GPS trajectory of the transport vehicle and the unloading depth recorded by the landfill depth sensor; the set of heavy metal contents of the VKU. The laboratory test results in the company's waste generation records; the soil permeability of the environmental attribute label. Groundwater flow velocity Historical monitoring data from the landfill environmental monitoring station; the company name on the source label. The name of the waste-generating enterprise in the transfer manifest, and the transfer manifest number. It serves as the unique identifier for the transfer manifest.

4. The knowledge cell-based spatiotemporal positioning and attribute association method for solid waste according to claim 1, characterized in that, The specific implementation method of the association rule in step (3) is as follows: First, the three-dimensional spatial coordinates of the abnormal volume element are quickly filtered out by spatial index. Error in the spatiotemporal overlap space error threshold The VKU candidate set is then processed by time filtering to retain the pouring time. Located within the abnormal time range The VKUs within the time filter are then used for attribute matching.

5. The knowledge cell-based spatiotemporal positioning and attribute association method for solid waste according to claim 1, wherein, The specific matching process for the three conditions in step (4) is as follows: First, the associated VKU is matched with the enterprise's waste production records by time to filter out the enterprise's waste production time. With the pouring time of the VKU Time difference ≤ The candidate set of enterprises is then used for spatial matching to filter out the GPS trajectory coordinates of transport vehicles. The tilt coordinates of the VKU Spatial difference ≤ The enterprises were then spatially matched; finally, attribute matching was performed on the enterprises after spatial matching to filter out the waste categories in their waste production records. Collection of heavy metal contents Enterprises whose waste attribute tags are completely consistent with those of the VKU are considered as the main sources of solid waste, and their spatiotemporal and attribute information in their waste generation / dumping records is associated with them.

6. The method for spatiotemporal localization and attribute association of solid waste based on volumetric knowledge units according to claim 1, characterized in that, The method for constructing the VKU library in step (2) is as follows: collect historical dumping event data of landfills, including transfer manifest data, enterprise waste generation records, transport vehicle GPS data, landfill unloading records, and environmental monitoring data; Extract spatiotemporal tags for each historical dumping event. Waste attribute tags Environmental attribute tags Source tags Generate the corresponding VKU; store all VKUs in a relational database or a non-relational database to form a VKU library, where the VKU indexing methods include spatial index, time index, and attribute index.

7. The knowledge cell-based spatiotemporal positioning and attribute association method for solid waste according to claim 1, wherein, The three-dimensional spatial coordinates of the abnormal volume element mentioned in step (1) The method for determining this is as follows: the planar coordinates of the waste pile are identified through multispectral analysis of UAV remote sensing images. The depth of the waste pile is obtained through depth sensors in the landfill. Abnormal time range The determination method is as follows: through time series analysis of IoT sensor data, the time when the environmental monitoring parameter first exceeds the threshold is identified as... The time it takes for environmental monitoring parameters to recover to within the threshold is used as... .

8. The method for spatiotemporal localization and attribute association of solid waste based on volumetric knowledge units according to claim 1, characterized in that, The causal mapping relationship determination method in step (3) is as follows: a machine learning model is used to establish a mapping relationship between the heavy metal content of waste and environmental anomaly indicators. The input of the machine learning model is the set of heavy metal content of VKU. Soil permeability Groundwater flow velocity The output is a theoretical environmental anomaly index. ; optimize model parameters through cross-validation to reduce the model's prediction error , It is the total number of samples used for validation. It is the first Measured environmental anomaly indicators of individual abnormal elements. It is determined by the machine learning model based on the first The theoretical environmental anomaly index is predicted from the waste attribute labels of each VKU. It is the maximum RMSE threshold allowed by the model; If the environmental anomaly indicators of the abnormal element With the model output correlation coefficient , If a preset correlation threshold is set, a causal mapping relationship is determined to exist. It is all The arithmetic mean of the measured values, i.e. , It is all The arithmetic mean of the predicted values, i.e. , It is the Pearson correlation coefficient, which measures the degree of linear correlation between measured and predicted values, and its value ranges from [value range missing]. .

9. The method for spatiotemporal localization and attribute association of solid waste based on volumetric knowledge units according to claim 1, characterized in that, The allowable error range for laboratory verification in step (5) is: for each element in the heavy metal content set of VKU. , The value is 0.1%; the allowable range for the trajectory verification time error threshold is: T 1 ≤ 30 minutes, the allowable range of spatial matching spatial error threshold is: ≤0.5 meters; the allowable range for the spatial error threshold of trajectory verification is: ≤0.5 meters; the allowable range for witness verification time error is: T 2 ≤ 60 minutes.

Images