A hazardous waste storage violation behavior identification method, system, device and medium

Abstract

This application relates to a method, system, equipment, and medium for identifying violations in hazardous waste storage. The method includes: acquiring images of hazardous waste packaging; calculating the attribute probability distribution of these images to obtain a digital twin of hazardous waste attributes containing attribute probability distribution vectors and uncertainty labels; constructing a spatial topology for the digital twin of hazardous waste attributes to generate a dynamic spatiotemporal topology network; calculating a basic risk coupling value by performing risk coupling integral calculation on the attribute probability distributions between nodes based on a pre-set hazardous waste taboo strength matrix and the dynamic spatiotemporal topology network; correcting the basic risk coupling value for environmental field factors based on real-time environmental data to generate a dynamic risk value; performing a time-series cumulative integral on the dynamic risk value and further determining a violation threshold to obtain the violation identification result. This method can solve the problem of inaccurate identification due to attribute uncertainty and lack of dynamic correlation, achieving accurate identification of potential chemical reaction risks.

Description

Technical Field

[0001] This invention belongs to the field of intelligent identification technology, and in particular relates to a method, system, equipment and medium for identifying violations in hazardous waste storage. Background Technology

[0002] With the continuous advancement of industrialization, the amount of hazardous waste generated is increasing daily. Its diverse types and complex properties make hazardous waste storage management a crucial link in environmental protection and safe production. To ensure storage safety and prevent accidents such as fires, explosions, or environmental pollution caused by improper storage, effective monitoring and identification of violations during the storage process are necessary. With the development of computer vision and Internet of Things (IoT) technologies, intelligent monitoring methods are gradually being applied to the field of hazardous waste storage. By using image acquisition and environmental sensing data to replace traditional manual inspections, real-time monitoring of the hazardous waste storage status can be achieved.

[0003] Traditional technologies typically employ rule-based matching or single-feature-based methods to identify violations in hazardous waste storage. Specifically, existing technologies often use optical character recognition (OCR) or target detection on captured images to obtain label information or appearance features of the hazardous waste, comparing this information with pre-stored storage planning data to determine if there are violations such as incorrect or mixed storage. Alternatively, sensors can be deployed to monitor temperature, humidity, and gas concentration within the warehouse to determine if environmental data exceeds preset safety thresholds, thereby triggering alarms for overt violations or environmental anomalies.

[0004] However, current identification methods primarily rely on deterministic feature matching and single-point threshold determination, which leads to inaccurate identification of potential risks. Because hazardous waste packaging may be damaged or obscured, existing technologies struggle to handle the ambiguity and uncertainty in the identification process, easily resulting in missed detections or misjudgments. Furthermore, current technologies typically analyze individual packaging attributes or independent environmental indicators in isolation, lacking dynamic correlation analysis of spatial proximity relationships between hazardous waste packaging and real-time environmental factors. This makes it difficult to effectively uncover and identify potential chemical reaction risks arising from uncertain hazardous waste properties and spatial proximity coupling, hindering accurate early warnings in the event of risks. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, system, equipment, and medium for identifying hazardous waste storage violations that can accurately identify potential chemical reaction risks caused by uncertain hazardous waste properties and spatial proximity coupling, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a method for identifying violations in hazardous waste storage, including:

[0007] S1. Obtain images of hazardous waste packaging, calculate the attribute probability distribution of the hazardous waste packaging images, and obtain a digital twin of hazardous waste attributes. The digital twin of hazardous waste attributes includes an attribute probability distribution vector and an uncertainty label.

[0008] S2. Construct a dynamic spatiotemporal topology network by using hazardous waste packaging as nodes and spatial proximity relationships as edges for the digital twin of hazardous waste attributes.

[0009] S3. Based on the pre-set hazardous waste taboo strength matrix and dynamic spatiotemporal topology network, risk coupling integral calculation is performed on the attribute probability distribution between nodes to obtain the basic risk coupling value, which is used to characterize the potential chemical reaction risk.

[0010] S4. Based on real-time environmental data, the basic risk coupling value is corrected by environmental field factors to generate a dynamic risk value. The dynamic risk value is used to reflect the real risk under the current environmental conditions.

[0011] S5. Perform time-series cumulative integration on the dynamic risk value to obtain the cumulative risk value;

[0012] S6. Determine the violation threshold based on the cumulative risk value to obtain the violation identification result.

[0013] In one embodiment, S1 includes:

[0014] S11. Perform target detection on the image of hazardous waste packaging to obtain the target region of the image of hazardous waste packaging;

[0015] S12. Perform feature mapping on the target region of the image to obtain the feature vector;

[0016] S13. Perform nonlinear activation on the feature vector to obtain class evidence values ​​corresponding to various hazardous waste attributes;

[0017] S14. Based on the Dirichlet distribution parameters, construct rules and perform distribution transformation on the class evidence values ​​to obtain a digital twin of hazardous waste attributes.

[0018] In one embodiment, S3 includes:

[0019] S31. Traverse the edges in the dynamic spatiotemporal topology network to obtain the attribute probability distribution vectors of the two nodes connected by the edges.

[0020] S32. Based on the hazardous waste taboo strength matrix and attribute probability distribution vector, perform probability weighted summation on attribute combinations to obtain the basic risk expectation value between nodes;

[0021] S33. The basic risk expectation value is fused with the spatial proximity weight of the edge to obtain the basic risk coupling value.

[0022] In one embodiment, S32 includes:

[0023] S321. Based on the attribute probability distribution vector, calculate the combination probability of the node category probability to obtain the attribute combination probability.

[0024] S322. Based on the hazardous waste taboo strength matrix and attribute combination probability, perform a matching query on the hazard level to obtain the target hazard level;

[0025] S323. Based on the attribute combination probability and the target hazard level, the risk values ​​of each attribute combination are accumulated to obtain the basic expected risk value between nodes. The formula for calculating the basic expected risk value is as follows:

[0026]

[0027] in, Basic risk expectation, and Each of the two nodes belongs to the first... Class and First The probability value of the class. For the first Class II hazardous waste and Class III Hazard classification of hazardous waste This is the preset total number of hazardous waste categories.

[0028] In one embodiment, S4 includes:

[0029] S41. Based on the real-time environmental data, the real-time environmental data includes the temperature, humidity and characteristic gas concentration data of the current area;

[0030] S42. Calculate the environmental correction factor based on temperature, humidity, and characteristic gas concentration data;

[0031] S43. Based on the basic risk coupling value and the environmental correction factor, perform product calculation to generate a dynamic risk value.

[0032] In one embodiment, S42 includes:

[0033] S421. Based on the preset reference temperature and reference humidity, calculate the difference between the temperature and humidity of the current area to obtain the temperature deviation value and humidity deviation value.

[0034] S422. Perform individual correction calculations on the temperature deviation value, humidity deviation value, and characteristic gas concentration data respectively to obtain the temperature correction term, humidity correction term, and gas concentration correction term;

[0035] S423. Multiply the temperature correction term, humidity correction term, and gas concentration correction term to obtain the environmental correction factor. The formula for calculating the environmental correction factor is:

[0036]

[0037] in, As an environmental correction factor, and These are the current temperature and humidity, respectively. and These are the reference temperature and reference humidity, respectively. Characteristic gas concentration, , and These are the temperature influence coefficient, humidity influence coefficient, and gas concentration sensitivity coefficient, respectively.

[0038] In one embodiment, S6 includes:

[0039] S61. Compare the cumulative risk value with the violation threshold to obtain the comparison result;

[0040] S62. When the comparison results meet the violation conditions, the risk contribution of each hazardous waste packaging node is calculated to obtain the risk source location information.

[0041] S63. Compare and encapsulate the results and risk source location information to obtain the violation identification results.

[0042] Secondly, this application also provides a hazardous waste storage violation identification system, including:

[0043] The attribute probability distribution calculation module is used to acquire images of hazardous waste packaging, calculate the attribute probability distribution of the images of hazardous waste packaging, and obtain a digital twin of hazardous waste attributes. The digital twin of hazardous waste attributes includes an attribute probability distribution vector and an uncertainty label.

[0044] The spatial topology construction module is used to construct a dynamic spatiotemporal topology network for the digital twin of hazardous waste attributes, with hazardous waste packaging as nodes and spatial proximity relationships as edges.

[0045] The risk coupling integral calculation module is used to perform risk coupling integral calculation on the attribute probability distribution between nodes based on the preset hazardous waste taboo strength matrix and dynamic spatiotemporal topology network, and obtain the basic risk coupling value. The basic risk coupling value is used to characterize the potential chemical reaction risk.

[0046] The environmental field factor correction module is used to correct the basic risk coupling value based on real-time environmental data and generate a dynamic risk value. The dynamic risk value is used to reflect the real risk under the current environmental conditions.

[0047] The time-series cumulative integration module is used to perform time-series cumulative integration on the dynamic risk value to obtain the cumulative risk value;

[0048] The violation threshold determination module is used to determine the violation threshold based on the cumulative risk value and obtain the violation behavior identification result.

[0049] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the hazardous waste storage violation identification method of the first aspect.

[0050] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for identifying violations in hazardous waste storage as described in the first aspect.

[0051] The aforementioned method, system, equipment, and medium for identifying violations in hazardous waste storage effectively overcome the limitations of traditional methods that rely on isolated analysis of a single target. This achieves accurate, dynamic, and forward-looking identification of violations in hazardous waste storage under complex and uncertain environments. By calculating the attribute probability distribution of hazardous waste packaging images, a digital twin of hazardous waste attributes containing uncertainty labels is constructed to address ambiguity in the identification process. Subsequently, a dynamic spatiotemporal topological network is built to characterize the spatial proximity relationships between packaging components. Combined with a pre-set hazardous waste taboo strength matrix, risk coupling integral calculations are performed on the attribute probability distribution between nodes, enabling the quantitative mining of potential chemical reaction risks. Furthermore, real-time environmental data is used to dynamically correct and accumulate the basic risk coupling values ​​over time. This effectively overcomes the limitations of traditional methods that rely on isolated analysis of a single target, achieving the technical effect of accurate, dynamic, and forward-looking identification of violations in hazardous waste storage under complex and uncertain environments. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 A flowchart illustrating a method for identifying violations in hazardous waste storage provided by the present invention;

[0054] Figure 2 A flowchart illustrating a method for generating violation identification results in an optional embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram of the structure of a hazardous waste storage violation identification system provided by the present invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0057] In one embodiment, such as Figure 1 As shown, a method for identifying violations in hazardous waste storage is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, or to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0058] S1. Obtain images of hazardous waste packaging, calculate the attribute probability distribution of the hazardous waste packaging images, and obtain a digital twin of hazardous waste attributes. The digital twin of hazardous waste attributes includes attribute probability distribution vectors and uncertainty labels.

[0059] Optionally, images of hazardous waste packaging acquired through image acquisition equipment are preprocessed before attribute probability distribution calculation. Preprocessing includes image denoising, grayscale conversion, and contour extraction. Image denoising employs a Gaussian filtering algorithm, which suppresses random noise by weighted averaging of neighboring pixels for each pixel. Grayscale conversion converts the color image to a single-channel grayscale image, reducing computation while preserving key appearance features of the hazardous waste packaging. Contour extraction uses the Canny edge detection algorithm to accurately extract the contour information of the hazardous waste packaging, avoiding background interference. Attribute probability distribution calculation is based on the appearance, label, and texture features of the hazardous waste packaging. By extracting these features and comparing them with a pre-set hazardous waste attribute feature library, the probability of the hazardous waste packaging belonging to various hazardous waste attributes is calculated, ultimately forming an attribute probability distribution vector. The attribute probability distribution vector is a one-dimensional vector composed of probability values ​​corresponding to multiple hazardous waste attribute categories. Each element corresponds to the probability of belonging to a specific hazardous waste attribute, and the sum of all elements is 1. Uncertainty labels are calculated based on the dispersion of the attribute probability distribution vector. Variance is used as the quantitative index of uncertainty. The larger the variance, the higher the uncertainty of the attribution of hazardous waste attributes. Uncertainty labels are used to characterize the reliability of attribute probability distribution. The digital twin of hazardous waste attributes is composed of the above attribute probability distribution vector and uncertainty labels, which can completely map the attribute characteristics of hazardous waste packaging and the uncertainty of attribute identification.

[0060] S2. Construct a dynamic spatiotemporal topology network by using hazardous waste packaging as nodes and spatial proximity relationships as edges for the digital twin of hazardous waste attributes.

[0061] Optionally, a digital twin of hazardous waste attributes corresponding to each hazardous waste package is used as a node. The attribute information of the node follows the attribute probability distribution vector and uncertainty label of the digital twin of hazardous waste attributes. The spatial proximity relationship is determined based on the coordinate information of each hazardous waste package in the image of the hazardous waste package. By calculating the straight-line distance between the contour centers of any two hazardous waste packages, two hazardous waste packages with a straight-line distance less than a preset proximity threshold are determined to have a spatial proximity relationship, and the corresponding two nodes are connected by an edge. In the process of generating the dynamic spatiotemporal topology network, the weight of the edge is determined by the spatial distance between the hazardous waste packages corresponding to the two nodes. The smaller the spatial distance, the larger the weight of the edge, which is used to characterize the degree of spatial association between the two hazardous waste packages. Preferably, the spatial distance is calculated using the Euclidean distance algorithm by calculating the square root of the sum of the squares of the coordinate differences between the two nodes. The dynamic spatiotemporal topology network is updated in real time. When the position of the hazardous waste package changes, the spatial distance and proximity relationship between each node are recalculated, and the edges and weights of the network are updated to ensure that the network can accurately reflect the real-time spatial distribution relationship of the hazardous waste packages.

[0062] S3. Based on the pre-set hazardous waste taboo strength matrix and dynamic spatiotemporal topology network, risk coupling integral calculation is performed on the attribute probability distribution between nodes to obtain the basic risk coupling value, which is used to characterize the potential chemical reaction risk.

[0063] Optionally, the pre-set hazardous waste incompatibility strength matrix is ​​a two-dimensional matrix pre-constructed based on the chemical properties and reaction characteristics of hazardous waste. The rows and columns of the matrix correspond to the hazardous waste attribute categories, and the element values ​​in the matrix represent the hazard level of chemical reactions between different types of hazardous waste. The larger the element value, the higher the risk of a hazardous chemical reaction when the two types of hazardous waste are stored adjacent to each other. The hazardous waste incompatibility strength matrix is ​​constructed in accordance with the hazardous waste safe storage specifications and chemical reaction manuals. It is obtained by sorting out the reaction characteristics of various types of hazardous waste and quantifying the hazard level of different combinations of hazardous waste. Based on the hazardous waste taboo strength matrix and the dynamic spatiotemporal topology network, risk coupling calculation is performed on the attribute probability distribution between nodes. Specifically, the edges in the dynamic spatiotemporal topology network are traversed first to obtain the attribute probability distribution vector of the two nodes connected by each edge. Then, based on the attribute probability distribution vector, the combination probability of the node category is calculated to obtain the attribute combination probability. Subsequently, based on the hazardous waste taboo strength matrix and the attribute combination probability, the hazard level is matched and queried to obtain the target hazard level. Finally, based on the attribute combination probability and the target hazard level, the risk values ​​of each attribute combination are accumulated to obtain the basic risk expectation value between nodes. The basic risk expectation value is multiplied and fused with the spatial proximity weight of the corresponding edge to obtain the risk coupling value corresponding to a single edge. The risk coupling values ​​corresponding to all edges in the dynamic spatiotemporal topology network are summed to obtain the overall basic risk coupling value. This basic risk coupling value is used to characterize the potential chemical reaction risk caused by attribute uncertainty and spatial proximity coupling of different hazardous waste packaging.

[0064] S4. Based on real-time environmental data, the basic risk coupling value is corrected by environmental field factors to generate a dynamic risk value, which is used to reflect the real risk under the current environmental conditions.

[0065] Optionally, based on real-time environmental data, the basic risk coupling value is corrected using environmental field factors to generate a dynamic risk value. Real-time environmental data includes temperature, humidity, and characteristic gas concentration data within the storage area. This data is acquired by synchronously reading real-time data transmitted from environmental sensing devices, ensuring consistency in the collection timestamps of the three types of data. After reading, the data is preprocessed to remove outlier data points. The determination of outlier data points is based on the reasonableness range of the data, which is pre-set according to the conventional environmental conditions of hazardous waste storage. Valid data is retained after removal for calculating the environmental correction factor. Based on preset reference temperature and humidity, the temperature and humidity differences in the current area are calculated to obtain temperature deviation and humidity deviation values. Individual correction calculations are then performed on the temperature deviation, humidity deviation, and characteristic gas concentration data to obtain temperature correction, humidity correction, and gas concentration correction terms. These three correction terms are multiplied to obtain the environmental correction factor. The basic risk coupling value is multiplied by the environmental correction factor to generate the dynamic risk value. The dynamic risk value calculation formula is as follows:

[0066]

[0067] in, This is a dynamic risk value. To accumulate risk value, Environmental correction factor, a dynamic risk value used to reflect the true risk under current environmental conditions.

[0068] S5. Perform time-series cumulative integration on the dynamic risk value to obtain the cumulative risk value.

[0069] Optionally, the time-series cumulative integral is used to calculate the cumulative effect of dynamic risk values ​​over time. Because the risk of hazardous waste chemical reactions is time-sensitive—short-term risk values ​​are low, but long-term accumulation may reach a dangerous threshold—it is necessary to quantify the degree of risk accumulation through the time-series cumulative integral. The calculation of the time-series cumulative integral uses time as the integration variable, integrating the dynamic risk values ​​over a continuous time period. The integration interval is a preset risk monitoring cycle. The integration process uses a trapezoidal integral algorithm, which divides the integration interval into several small trapezoids, calculates the area of ​​each small trapezoid, and sums them to obtain the cumulative risk value. The specific integration formula is as follows:

[0070]

[0071] in, To accumulate risk value, The start time of the monitoring period. This is the end time of the monitoring period. for The dynamic risk value at any given time is the product of the basic risk coupling value and the environmental correction factor. For example, the monitoring period can be preset according to the characteristics of hazardous waste storage. During the integration process, the dynamic risk value at each time point is read in real time, calculated and accumulated segment by segment, and finally the cumulative risk value over the entire monitoring period is obtained. This cumulative risk value can reflect the time-accumulated effect of hazardous waste storage risk.

[0072] S6. Determine the violation threshold based on the cumulative risk value to obtain the violation identification result.

[0073] Optionally, the violation threshold is a pre-set risk threshold based on hazardous waste storage safety standards. It is used to determine whether the accumulated risk value reaches the level of violation. The violation threshold is set according to the hazard level of different hazardous wastes and storage safety regulations, and is preset separately for each hazardous waste category to avoid misjudgment or omission due to a single threshold. The accumulated risk value is compared with the violation threshold for the corresponding hazardous waste category. If the accumulated risk value is greater than the violation threshold, it is determined that the violation condition is met; if the accumulated risk value is less than or equal to the violation threshold, it is determined that the violation condition is not met. When the comparison result meets the violation condition, the risk contribution of each hazardous waste packaging node is calculated. The risk contribution refers to the proportion of a single hazardous waste packaging node's contribution to the overall accumulated risk value. By selecting nodes with higher risk contribution values ​​as core risk sources, the unique identifier of the node, the risk contribution value, and the corresponding hazardous waste attribute information are integrated to obtain the risk source location information. The comparison results and risk source location information are encapsulated, and key information such as comparison conclusions, hazardous waste categories, violation thresholds (including if there is a violation), risk source node identifiers, and risk contribution are organized in a preset format to obtain the violation identification results.

[0074] In the aforementioned method for identifying violations in hazardous waste storage, images of hazardous waste packaging are acquired and a digital twin of hazardous waste attributes is constructed to accurately characterize the attributes of hazardous waste and identify uncertainties. Then, through spatial topology construction, risk coupling integration, environmental factor correction, temporal cumulative integration, and threshold determination, the potential risks and time-accumulated effects of hazardous waste storage are gradually quantified. At the same time, spatial correlation and environmental impact are taken into account, ultimately achieving accurate identification of violations in hazardous waste storage, effectively uncovering potential risks and achieving early warning, thereby improving the safety and intelligence level of hazardous waste storage management.

[0075] In one embodiment, S1 includes:

[0076] S11. Perform target detection on the image of hazardous waste packaging to obtain the target region of the image of hazardous waste packaging.

[0077] Optionally, the YOLO algorithm (You Only Look Once, a single-pass detection algorithm) is used for target detection. The YOLO algorithm completes the localization and classification of targets in the image through a single forward propagation, balancing detection speed and accuracy. Specifically, the input hazardous waste packaging image is first preprocessed. The preprocessing process includes image normalization, denoising, and resizing. Image normalization maps the image pixel values ​​to the same interval to avoid the impact of pixel value differences on the detection results. Denoising uses a Gaussian filtering algorithm, which suppresses random noise by weighted averaging of the image pixel neighborhood. Resizing scales the image to a preset fixed size to ensure that the algorithm input specifications are consistent. After preprocessing, the image is input into the YOLO algorithm. The YOLO algorithm performs multi-scale feature extraction on the image using preset anchor boxes. The anchor boxes are pre-defined rectangles of different sizes used to match hazardous waste packaging of different sizes. By calculating the intersection-union ratio (IUU) between the anchor boxes and the hazardous waste packaging, candidate boxes that meet the conditions are selected. Then, confidence filtering and non-maximum suppression are performed on the candidate boxes to remove overlapping candidate boxes. Finally, the image target region corresponding to each hazardous waste packaging is output. The image target region is represented by bounding box coordinates, accurately selecting the area where the hazardous waste packaging is located and eliminating interference from the background and other irrelevant areas.

[0078] S12. Perform feature mapping on the target region of the image to obtain the feature vector.

[0079] Optionally, based on the target region of the image, feature mapping is performed using convolution operations. The convolution operation involves sliding a preset convolution kernel across the target region of the image to extract texture, contour, and color features within the region. These features uniquely characterize the appearance attributes of hazardous waste packaging, providing support for hazardous waste attribute identification. The convolution kernel is a preset fixed-size matrix, and the elements in the matrix are preset weight values. These weight values ​​are obtained by fitting the correlation between the features and attributes of the hazardous waste packaging. Specifically, the target region of the image is converted into a single-channel grayscale image to reduce computation while retaining key features. Then, multiple rounds of convolution are performed on the grayscale image using convolution kernels of different sizes. After each round of convolution, a corresponding feature map is obtained. Each pixel in the feature map corresponds to a local feature of the target region of the image. All feature maps are concatenated in channel order, and then a flattening operation is used to convert the two-dimensional feature map into a one-dimensional vector. This one-dimensional vector is the feature vector, and each element of the feature vector corresponds to a local feature of the hazardous waste packaging. The vector dimension is consistent with the total number of extracted features.

[0080] S13. Perform nonlinear activation on the feature vector to obtain class evidence values ​​corresponding to various hazardous waste attributes.

[0081] Optionally, the ReLU function (Rectified Linear Unit) is used to nonlinearly activate the feature vector. This mainly introduces a nonlinear mapping to enhance the expressive power of the feature vector while suppressing interference from irrelevant features, highlighting key features related to the hazardous waste's attributes. The formula for calculating the ReLU function is:

[0082]

[0083] in, This function, which takes a single element value from a feature vector, performs non-linear filtering by setting elements less than 0 to 0 and retaining those greater than or equal to 0. Specifically, the feature vector is input element-by-element into the ReLU function, which calculates the activated feature values. These activated feature values ​​are then weighted and summed with pre-defined hazardous waste attribute feature weights to obtain class evidence values ​​for each type of hazardous waste attribute. The class evidence value characterizes the degree of matching between the feature vector and each type of hazardous waste attribute; a larger class evidence value indicates a higher probability that the hazardous waste packaging corresponding to the feature vector belongs to that type of hazardous waste attribute.

[0084] S14. Based on the Dirichlet distribution parameters, construct rules and perform distribution transformation on the class evidence values ​​to obtain a digital twin of hazardous waste attributes.

[0085] Optionally, the Dirichlet distribution is a multivariate probability distribution used to describe the probability distribution of multiple discrete variables. Its parameters are composed of concentration parameters; the larger the concentration parameter, the more concentrated the probability distribution and the lower the uncertainty. The Dirichlet distribution parameter construction rule is a pre-defined mapping rule. This mapping rule sets the concentration parameters of the Dirichlet distribution based on the magnitude of the class evidence values. Specifically, each class evidence value is used as the initial concentration parameter for the corresponding dimension of the Dirichlet distribution. Then, through normalization, the sum of all concentration parameters is made to a preset fixed value, ensuring the rationality of the distribution. Based on this mapping rule, the class evidence values ​​are transformed into a distribution, and the probability values ​​of various hazardous waste attributes are calculated using the probability density function of the Dirichlet distribution. The probability density function is:

[0086]

[0087] in, This is a vector composed of the probability values ​​of various hazardous waste attributes. Let be the concentration parameter vector of the Dirichlet distribution. This represents the total number of categories of hazardous waste attributes. The gamma function is used to normalize the concentration parameter. The probability vector obtained after the distribution transformation is the attribute probability distribution vector. The uncertainty label is then calculated by combining the dispersion of this vector. The attribute probability distribution vector and the uncertainty label together constitute a digital twin of hazardous waste attributes.

[0088] In the above embodiments, the target region of the image of hazardous waste packaging is accurately extracted by target detection, irrelevant interference is eliminated, key features are extracted by feature mapping and feature vectors are generated, effective features are highlighted by nonlinear activation and class evidence values ​​are obtained, and finally, a distribution transformation is completed based on Dirichlet distribution to obtain a digital twin of hazardous waste attributes, which accurately represents the attributes of hazardous waste and identifies uncertainties.

[0089] In one embodiment, S3 includes:

[0090] S31. Traverse the edges in the dynamic spatiotemporal topology network to obtain the attribute probability distribution vectors of the two nodes connected by the edges.

[0091] Optionally, the edges in the dynamic spatiotemporal topology network are traversed sequentially, starting from the edge corresponding to the network's starting node and visiting each edge in turn until all edges have been traversed, ensuring no omissions or duplicates. During the traversal, for each edge visited, its association information is extracted. This association information includes the unique identifiers of the two nodes connected by the edge. Using these unique identifiers, the basic node data stored in the dynamic spatiotemporal topology network can be directly accessed, thereby obtaining the attribute probability distribution vectors corresponding to the two nodes. The attribute probability distribution vector is a set of probabilities representing the hazardous waste packaging belonging to various hazardous waste attributes, with each element corresponding to the probability of belonging to a particular hazardous waste attribute, and the sum of all elements being 1. After the traversal is complete, a dedicated association data group is established for each edge. Each data group contains the edge's identifier and the attribute probability distribution vectors of the two nodes connected by that edge.

[0092] S32. Based on the hazardous waste taboo strength matrix and attribute probability distribution vector, perform probability weighted summation on attribute combinations to obtain the basic risk expectation value between nodes.

[0093] Optionally, the hazardous waste incompatibility strength matrix is ​​a two-dimensional matrix pre-constructed based on the chemical properties, reaction characteristics, and safe storage specifications of various hazardous wastes. The rows and columns of the matrix correspond to hazardous waste attribute categories. Each element value in the matrix represents the degree of incompatibility for a hazardous chemical reaction to occur when two hazardous waste attributes are stored adjacently. The larger the element value, the higher the probability that the coupling of the two hazardous waste attributes will cause a risk. Based on this hazardous waste incompatibility strength matrix and the attribute probability distribution vectors of the two nodes, all possible combinations of hazardous waste attributes in the two attribute probability distribution vectors are first enumerated. Each attribute combination consists of one hazardous waste attribute from the first node and one hazardous waste attribute from the second node. Then, a probability weighting calculation is performed on each attribute combination. Specifically, the probability value of the attribute corresponding to the first node is multiplied by the probability value of the attribute corresponding to the second node to obtain the joint probability of the attribute combination. This joint probability is then multiplied by the element values ​​of the two corresponding attributes in the hazardous waste incompatibility strength matrix to obtain the weighted risk value of the attribute combination. Finally, the weighted risk values ​​of all attribute combinations are summed to obtain the basic expected risk value between nodes. This basic expected risk value is used to characterize the potential risk level of hazardous waste packaging corresponding to two nodes based solely on attribute coupling.

[0094] S33. The basic risk expectation value is fused with the spatial proximity weight of the edge to obtain the basic risk coupling value.

[0095] Optionally, the spatial proximity weight of an edge is an inherent parameter of each edge in the dynamic spatiotemporal topology network. It is calculated based on the spatial distance between the two hazardous waste packages corresponding to the nodes connected by the edge. The smaller the spatial distance, the larger the spatial proximity weight of the edge, which is used to characterize the degree of spatial association between the two hazardous waste packages. This weight is generated and stored synchronously during the construction of the dynamic spatiotemporal topology network. The fusion process involves multiplying the expected value of the basic risk by the spatial proximity weight of the corresponding edge to obtain the basic risk coupling value. Preferably, the fusion process can first normalize the expected value of the basic risk to ensure that its value range is consistent with the spatial proximity weight, avoiding fusion deviations caused by differences in numerical ranges. The normalization process uses linear normalization to map the expected value of the basic risk to the same interval as the spatial proximity weight. Through this fusion method, the basic risk coupling value simultaneously considers the attribute coupling risk and spatial association degree of the hazardous waste packages, accurately characterizing the potential chemical reaction risk caused by the attribute matching degree and spatial proximity of the hazardous waste packages corresponding to the two nodes.

[0096] In the above embodiments, the probability distribution vector of node attributes is obtained by traversing the edges of the dynamic spatiotemporal topology network. Based on the hazardous waste taboo strength matrix, the attribute combination is summed by probability weighting to obtain the basic risk expectation value. Then, the spatial proximity weight of the edges is fused to obtain the basic risk coupling value. Through traversal, weighted summation and multiplication fusion, the comprehensive risk of hazardous waste packaging attribute coupling and spatial correlation is accurately quantified, providing reliable basic risk data for environmental factor correction and violation judgment.

[0097] In one embodiment, S32 includes:

[0098] S321. Based on the attribute probability distribution vector, calculate the combination probability of the node category probability to obtain the attribute combination probability.

[0099] Optionally, the node category probability is a single element in the attribute probability distribution vector. Each element corresponds to the probability that the hazardous waste packaging of the node belongs to a certain preset hazardous waste category. The dimension of the attribute probability distribution vector is the same as the total number of preset hazardous waste categories, and the sum of all elements is 1. The combination probability calculation does not involve pairwise multiplication. Specifically, the attribute probability distribution vectors of the two nodes are extracted. The probability values ​​of all categories in the attribute probability distribution vector of the first node and the attribute probability distribution vector of the second node are enumerated. Each category probability of the first node is multiplied by each category probability of the second node to obtain the probability value corresponding to each category combination. This probability value is the attribute combination probability. For example, if the attribute probability distribution vector of the first node contains K category probabilities, and the attribute probability distribution vector of the second node also contains K category probabilities, then K×K attribute combination probabilities can be obtained. Each attribute combination probability uniquely corresponds to a hazardous waste attribute combination of "a certain category of the first node - a certain category of the second node".

[0100] S322. Based on the hazardous waste taboo intensity matrix and attribute combination probability, perform a matching query on the hazard level to obtain the target hazard level.

[0101] Optionally, the hazardous waste incompatibility strength matrix is ​​a two-dimensional matrix pre-constructed based on the chemical properties, reactivity characteristics, and safe storage specifications of various types of hazardous waste. The rows and columns of the matrix correspond to preset hazardous waste categories, and each element in the matrix... All characterize the first Class II Hazardous Waste and Class III The hazard level of a hazardous chemical reaction that may occur when adjacent hazardous wastes are stored is determined by pre-setting the severity and harmful consequences of the reaction. Matching queries locate target elements in the hazardous waste incompatibility matrix based on the hazardous waste categories corresponding to attribute combinations. Specifically, each attribute combination corresponds to a specific set of hazardous waste categories (the first...). Class, No. (Class), based on the combination of this category, directly query the first element in the hazardous waste taboo strength matrix. Line number The value of each element in the column represents the target hazard level corresponding to that attribute combination. Furthermore, if the order of the categories corresponding to the attribute combinations is different, the matrix elements queried will also be different, ensuring the accuracy of hazard level matching.

[0102] S323. Based on the attribute combination probability and the target hazard level, the risk values ​​of each attribute combination are accumulated to obtain the basic expected risk value between nodes. The formula for calculating the basic expected risk value is as follows:

[0103]

[0104] in, Basic risk expectation, and Each of the two nodes belongs to the first... Class and First The probability value of the class. For the first Class II hazardous waste and Class III Hazard classification of hazardous waste This is the preset total number of hazardous waste categories.

[0105] Optionally, the cumulative calculation involves summing the risk values ​​of each attribute combination to obtain the basic expected risk value between the two nodes. The risk value of each attribute combination is obtained by multiplying the probability of that attribute combination by the target hazard level. This calculation quantifies the contribution of different attribute combinations to the overall risk. Specifically, for each attribute combination, the probability of the attribute combination is multiplied by the target hazard level to obtain the individual risk value of that attribute combination; then, all attribute combinations are iterated over, and all individual risk values ​​are summed sequentially. The final sum is the basic expected risk value between the nodes. For the first The expected value of the basic risk between node j and node j. For the first The node belongs to the node. The probability value of hazardous waste. For the j-th node to belong to the j-th node The probability value of hazardous waste. For the first Class II hazardous waste and Class III Hazard classification of hazardous waste The total number of hazardous waste categories is preset. The accumulation process follows a double-loop logic based on a given formula, first fixing the total number of categories. Value, iterate through all After completing the inner accumulation, iterate through all values ​​in sequence. The outer layer is accumulated to ensure that the risk values ​​of all attribute combinations are included in the calculation, without omission or duplication.

[0106] In the above embodiment, the attribute combination probability is obtained by combining the node category probabilities, the target hazard level is obtained by matching the hazardous waste taboo strength matrix, and the basic risk expectation value is obtained by accumulating the risk values ​​of each attribute combination. This embodiment clearly quantifies the risk contribution of different hazardous waste attribute combinations, taking into account both probability accuracy and hazard level matching precision.

[0107] In one embodiment, S4 includes:

[0108] S41. Based on the real-time environmental data, the real-time environmental data includes the temperature, humidity and characteristic gas concentration data of the current area.

[0109] Optionally, characteristic gas concentration data refers to the concentration data of specific gases that may be generated by chemical reactions of hazardous waste within the hazardous waste storage area and can characterize the degree of risk. The types of these gases are pre-determined based on preset hazardous waste categories to ensure that the collected data is directly related to the potential reaction risks of hazardous waste. Specifically, by synchronously reading real-time data transmitted from preset environmental sensing devices, the temperature, humidity, and characteristic gas concentration data of the current area are extracted respectively, ensuring that the collection timestamps of the three types of data are consistent, and the data is preprocessed after reading.

[0110] S42. Based on the temperature, humidity and characteristic gas concentration data, the environmental correction factor is calculated.

[0111] Optionally, the environmental correction factor is a comprehensive coefficient used to quantify the impact of real-time environmental conditions on the risk of hazardous waste chemical reactions. It is obtained based on a preset linear correlation by fitting the relationship between the degree of deviation of different environmental parameters from the baseline value and the hazardous waste reaction rate. The preset reference temperature and reference humidity are fixed values ​​set based on hazardous waste storage safety regulations and suitable environmental conditions for stable hazardous waste storage. Their settings are based on the chemical characteristics and storage requirements of different hazardous wastes to ensure accurate reflection of the safe environmental baseline for hazardous waste storage. First, based on the aforementioned reference temperature and reference humidity, the difference between the current temperature and current humidity is calculated to obtain temperature deviation and humidity deviation values. These deviation values ​​characterize the degree of deviation of the current environmental parameters from the safe baseline. Next, individual correction calculations are performed on the temperature deviation, humidity deviation, and characteristic gas concentration data. The temperature correction is obtained by combining the temperature deviation with a preset temperature influence coefficient; the humidity correction is obtained by combining the humidity deviation with a preset humidity influence coefficient; and the gas concentration correction is obtained by combining the characteristic gas concentration data with a preset gas concentration sensitivity coefficient. All of these influence coefficients are fitted based on the chemical reaction characteristics of hazardous waste and are used to quantify the influence weight of each environmental parameter on the hazardous waste reaction risk. Finally, the three individual correction items are multiplied to obtain a comprehensive environmental correction factor. This multiplication operation reflects the synergistic influence of each environmental parameter on hazardous waste risk, ensuring that the environmental correction factor can comprehensively characterize the degree of influence of real-time environmental conditions.

[0112] S43. Based on the basic risk coupling value and the environmental correction factor, perform product calculation to generate a dynamic risk value.

[0113] Optionally, the basic risk coupling value is obtained by traversing the edges of the dynamic spatiotemporal topology network, calculating the probability of attribute combinations, matching hazard levels, and accumulating the expected basic risk value. This value is then fused with the spatial proximity weights of the edges to characterize the potential chemical reaction risk of hazardous waste packaging based on attribute coupling and spatial association. This value is pre-stored and can be directly accessed. Specifically, the stored basic risk coupling value is directly accessed and multiplied by an environmental correction factor to generate a dynamic risk value. By quantifying and fusing the spatial risk of the hazardous waste's attributes with the impact of the real-time environment, the dynamic risk value can truly reflect the actual risk of hazardous waste storage under current environmental conditions, avoiding misjudgments caused by considering only spatial attribute risk while ignoring environmental impact.

[0114] In the above embodiments, by acquiring the current regional temperature, humidity and characteristic gas concentration data, an environmental correction factor is calculated based on a preset correlation function. Then, the basic risk coupling value is multiplied by the environmental correction factor to generate a dynamic risk value. This embodiment accurately integrates the impact of environmental factors on hazardous waste risk, making risk quantification more in line with actual storage scenarios and effectively improving the authenticity and accuracy of hazardous waste storage risk identification.

[0115] In one embodiment, S42 includes:

[0116] S421. Based on the preset reference temperature and reference humidity, calculate the difference between the temperature and humidity of the current area to obtain the temperature deviation value and humidity deviation value.

[0117] Optionally, the reference temperature and reference humidity are fixed values ​​pre-set based on hazardous waste storage safety regulations and suitable environmental conditions for stable hazardous waste storage. These values ​​are set according to the chemical characteristics and storage requirements of different hazardous wastes to ensure accurate reflection of the safe environmental baseline for hazardous waste storage. Specifically, the pre-stored reference temperature and reference humidity are directly retrieved, along with the current area's temperature and humidity data. Differences are calculated separately: the current temperature is subtracted from the reference temperature to obtain the temperature deviation value, and the current humidity is subtracted from the reference humidity to obtain the humidity deviation value. The temperature deviation value characterizes the degree of deviation of the current temperature from the safe baseline, and the humidity deviation value characterizes the degree of deviation of the current humidity from the safe baseline. A positive deviation value indicates that the current environmental parameters are higher than the baseline, while a negative deviation value indicates that the current environmental parameters are lower than the baseline.

[0118] S422. Perform individual correction calculations on the temperature deviation value, humidity deviation value, and characteristic gas concentration data respectively to obtain the temperature correction term, humidity correction term, and gas concentration correction term.

[0119] Optionally, the individual correction calculation adopts a linear correlation approach, constructing independent calculation logics for temperature deviation, humidity deviation, and characteristic gas concentration data respectively, ensuring that the impact of each environmental parameter can be accurately quantified. The temperature influence coefficient, humidity influence coefficient, and gas concentration sensitivity coefficient are fixed coefficients pre-fitted based on the chemical reaction characteristics of hazardous waste, and their values ​​are determined according to the sensitivity of different hazardous waste types to temperature, humidity, and characteristic gas concentration. Specifically, multiplying the temperature deviation value by the temperature influence coefficient and adding 1 yields the temperature correction term, which characterizes the degree of impact of temperature deviation from the baseline on hazardous waste risk; multiplying the humidity deviation value by the humidity influence coefficient and adding 1 yields the humidity correction term, which characterizes the degree of impact of humidity deviation from the baseline on hazardous waste risk; and multiplying the characteristic gas concentration data by the gas concentration sensitivity coefficient and adding 1 yields the gas concentration correction term, which characterizes the real-time impact of characteristic gas concentration on hazardous waste risk.

[0120] S423. Multiply the temperature correction term, humidity correction term, and gas concentration correction term to obtain the environmental correction factor. The formula for calculating the environmental correction factor is:

[0121]

[0122] in, As an environmental correction factor, and These are the current temperature and humidity, respectively. and These are the reference temperature and reference humidity, respectively. Characteristic gas concentration, , and These are the temperature influence coefficient, humidity influence coefficient, and gas concentration sensitivity coefficient, respectively.

[0123] Optionally, the product operation synergistically integrates the impacts of three environmental parameters—temperature, humidity, and characteristic gas concentration—on hazardous waste risk. Since the effects of these three environmental parameters on the chemical reactions of hazardous waste are interactive and additive, the product method accurately reflects this synergistic effect, ensuring that the environmental correction factor comprehensively characterizes the combined impact of the real-time environment on hazardous waste risk. Specifically, the temperature correction term, humidity correction term, and gas concentration correction term are directly invoked, and the product operation is performed sequentially. First, the temperature correction term is multiplied by the humidity correction term to obtain their synergistic influence coefficient. Then, this coefficient is multiplied by the gas concentration correction term, and the final product is the environmental correction factor. Performing the multiplication operation sequentially according to the given calculation formula ensures computational efficiency and result accuracy, while also ensuring that the environmental correction factor accurately matches real-time environmental conditions.

[0124] In the above embodiment, temperature and humidity deviation values ​​are calculated by preset reference benchmarks, and three individual correction calculations are completed based on the influence coefficients obtained by fitting. Then, the environmental correction factor is obtained by product operation. This embodiment accurately quantifies the synergistic impact of each environmental parameter on hazardous waste risk, making the environmental correction factor more consistent with the actual storage scenario.

[0125] In one embodiment, S6 includes:

[0126] S61. Compare the cumulative risk value with the violation threshold to obtain the comparison result.

[0127] Optionally, the violation threshold is a fixed critical value pre-set based on the hazard level, storage safety regulations, and risk control requirements of different hazardous waste categories. Each preset hazardous waste category corresponds to a unique violation threshold, which is set based on the chemical characteristics, reaction risks, and storage limits of the hazardous waste to ensure accurate differentiation between compliance and violation status. Specifically, the cumulative risk value is directly called, and according to the hazardous waste packaging category corresponding to the cumulative risk value, the pre-stored violation threshold for the corresponding category is called. Through numerical comparison, the comparison only judges the relationship between the cumulative risk value and the violation threshold, resulting in three comparison results: cumulative risk value greater than the violation threshold, cumulative risk value equal to the violation threshold, and cumulative risk value less than the violation threshold. Among these, a cumulative risk value greater than the violation threshold indicates that the violation condition is met, while the other two indicate that the violation condition is not met.

[0128] S62. When the comparison results meet the violation conditions, calculate the risk contribution of each hazardous waste packaging node to obtain the risk source location information.

[0129] Optionally, the risk contribution rate refers to the proportion of a single hazardous waste packaging node's contribution to the overall cumulative risk value. It is used to accurately pinpoint the core risk sources causing violations. The calculation is performed through proportional allocation. Specifically, first, the stored individual cumulative risk value corresponding to each hazardous waste packaging node is retrieved. This value represents the risk accumulation generated by each node during the time-series cumulative integration process. Then, the overall cumulative risk value is retrieved, and the individual cumulative risk value of each node is divided by the overall cumulative risk value to obtain the risk contribution rate of each node. The calculation formula is as follows:

[0130]

[0131] in, The risk contribution of the i-th hazardous waste packaging node. Let be the cumulative risk value for the i-th hazardous waste packaging node. This represents the overall cumulative risk value. After calculation, the risk contribution of all hazardous waste packaging nodes is ranked, and nodes with higher risk contribution are selected as core risk sources. The unique identifier of each node, its risk contribution value, and the corresponding hazardous waste attribute information are integrated to obtain risk source location information. This information can accurately pinpoint the specific hazardous waste packaging that caused the violation.

[0132] S63. Compare and encapsulate the results and risk source location information to obtain the violation identification results.

[0133] Optionally, encapsulation involves organizing the comparison results and risk source location information into structured output data. Specifically, the comparison results are first formalized into text, clearly indicating the hazardous waste category, violation threshold, and comparison conclusion corresponding to the current cumulative risk value, i.e., whether the violation conditions are met. Then, the risk source location information is associated with this comparison conclusion, ensuring that each comparison result corresponds to a unique risk source. If the comparison result does not meet the violation conditions, the risk source location information is empty. Further, all associated data is organized according to a preset format, clarifying the meaning and order of data fields to form a complete violation identification result. This result includes key information such as the comparison conclusion, hazardous waste category, violation threshold (if applicable), risk source node identifier, and risk contribution, ensuring that the output data is clear and standardized, directly providing clear evidence for violation judgment and risk tracing in hazardous waste storage management.

[0134] In the above embodiments, a clear comparison result is obtained by comparing the accumulated risk value with the preset violation threshold. When the violation conditions are met, the risk contribution of each hazardous waste packaging node is calculated and the risk source is located. Finally, the comparison result and the risk source location information are combined to obtain the violation behavior identification result. This achieves accurate judgment of violation behavior and rapid tracing of risk source, provides a clear basis for the illegal disposal of hazardous waste storage, and improves the pertinence and safety of hazardous waste storage management.

[0135] In the aforementioned method, system, equipment, and medium for identifying violations in hazardous waste storage, a digital twin of hazardous waste attributes containing attribute probability distribution vectors and uncertainty labels is constructed by performing target detection, feature mapping, nonlinear activation, and Dirichlet distribution transformation on images of hazardous waste packaging. Then, a dynamic spatiotemporal topology network is built using hazardous waste packaging as nodes and spatial proximity relationships as edges. The node attribute probability distribution vectors are obtained by traversing the network edges. A basic risk expectation value is obtained through attribute combination probability calculation, hazard level matching, and accumulation. A basic risk coupling value is obtained by fusing spatial proximity weights. Combining real-time temperature, humidity, and characteristic gas concentrations, an environmental correction factor is obtained through deviation calculation, single-item correction, and multiplication operations. This corrects the basic risk coupling value, generating a dynamic risk value. A cumulative risk value is obtained through time-series cumulative integration. Finally, through violation threshold comparison, risk contribution calculation, and result encapsulation, the identification of violations and the location of risk sources are completed. This technical solution addresses the technical problems of existing technologies, such as relying on deterministic feature matching and single-point threshold determination, being prone to missed detections and misjudgments, and being unable to effectively explore potential risks caused by spatial proximity coupling and environmental impacts, by accurately characterizing the uncertainty of hazardous waste properties, mining the spatial proximity correlation of nodes, combining dynamic risk correction in the environmental field, and quantifying the risk accumulation effect over time. It enables accurate identification and early warning of potential risks in hazardous waste storage.

[0136] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0137] Based on the same inventive concept, this application also provides a system for implementing the above-mentioned method for identifying violations in hazardous waste storage. The solution provided by this system is similar to the solution described in the above method; therefore, the specific limitations of one or more embodiments of the hazardous waste storage violation identification system provided below can be found in the above-described limitations of the hazardous waste storage violation identification method, and will not be repeated here.

[0138] In one exemplary embodiment, such as Figure 3 As shown, a hazardous waste storage violation identification system 10 is provided, including:

[0139] The attribute probability distribution calculation module 11 is used to acquire images of hazardous waste packaging, calculate the attribute probability distribution of the images of hazardous waste packaging, and obtain a digital twin of hazardous waste attributes. The digital twin of hazardous waste attributes includes an attribute probability distribution vector and an uncertainty label.

[0140] Spatial topology construction module 12 is used to construct a spatial topology for the digital twin of hazardous waste attributes, with hazardous waste packaging as nodes and spatial proximity as edges, to generate a dynamic spatiotemporal topology network.

[0141] The risk coupling integral calculation module 13 is used to perform risk coupling integral calculation on the attribute probability distribution between nodes based on the preset hazardous waste taboo strength matrix and dynamic spatiotemporal topology network to obtain the basic risk coupling value, which is used to characterize the potential chemical reaction risk.

[0142] The environmental field factor correction module 14 is used to correct the basic risk coupling value based on real-time environmental data and generate a dynamic risk value. The dynamic risk value is used to reflect the real risk under the current environmental conditions.

[0143] The time-series cumulative integration module 15 is used to perform time-series cumulative integration on the dynamic risk value to obtain the cumulative risk value;

[0144] The violation threshold determination module 16 is used to determine the violation threshold based on the cumulative risk value and obtain the violation behavior identification result.

[0145] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the hazardous waste storage violation identification method as described above.

[0146] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the hazardous waste storage violation identification method as described above.

[0147] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0148] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for identifying violations in hazardous waste storage, characterized in that, The method includes: S1. Obtain an image of a hazardous waste packaging piece, calculate the attribute probability distribution of the image of the hazardous waste packaging piece, and obtain a digital twin of hazardous waste attributes. The digital twin of hazardous waste attributes includes an attribute probability distribution vector and an uncertainty label. S2. Construct a dynamic spatiotemporal topology network by using hazardous waste packaging as nodes and spatial proximity relationships as edges on the digital twin of the hazardous waste attributes. S3. Based on the preset hazardous waste taboo strength matrix and the dynamic spatiotemporal topology network, perform risk coupling integral calculation on the attribute probability distribution between nodes to obtain the basic risk coupling value, which is used to characterize the potential chemical reaction risk. S4. Based on real-time environmental data, the basic risk coupling value is corrected by environmental field factors to generate a dynamic risk value, which is used to reflect the real risk under the current environmental conditions. S5. Perform time-series cumulative integration on the dynamic risk value to obtain the cumulative risk value; S6. Determine the violation threshold based on the accumulated risk value to obtain the violation identification result.

2. The method according to claim 1, characterized in that, S1 includes: S11. Perform target detection on the image of the hazardous waste packaging to obtain the target region of the image of the hazardous waste packaging; S12. Perform feature mapping on the target region of the image to obtain a feature vector; S13. Perform nonlinear activation on the feature vector to obtain class evidence values ​​corresponding to various hazardous waste attributes; S14. Based on the Dirichlet distribution parameters, construct rules and perform distribution transformation on the class evidence values ​​to obtain the digital twin of the hazardous waste attributes.

3. The method according to claim 1, characterized in that, S3 includes: S31. Traverse the edges in the dynamic spatiotemporal topology network to obtain the attribute probability distribution vectors of the two nodes connected by the edges; S32. Based on the hazardous waste prohibition intensity matrix and the attribute probability distribution vector, perform probability weighted summation on the attribute combination to obtain the basic risk expectation value between nodes; S33. The basic risk expectation value is fused with the spatial proximity weight of the edge to obtain the basic risk coupling value.

4. The method according to claim 3, characterized in that, S32 includes: S321. Based on the attribute probability distribution vector, calculate the combination probability of the node category probability to obtain the attribute combination probability; S322. Based on the hazardous waste taboo strength matrix and the attribute combination probability, perform a matching query on the hazard level to obtain the target hazard level; S323. Based on the attribute combination probability and the target hazard level, the risk values ​​of each attribute combination are accumulated to obtain the basic expected risk value between nodes. The formula for calculating the basic expected risk value is as follows: in, Basic risk expectation, and Each of the two nodes belongs to the first... Class and First The probability value of the class. For the first Class II hazardous waste and Class III Hazard classification of hazardous waste This is the preset total number of hazardous waste categories.

5. The method according to claim 1, characterized in that, S4 includes: S41. Based on the real-time environmental data, the real-time environmental data includes the temperature, humidity and characteristic gas concentration data of the current area; S42. Calculate the environmental correction factor based on the temperature, humidity, and characteristic gas concentration data; S43. Based on the basic risk coupling value and the environmental correction factor, perform a product calculation to generate the dynamic risk value.

6. The method according to claim 5, characterized in that, S42 includes: S421. Based on the preset reference temperature and reference humidity, calculate the difference between the temperature and humidity of the current area to obtain the temperature deviation value and humidity deviation value. S422. Perform individual correction calculations on the temperature deviation value, the humidity deviation value, and the characteristic gas concentration data respectively to obtain the temperature correction term, humidity correction term, and gas concentration correction term; S423. Multiply the temperature correction term, the humidity correction term, and the gas concentration correction term to obtain the environmental correction factor. The formula for calculating the environmental correction factor is as follows: in, As an environmental correction factor, and These are the current temperature and humidity, respectively. and These are the reference temperature and reference humidity, respectively. Characteristic gas concentration, , and These are the temperature influence coefficient, humidity influence coefficient, and gas concentration sensitivity coefficient, respectively.

7. The method according to claim 1, characterized in that, S6 includes: S61. Compare the accumulated risk value with the violation threshold to obtain the comparison result; S62. When the comparison result meets the violation conditions, the risk contribution of each hazardous waste packaging node is calculated to obtain the risk source location information; S63. The comparison result and the risk source location information are encapsulated to obtain the violation identification result.

8. A system for identifying violations in hazardous waste storage, characterized in that, The system includes: The attribute probability distribution calculation module is used to acquire images of hazardous waste packaging, perform attribute probability distribution calculation on the images of hazardous waste packaging, and obtain a digital twin of hazardous waste attributes. The digital twin of hazardous waste attributes includes an attribute probability distribution vector and an uncertainty label. The spatial topology construction module is used to construct a dynamic spatiotemporal topology network for the digital twin of hazardous waste attributes, with hazardous waste packaging as nodes and spatial proximity relationships as edges. The risk coupling integral calculation module is used to perform risk coupling integral calculation on the attribute probability distribution between nodes based on the preset hazardous waste taboo strength matrix and the dynamic spatiotemporal topology network, so as to obtain the basic risk coupling value, which is used to characterize the potential chemical reaction risk. The environmental field factor correction module is used to correct the basic risk coupling value based on real-time environmental data and generate a dynamic risk value, which is used to reflect the real risk under the current environmental conditions. The time-series cumulative integration module is used to perform time-series cumulative integration on the dynamic risk value to obtain the cumulative risk value; The violation threshold determination module is used to determine the violation threshold of the accumulated risk value and obtain the violation behavior identification result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images