A method and system for intelligent treatment of hazardous waste based on knowledge graph

Abstract

The present invention discloses a method and system for intelligently treating hazardous waste based on a knowledge graph. After receiving hazardous waste transported by a waste-producing unit, the hazardous waste is intelligently classified in advance based on the basic data provided by the waste-producing unit, and is matched with the corresponding hazardous waste treatment process through an intelligent recommendation processing algorithm. There is no need to specify the corresponding process after manual review, which solves the problems of cumbersome manual recording procedures, low work efficiency and easy errors, and prevents environmental pollution or personnel safety hazards caused by the loss of hazardous waste. Supervision, disposal and traceability can be implemented by reviewing the historical data in the operator's receiving terminal.

Description

Technical Field

[0001] The present invention belongs to the technical field of safe management of hazardous waste, and specifically relates to a method and system for intelligent treatment of hazardous waste based on a knowledge graph. Background Art

[0002] Most hazardous chemical wastes are a multi-component mixture, which also leads to the existence of multiple hazardous properties and unstable characteristics in the same hazardous chemical waste, making the safe management of hazardous chemical wastes more difficult. Due to the diverse types of hazardous chemical wastes, diverse hazards, diverse packaging forms, different professional capabilities of managers, and the influence of storage environments such as weather and facilities and equipment, the storage link of hazardous chemical wastes is the link in the whole process of hazardous chemical waste management where safety accidents are most likely to occur, and there are many safety accident lessons.

[0003] At present, in the actual storage management of hazardous waste, warehouse managers generally divide the warehouse, storage area and storage location into functional groups. Storage areas are generally divided according to the physical state of hazardous waste, disposal process flow or simple hazardous waste properties, and storage locations are divided according to the fire distance requirements of each enterprise's own regulations and building fire protection regulations. Such warehouse planning does not take into account the specific hazards of hazardous waste and the mixed storage and compatibility of hazardous waste. In the case of special materials or management negligence, it is very easy to cause safety accidents, increasing the cost and pressure of enterprise safety management.

[0004] In addition, the existing method of collecting and treating hazardous chemical waste is: first, qualified collection units go to the hazardous chemical waste generating station to collect, and during collection, operators manually record relevant information such as the generation time, generating unit, type, and weight of hazardous solid and liquid waste; collection operators submit relevant documents in paper form, and hazardous chemical waste is sent to the treatment process for treatment.

[0005] The above-mentioned processing steps mainly have the following problems: since hazardous chemical waste involves a wide range of industries, a wide variety of types, a large amount of production, and is difficult to supervise, and the current treatment process cannot intelligently classify hazardous chemical waste in advance to make it correspond to the corresponding hazardous chemical waste treatment process, resulting in the need for processing personnel to manually review the relevant documents submitted in paper form and then specify the corresponding process. In this process, the manual recording procedure is cumbersome, the work efficiency is low, and it is prone to errors. In addition, when hazardous chemical waste is lost and causes environmental pollution or personnel safety hazards, it is difficult to implement supervision, disposal and traceability.

[0006] Therefore, it is necessary to solve the limitations of existing technologies through a knowledge graph-based intelligent treatment method and system for hazardous chemical waste, to intelligently classify hazardous chemical waste and allocate treatment processes, thereby playing an important role in promoting the supervision and management of hazardous waste in my country. Summary of the invention

[0007] In view of the above-mentioned deficiencies in the prior art, the present application provides a method and system for intelligent treatment of hazardous waste based on a knowledge graph.

[0008] In the first aspect, the present application proposes a method for intelligent treatment of hazardous waste based on a knowledge graph, comprising the following steps:

[0009] Obtain the hazardous waste knowledge data involved in the target waste-producing units and build a hazardous waste knowledge graph;

[0010] Obtaining basic data of the hazardous waste to be processed, and determining the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph;

[0011] Acquire image data corresponding to the hazardous chemical waste to be processed, perform morphological recognition on the image data through machine vision technology to obtain a morphological recognition result of the hazardous chemical waste to be processed, and perform machine labeling on the hazardous chemical waste to be processed based on the morphological recognition result to obtain a morphological label of the hazardous chemical waste to be processed;

[0012] Constructing an intelligent recommendation processing algorithm, inputting the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed into the intelligent recommendation processing algorithm to obtain the hazardous waste intelligent recommendation processing results;

[0013] According to the intelligent recommended treatment result of hazardous waste, a corresponding hazardous waste treatment process is allocated to the hazardous waste to be treated, and the process is sent to a receiving terminal of an operator, so that the hazardous waste is treated accordingly.

[0014] In some embodiments, the step of obtaining hazardous waste knowledge data related to the target waste-generating unit and constructing a hazardous waste knowledge graph includes:

[0015] Sentence processing is performed on the hazardous waste knowledge data to establish a hazardous waste generation process table, wherein the hazardous waste generation process table includes entities and attributes;

[0016] Establish an association relationship between entities and attributes, and establish a hazardous waste knowledge graph based on the association relationship.

[0017] In some embodiments, the obtaining of basic data of the hazardous waste to be processed and determining the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph include:

[0018] Determine whether the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base;

[0019] If the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base, determine the type of hazardous waste to which the hazardous waste to be processed belongs according to the preset hazardous waste knowledge base;

[0020] If the hazardous waste to be processed cannot be matched with the hazardous waste information in the preset hazardous waste knowledge base, the type of hazardous waste to which the hazardous waste to be processed belongs is determined based on the hazardous waste knowledge graph and the preset hazardous waste knowledge base.

[0021] In some embodiments, the acquiring of image data corresponding to the hazardous waste to be processed, performing morphological recognition on the image data by machine vision technology to obtain a morphological recognition result of the hazardous waste to be processed, and machine labeling the hazardous waste to be processed based on the morphological recognition result to label the morphological label for the hazardous waste to be processed includes:

[0022] Receive image data corresponding to the hazardous waste to be processed, extract, eliminate, and color cluster the image target morphology on the image data to obtain the two-dimensional coordinates, angle, and time of the detection target on the image in the image coordinate system;

[0023] Focus on tracking and displaying the detected target on the image data to determine whether the target information is consistent with

[0024] The target information identified in the previous frame is the same target. If it is not the same target, the processed image and the coordinates, angles and time information of the detected target on the image are manually corrected to obtain the corrected coordinate information;

[0025] The error correction coordinate information is converted into an image coordinate system to obtain the coordinates of the error target or the newly added target, and the information of the error target is removed from the coordinates, angles and time information of the detection target in each frame of the received image in the image coordinate system to obtain the coordinates, angles and time information of the identified target in the image coordinate system, and at the same time, the emphasis on the error target is removed from the received visual feedback image to complete the morphological recognition and obtain the morphological recognition result;

[0026] Based on the morphological recognition result, the hazardous waste to be processed is machine labeled to obtain a morphological label of the hazardous waste to be processed.

[0027] In some embodiments, the intelligent recommendation processing algorithm is constructed, and the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm. Before obtaining the intelligent recommendation processing results of hazardous waste, the following steps are further included:

[0028] Constructing a feature vector extraction model, wherein the feature vector extraction model includes an input layer, a convolution layer, a normalization layer, a pooling layer, and a fully connected layer;

[0029] The output end of the input layer is connected to the input end of the convolution layer through an information transmission channel; the output end of the convolution layer is connected to the input end of the normalization layer through an information transmission channel; the output end of the normalization layer is connected to the input end of the pooling layer through an information transmission channel; the output end of the pooling layer is connected to the input end of the fully connected layer through an information transmission channel;

[0030] Obtain basic data of hazardous waste corresponding to the type of hazardous waste; transmit the basic data of hazardous waste corresponding to the type of hazardous waste as input to the input layer, and finally obtain the feature vector of the hazardous waste to be processed after feature extraction through the fully connected layer.

[0031] In some embodiments, the step of constructing an intelligent recommendation processing algorithm includes:

[0032] An input relationship graph is constructed based on the hazardous waste classification results, the hazardous waste form labels to be processed, the preset hazardous waste treatment process and the feature vectors of the hazardous waste to be processed as basic elements, wherein the hazardous waste types in the hazardous waste classification results are used as nodes, the attributes of each node correspond to the feature vectors of the hazardous waste to be processed, the hazardous waste form labels to be processed are used as labels on the nodes, and the preset hazardous waste treatment process is used as edges between the nodes;

[0033] If two hazardous waste types have the same label, it is considered that the two hazardous waste types are applicable to the same hazardous waste treatment process, and an edge is created for the corresponding two nodes;

[0034] A graph encoder and a graph decoder are formed according to the input relationship graph, and the graph encoder and the graph decoder together constitute the intelligent recommendation processing algorithm.

[0035] In some embodiments, the hazardous waste classification results, the hazardous waste form labels to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm to obtain the hazardous waste intelligent recommendation processing results, including:

[0036] Step 1: Input the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the feature vectors of the hazardous waste to be processed into the graph encoder and the graph decoder, and initialize the graph encoder, the graph decoder and the model parameters;

[0037] Step 2: Calculate the feature representation corresponding to any node in the graph encoder;

[0038] Step 3: Calculate the output and loss of the softmax layer for node classification in the graph encoder according to the feature representation;

[0039] Step 4: Calculate the softmax layer output and loss of link prediction in the graph decoder according to the feature representation;

[0040] Step 5: Calculate the total target loss;

[0041] Step 6: Use the gradient descent algorithm to update the model parameters and optimize the total target loss;

[0042] Step 7: Output the optimal parameter vectors of the graph encoder and the graph decoder to obtain a classification prediction network for node classification and link prediction;

[0043] Step 8: Import the classification prediction network into the graph neural network model to obtain a more standard node embedding, and use the machine learning algorithm in the downstream task to classify and predict the corresponding relationship between nodes and edges; thereby obtaining the intelligent recommended treatment results of hazardous waste.

[0044] In the second aspect, the present application proposes a hazardous waste intelligent treatment system based on a knowledge graph, including a graph construction module, a data classification module, a morphology recognition and annotation module, an intelligent prediction and recommendation module, and a data transmission module;

[0045] The graph construction module is used to obtain the hazardous waste knowledge data involved in the target waste-producing unit and construct a hazardous waste knowledge graph;

[0046] The data classification module is used to obtain basic data of the hazardous waste to be processed, and determine the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph;

[0047] The morphological recognition and labeling module is used to obtain image data corresponding to the hazardous chemical waste to be processed, perform morphological recognition on the image data through machine vision technology to obtain a morphological recognition result of the hazardous chemical waste to be processed, and perform machine labeling on the hazardous chemical waste to be processed based on the morphological recognition result to obtain a morphological label of the hazardous chemical waste to be processed;

[0048] The intelligent prediction and recommendation module is used to construct an intelligent recommendation processing algorithm, input the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the feature vectors corresponding to the hazardous waste to be processed into the intelligent recommendation processing algorithm, and obtain the intelligent recommendation processing results of the hazardous waste;

[0049] The data transmission module is used to allocate the corresponding hazardous waste treatment process to the hazardous waste to be treated according to the intelligent recommended treatment result of the hazardous waste, and send it to the receiving terminal of the operator, so as to perform corresponding treatment on the hazardous waste.

[0050] In a third aspect, the present application proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0051] In a fourth aspect, the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0052] Beneficial effects of the present invention:

[0053] After receiving the hazardous waste from the waste-producing units, the hazardous waste is intelligently classified in advance based on the basic data provided by the waste-producing units, and it is matched with the corresponding hazardous waste treatment process through the intelligent recommendation processing algorithm. There is no need to manually review and specify the corresponding process, which solves the problems of cumbersome manual recording procedures, low work efficiency and easy errors. It prevents environmental pollution or personnel safety hazards caused by the loss of hazardous waste. Supervision, disposal and traceability can be implemented by reviewing the historical data in the operator's receiving terminal. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is the overall flow chart of the present invention.

[0055] Figure 2 It is a system principle diagram of the present invention. DETAILED DESCRIPTION

[0056] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein; on the contrary, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0057] In the first aspect, this application proposes a method for intelligent treatment of hazardous waste based on knowledge graph, such as Figure 1 As shown, the following steps are included:

[0058] S100: Acquire the hazardous waste knowledge data involved in the target waste-producing unit and construct a hazardous waste knowledge graph;

[0059] In some embodiments, the step of obtaining hazardous waste knowledge data related to the target waste-generating unit and constructing a hazardous waste knowledge graph includes:

[0060] Sentence processing is performed on the hazardous waste knowledge data to establish a hazardous waste generation process table, wherein the hazardous waste generation process table includes entities and attributes;

[0061] Establish an association relationship between entities and attributes, and establish a hazardous waste knowledge graph based on the association relationship.

[0062] Among them, the hazardous waste knowledge data obtained in this embodiment includes the waste data provided by the waste generating unit, the General Principles of Hazardous Waste Identification Standards, the National Hazardous Waste List, and the technical specifications for pollutant discharge permits issued by the state and other relevant valid documents related to hazardous wastes;

[0063] Sentence processing refers to extracting data information such as the name of hazardous waste, name of waste-producing unit, industry source, hazardous waste code, hazardous waste category, etc. from the above data to establish a generation process table for each hazardous waste.

[0064] Table 1 is an example of the generation process of hazardous waste of waste category HW01 medical waste.

[0065] Table 1

[0066]

[0067] The entities include: name of hazardous waste, name of waste producing unit, industry source, hazardous waste code / hazardous waste category; the corresponding attributes include: medical waste, XX Medical Co., Ltd., hygiene, 841-001-01 / infectious waste, 841-002-01 / injurious waste, 841-003-01 / pathological waste, 841-004-01 / chemical waste, 841-005-01 / pharmaceutical waste.

[0068] Among them, the association relationships are: the name of hazardous waste, name of waste producing unit, industry source, hazardous waste code / hazardous waste category and medical waste, XX Medical Co., Ltd., sanitation, 841-001-01 / infectious waste, 841-002-01 / injurious waste, 841-003-01 / pathological waste, 841-004-01 / chemical waste, 841-005-01 / pharmaceutical waste.

[0069] For example:

[0070] The name of the hazardous waste transported by the waste-generating unit is medical waste;

[0071] The name of the waste generating unit is XX Medical Co., Ltd.;

[0072] The industry source of the waste generating unit is health;

[0073] The hazardous waste code / hazardous waste category corresponds to 841-001-01 / infectious waste.

[0074] A hazardous waste knowledge graph is established based on the above-mentioned association relationships, and the obtained entity and attribute data are used to construct the hazardous waste knowledge graph.

[0075] S200: Obtain basic data of the hazardous waste to be processed, and determine the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph;

[0076] Among them, the basic data of the hazardous waste to be processed includes the name of the hazardous waste, the name of the waste producing unit, the industry source, the hazardous waste code, the hazardous waste category and other data information mentioned above. The hazardous waste knowledge base used in this embodiment can be an internally built hazardous waste knowledge base or an external hazardous waste knowledge base. The embodiment of the present invention does not limit the source and specific structure of the hazardous waste knowledge base, as long as it can be used to query hazardous waste and determine the standard hazardous waste type.

[0077] In some embodiments, the obtaining of basic data of the hazardous waste to be processed and determining the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph include:

[0078] Determine whether the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base;

[0079] If the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base, determine the type of hazardous waste to which the hazardous waste to be processed belongs according to the preset hazardous waste knowledge base;

[0080] If the hazardous waste to be processed cannot be matched with the hazardous waste information in the preset hazardous waste knowledge base, the type of hazardous waste to which the hazardous waste to be processed belongs is determined based on the hazardous waste knowledge graph and the preset hazardous waste knowledge base.

[0081] Among them, since the existing hazardous waste knowledge base usually has limited items, there are many queried hazardous wastes that are not in the hazardous waste knowledge base, so it is impossible to query all queried hazardous wastes only through the hazardous waste knowledge base. In the embodiment of the present invention, when the queried hazardous waste cannot match the hazardous waste information in the hazardous waste knowledge base, other potential information about the queried hazardous waste is discovered through the knowledge graph, and the discovered potential information is used to correspond to the hazardous waste information in the hazardous waste knowledge base, so that a relatively accurate hazardous waste type result can be provided for various queried hazardous wastes, effectively improving the efficiency of hazardous waste query.

[0082] Furthermore, if the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base, the hazardous waste type to which the hazardous waste to be processed belongs is determined according to the preset hazardous waste knowledge base;

[0083] Among them, when the query question is "What kind of hazardous waste are cotton balls and cotton swabs?", it can be identified that the queried hazardous waste is "cotton balls and cotton swabs", and cotton balls and cotton swabs can be directly matched with the hazardous waste information "medical waste" in the hazardous waste knowledge base. Therefore, it can be directly determined based on the hazardous waste knowledge base that the target hazardous waste type to which cotton balls and cotton swabs belong is the hazardous waste type corresponding to "medical waste" - infectious waste.

[0084] When the hazardous waste being queried does not match the hazardous waste information in the hazardous waste knowledge base, that is, when the hazardous waste being queried does not directly belong to the existing hazardous waste information in the hazardous waste knowledge base, the one-dimensional information of the hazardous waste being queried can be determined based on the knowledge graph, and the one-dimensional information of the hazardous waste being queried can match the hazardous waste information in the hazardous waste knowledge base, and then the target hazardous waste category to which the hazardous waste being queried belongs is determined based on the one-dimensional information of the hazardous waste being queried through the hazardous waste knowledge base. The dimensional information in the embodiment of the present invention can be label information, category attribute information, hypernyms or synonyms, abbreviations, etc. of the hazardous waste being queried.

[0085] Specifically: determine and query multiple dimensions of hazardous waste information based on the knowledge graph;

[0086] Determine whether the multiple dimensions of hazardous waste information being queried match the hazardous waste information in the hazardous waste knowledge base;

[0087] When the multiple dimensional information of the queried hazardous waste matches the hazardous waste information in the hazardous waste knowledge base, the target hazardous waste category to which the queried hazardous waste belongs is determined through the hazardous waste knowledge base according to the multiple matching dimensional information of the queried hazardous waste.

[0088] In this embodiment, when the category attribute information, label information and hypernyms of the queried hazardous waste can all match the hazardous waste information in the hazardous waste knowledge base, the hazardous waste knowledge base can first determine one or more hazardous waste types to which the queried hazardous waste belongs according to multiple matching dimension information of the queried hazardous waste, and then determine the target hazardous waste type according to the frequency information of the hazardous waste type. The dimensional information includes: category attribute information and hypernyms.

[0089] Taking category attribute information as an example, since the category attribute information of hazardous waste is used to summarize a large range of information, the category attribute information is used as the dimensional information of hazardous waste to match the hazardous waste information in the hazardous waste knowledge base. The target hazardous waste types determined in the hazardous waste knowledge base based on the category attribute information of hazardous waste are often more accurate, and query errors caused by matching queries after narrowing semantics rarely occur. Specifically, when the hazardous waste to be queried is "Corica scalpel", since Corica scalpel is not in the hazardous waste knowledge base, the queried hazardous waste cannot be matched with the item information in the hazardous waste knowledge base. At this time, the category attribute information of the queried hazardous waste, that is, "damage waste", is determined according to the knowledge graph, and then it is judged whether the queried hazardous waste matches the hazardous waste information in the hazardous waste knowledge base. It is judged that "damage waste" matches the hazardous waste information "damage waste" in the hazardous waste knowledge base. At this time, the type of target hazardous waste can be determined through the hazardous waste knowledge base according to the category attribute information of the queried hazardous waste, that is, after determining that the hazardous waste type of "damage waste" in the hazardous waste knowledge base is "medical waste", the type of target hazardous waste is also "medical waste".

[0090] When hypernyms are used as one-dimensional information for querying hazardous waste, due to the same query

[0091] There may be multiple hypernyms for hazardous waste, for example, the hypernyms for "beaker" include "chemical measuring tools", "medical measuring tools", "physical measuring tools", etc. Therefore, it is necessary to first count the hazardous waste type information corresponding to the hypernyms of the query hazardous waste, and then determine the target hazardous waste type according to the matching frequency of the hazardous waste type information. In an embodiment of the present invention, determining the target hazardous waste type according to the matching frequency of the hazardous waste type information can specifically be to determine the hazardous waste type information with the highest matching frequency as the target hazardous waste type corresponding to the query hazardous waste, wherein, if the matching frequency information of the hazardous waste type information is the same, such as the hazardous waste type information is "chemical hazardous waste" and "medical hazardous waste" are both twice, then the two hazardous waste type information are jointly determined as the target hazardous waste type.

[0092] When the category attribute information or hypernym query does not match the hazardous waste information in the hazardous waste knowledge base, the query hazardous waste and each entity data in the hazardous waste knowledge base are converted into the query hazardous waste word vector and the knowledge base hazardous waste word vector respectively; the knowledge base hazardous waste word vector with the highest similarity to the query hazardous waste word vector is determined by performing word vector similarity calculation, so that the hazardous waste type corresponding to the hazardous waste information of the knowledge base hazardous waste word vector in the hazardous waste knowledge base is determined as the target hazardous waste type. The method of calculating word vector similarity can simply and effectively find hazardous waste information with a high matching degree in the hazardous waste knowledge base, thereby ensuring that the target hazardous waste type can be determined for various query hazardous wastes.

[0093] Methods for determining the similarity between the query hazardous waste word vector and the knowledge base hazardous waste word vector include: calculating the cosine similarity of the word vector, neural network judgment, and calculating the spatial distance. Among them, calculating the spatial distance of the word vector can also specifically include calculating the Euclidean distance, the Mahalanobis distance, and other methods. Of course, the similarity can also be calculated by using two or more of these similarity judgment methods at the same time, and each similarity calculation method can also be assigned a weight value, and finally the target hazardous waste type is determined based on the comprehensive calculation final value.

[0094] S300: Obtain image data corresponding to the hazardous waste to be processed, perform morphological recognition on the image data through machine vision technology to obtain a morphological recognition result of the hazardous waste to be processed, and perform machine labeling on the hazardous waste to be processed based on the morphological recognition result to obtain a morphological label of the hazardous waste to be processed;

[0095] In some embodiments, the acquiring of image data corresponding to the hazardous waste to be processed, performing morphological recognition on the image data by machine vision technology to obtain a morphological recognition result of the hazardous waste to be processed, and machine labeling the hazardous waste to be processed based on the morphological recognition result to label the morphological label for the hazardous waste to be processed includes:

[0096] Receive image data corresponding to the hazardous waste to be processed, extract, eliminate, and color cluster the image target morphology on the image data to obtain the two-dimensional coordinates, angle, and time of the detection target on the image in the image coordinate system;

[0097] Focus on tracking and displaying the detected target on the image data to determine whether the target information is consistent with

[0098] The target information identified in the previous frame is the same target. If it is not the same target, the processed image and the coordinates, angles and time information of the detected target on the image are manually corrected to obtain the corrected coordinate information;

[0099] The error correction coordinate information is converted into an image coordinate system to obtain the coordinates of the error target or the newly added target, and the information of the error target is removed from the coordinates, angles and time information of the detection target in each frame of the received image in the image coordinate system to obtain the coordinates, angles and time information of the identified target in the image coordinate system, and at the same time, the emphasis on the error target is removed from the received visual feedback image to complete the morphological recognition and obtain the morphological recognition result;

[0100] Based on the morphological recognition result, the hazardous waste to be processed is machine labeled to obtain a morphological label of the hazardous waste to be processed.

[0101] Among them, the main purpose of morphological recognition is to classify the morphology of hazardous waste to be treated after classification and treatment, and further classify the hazardous waste to be treated into solid waste and liquid waste, and mark the corresponding solid waste labels and liquid waste labels of the identification results of hazardous waste to be treated through machine.

[0102] S400: constructing an intelligent recommendation processing algorithm, inputting the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed into the intelligent recommendation processing algorithm, and obtaining the intelligent recommendation processing results of the hazardous waste;

[0103] In some embodiments, the intelligent recommendation processing algorithm is constructed, and the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm. Before obtaining the intelligent recommendation processing results of hazardous waste, the following steps are further included:

[0104] Constructing a feature vector extraction model, wherein the feature vector extraction model includes an input layer, a convolution layer, a normalization layer, a pooling layer, and a fully connected layer;

[0105] The output end of the input layer is connected to the input end of the convolution layer through an information transmission channel; the output end of the convolution layer is connected to the input end of the normalization layer through an information transmission channel; the output end of the normalization layer is connected to the input end of the pooling layer through an information transmission channel; the output end of the pooling layer is connected to the input end of the fully connected layer through an information transmission channel;

[0106] Obtain basic data of hazardous waste corresponding to the type of hazardous waste; transmit the basic data of hazardous waste corresponding to the type of hazardous waste as input to the input layer, and finally obtain the feature vector of the hazardous waste to be processed after feature extraction through the fully connected layer.

[0107] The basic data of the hazardous waste corresponding to the type of hazardous waste is transmitted to the input layer as input, and finally the feature vector of the hazardous waste to be processed is obtained after feature extraction through the fully connected layer.

[0108] Among them, the number of convolution kernels in the convolution layer is 25, and the number of neurons in the fully connected layer is 64. The input data of the input layer is the preprocessed frequency feature data, spatiotemporal feature data, and turnover feature data. The convolution layer consists of 25 convolution kernels with a step size of 1 and a size of 3×3, without an activation function. The normalization layer consists of 25 channels, each with a size of 3×3, with a ReLU function. The pooling layer consists of 25 pooling kernels with a size of 2×2, without an activation function, and the fully connected layer consists of 64 nodes with a ReLU function. After the valid data of the input layer is extracted by the convolution layer, the feature data output by the convolution layer is passed to the normalization layer for feature standardization, and then passed to the pooling layer for feature compression and information filtering. The output data of the pooling layer after compression and filtering is extracted by the fully connected layer to obtain the feature vector.

[0109] After the data processing in the above steps, the characteristic vector of the hazardous waste to be treated is finally obtained.

[0110] In some embodiments, the step of constructing an intelligent recommendation processing algorithm includes:

[0111] An input relationship graph is constructed based on the hazardous waste classification results, the hazardous waste form labels to be processed, the preset hazardous waste treatment process and the feature vectors of the hazardous waste to be processed as basic elements, wherein the hazardous waste types in the hazardous waste classification results are used as nodes, the attributes of each node correspond to the feature vectors of the hazardous waste to be processed, the hazardous waste form labels to be processed are used as labels on the nodes, and the preset hazardous waste treatment process is used as edges between the nodes;

[0112] If two hazardous waste types have the same label, it is considered that the two hazardous waste types are applicable to the same hazardous waste treatment process, and an edge is created for the corresponding two nodes;

[0113] Among them, the input relationship graph is constructed based on all the nodes, labels, edges and feature vectors created , Represents a collection of nodes. Represents a collection of attributes on a node. Represents the set of labels on the node, represents the set of labeled edges, represents the set of all edge types, Represents the set of all node types, Represents the edge weight matrix.

[0114] in, Represents a collection of nodes. Representation diagram The total number of nodes; Represents the attribute set on the node, that is Is a node The eigenvectors on Represents the set of labels on the node, node One-hot label vector on the node Belong to Class, then , otherwise, ;

[0115] At the same time, remember Represents the label set of labeled nodes in the graph; represents the set of labeled edges, , a triple Represents a relationship (edge ​​with a label); The number of elements of , The number of elements of , the ijth element of the weight matrix is Measured edge Node on and strength of relationship;

[0116] A graph encoder and a graph decoder are formed according to the input relationship graph, and the graph encoder and the graph decoder together constitute the intelligent recommendation processing algorithm.

[0117] Among them, the graph encoder is constructed according to the input relationship graph and Graph Decoder , by the graph encoder and Graph Decoder Together they constitute the intelligent recommendation processing algorithm, represents the input graph with attributes having multiple entity types and multiple relationship types, Represents all parameter matrices of the graph encoder The parameter vector is composed of Representation Graph Encoder The matrix composed of all output features of the last layer, , Represents all parameter matrices of the graph decoder The parameter vector.

[0118] In some embodiments, the hazardous waste classification results, the hazardous waste form labels to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm to obtain the hazardous waste intelligent recommendation processing results, including:

[0119] Step 1: Input the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the feature vectors of the hazardous waste to be processed into the graph encoder and the graph decoder, and initialize the graph encoder, the graph decoder and the model parameters;

[0120] Initialize the graph encoder and Graph Decoder parameter , ;

[0121] Step 2: Calculate the feature representation corresponding to any node in the graph encoder;

[0122] The specific formula is:

[0123]

[0124] in, , Is a node The graph encoder The output features of the layer, Representation Node The graph encoder The output features of the layer, Is a node The graph encoder The output features of the layer, It is The characteristic dimension of the layer, , represents the maximum activation function of each element, Representation and Nodes Related The index set of the node, Represents the normalization constant corresponding to the problem, which can be a learnable parameter or given in advance, for example, Representing a collection The number of elements, It is The unknown weight parameter matrix of the layer, is the number of hidden layers of the graph encoder.

[0125] Step 3: Calculate the output and loss of the softmax layer for node classification in the graph encoder according to the feature representation;

[0126] The specific formula for calculating the output of the softmax layer for node classification is:

[0127]

[0128] Among them, the vector of Elements , ;

[0129] The specific formula for calculating the loss for node classification is:

[0130]

[0131] in, is the set of indices for labeled nodes, It is a one-hot tag No. elements;

[0132] Step 4: Calculate the softmax layer output and loss of link prediction in the graph decoder according to the feature representation;

[0133] Compute the output of the softmax layer for link prediction in the graph decoder:

[0134]

[0135] in, Representing relationships The observation vector, corresponds to the relationship The parameter matrix of

[0136] Calculate the loss for link prediction:

[0137]

[0138] in, Representing relationships The observation label of and Relationship Type When ,otherwise .

[0139] Step 5: Calculate the total target loss;

[0140]

[0141] in, is the balance factor, which is a hyperparameter;

[0142] Step 6: Use the gradient descent algorithm to update the model parameters and optimize the total target loss;

[0143] Among them, the gradient descent algorithm is used to update the model parameters and (i.e. all and ), optimize the total target loss .

[0144] Step 7: Output the optimal parameter vectors of the graph encoder and the graph decoder to obtain a classification prediction network for node classification and link prediction;

[0145] Among them, the network that can be used for node classification and link prediction is obtained, including: , ,in and represents the optimal model parameters;

[0146] Step 8: Import the classification prediction network into the graph neural network model to obtain a more standard node embedding, and use the machine learning algorithm in the downstream task to classify and predict the corresponding relationship between nodes and edges; thereby obtaining the intelligent recommended treatment results of hazardous waste.

[0147] The corresponding relationship graph in the classification prediction network is imported into the graph neural network model. At the same time, this model provides three classic clustering algorithms for division, namely K-means clustering algorithm, spectral clustering algorithm, and self-organizing mapping algorithm. A more accurate embedding representation of the node is calculated, which is then used for downstream classification prediction tasks, and the final relationship graph after classification prediction is output. According to the relationship between the hazardous waste to be processed, the morphological label and the preset hazardous waste treatment process in the relationship graph, the final intelligent recommendation treatment result of hazardous waste is obtained.

[0148] S500: allocating a corresponding hazardous waste treatment process to the hazardous waste to be treated according to the intelligent recommended treatment result of the hazardous waste, and sending it to a receiving terminal of an operator, so as to perform corresponding treatment on the hazardous waste.

[0149] For example: when the hazardous waste to be processed received from the waste-generating unit is "used disposable medical supplies and disposable medical devices", at this time, the hazardous waste classification and treatment method of this solution is analyzed as medical waste-infectious waste, the form is identified as solid, and the form label is solid hazardous waste. The corresponding hazardous waste treatment process is: solid hazardous waste treatment process; after the operator receives the above information through the receiving terminal, he / she will perform specific treatment processes for solid hazardous waste such as physical volume reduction, compression and crushing on the hazardous solid waste.

[0150] Among them, the hazardous waste classification and treatment method in this embodiment can be applied to a client or a server, where the client includes, for example, a desktop computer, a mobile phone, or even an application software client. The embodiment of the present invention does not limit the specific form of the client to which the method is applied. The server may include a single server, a server cluster, or even a platform built based on a server cluster.

[0151] In the second aspect, this application proposes a hazardous waste intelligent treatment system based on knowledge graph, such as Figure 2 As shown, it includes a graph construction module, a data classification module, a morphology recognition and annotation module, an intelligent prediction and recommendation module, and a data transmission module;

[0152] The graph construction module is used to obtain the hazardous waste knowledge data involved in the target waste-producing unit and construct a hazardous waste knowledge graph;

[0153] The data classification module is used to obtain basic data of the hazardous waste to be processed, and determine the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph;

[0154] The morphological recognition and labeling module is used to obtain image data corresponding to the hazardous chemical waste to be processed, perform morphological recognition on the image data through machine vision technology to obtain a morphological recognition result of the hazardous chemical waste to be processed, and perform machine labeling on the hazardous chemical waste to be processed based on the morphological recognition result to obtain a morphological label of the hazardous chemical waste to be processed;

[0155] The intelligent prediction and recommendation module is used to construct an intelligent recommendation processing algorithm, input the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the feature vectors corresponding to the hazardous waste to be processed into the intelligent recommendation processing algorithm, and obtain the intelligent recommendation processing results of the hazardous waste;

[0156] The data transmission module is used to allocate the corresponding hazardous waste treatment process to the hazardous waste to be treated according to the intelligent recommended treatment result of the hazardous waste, and send it to the receiving terminal of the operator, so as to perform corresponding treatment on the hazardous waste.

[0157] In a third aspect, the present application proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0158] In a fourth aspect, the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0159] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0160] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0161] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0162] In the embodiments provided in the present disclosure, it should be understood that the disclosed apparatus / computer equipment and methods can be implemented in other ways. For example, the apparatus / computer equipment embodiments described above are merely schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, multiple units or components may be combined or may be integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection of the apparatus or unit, which may be electrical, mechanical or other forms.

[0163] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0164] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0165] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present disclosure implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, and the computer program code may be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electric carrier signals and telecommunication signals.

[0166] The above are only preferred implementations of the present invention. It should be pointed out that a number of modifications and improved technical solutions made by those skilled in the art without departing from the technical solution should also be deemed to fall within the scope of protection required by the claims.

Claims

1. A method for intelligent treatment of hazardous waste based on knowledge graph, characterized by: The following steps are involved: Obtain the hazardous waste knowledge data involved in the target waste-producing units and build a hazardous waste knowledge graph; Obtaining basic data of the hazardous waste to be processed, and determining the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph; Obtain image data corresponding to the hazardous waste to be processed, perform morphological recognition on the image data through machine vision technology to obtain a morphological recognition result of the hazardous waste to be processed, and perform machine labeling on the hazardous waste to be processed based on the morphological recognition result to obtain a morphological label of the hazardous waste to be processed, including: Receive image data corresponding to the hazardous waste to be processed, extract, eliminate, and color cluster the image target morphology on the image data to obtain the two-dimensional coordinates, angle, and time of the detection target on the image in the image coordinate system; Focus on tracking and displaying the detected target on the image data to determine whether the target information is consistent with The target information identified in the previous frame is the same target. If it is not the same target, the processed image and the coordinates, angles and time information of the detected target on the image are manually corrected to obtain the corrected coordinate information; The error correction coordinate information is converted into an image coordinate system to obtain the coordinates of the error target or the newly added target, and the information of the error target is removed from the coordinates, angles and time information of the detection target in each frame of the received image in the image coordinate system to obtain the coordinates, angles and time information of the identified target in the image coordinate system, and at the same time, the emphasis on the error target is removed from the received visual feedback image to complete the morphological recognition and obtain the morphological recognition result; Based on the morphological recognition result, the hazardous chemical waste to be processed is machine labeled to obtain a morphological label of the hazardous chemical waste to be processed; Constructing an intelligent recommendation processing algorithm, including constructing an input relationship graph based on the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the characteristic vector of the hazardous waste to be processed as basic elements, wherein the hazardous waste types in the hazardous waste classification results are used as nodes, the attributes of each node correspond to the characteristic vector of the hazardous waste to be processed, the morphological labels of the hazardous waste to be processed are used as labels on the nodes, and the preset hazardous waste processing process is used as the edge between the nodes; if two hazardous waste types have the same label, it is considered that the two hazardous waste types are applicable to the same hazardous waste processing process, and an edge is created for the corresponding two nodes; forming a graph encoder and a graph decoder according to the input relationship graph, and the graph encoder and the graph decoder together constitute the intelligent recommendation processing algorithm; inputting the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the characteristic vector corresponding to the hazardous waste to be processed into the intelligent recommendation processing algorithm to obtain the intelligent recommendation processing results of hazardous waste; According to the intelligent recommended treatment result of hazardous waste, a corresponding hazardous waste treatment process is allocated to the hazardous waste to be treated, and the process is sent to a receiving terminal of an operator, so that the hazardous waste is treated accordingly.

2. The method according to claim 1, characterized in that: The step of obtaining the hazardous waste knowledge data involved in the target waste-generating unit and constructing a hazardous waste knowledge graph includes: Sentence processing is performed on the hazardous waste knowledge data to establish a hazardous waste generation process table, wherein the hazardous waste generation process table includes entities and attributes; Establish an association relationship between entities and attributes, and establish a hazardous waste knowledge graph based on the association relationship.

3. The method according to claim 1, characterized in that: The obtaining of basic data of the hazardous waste to be processed and determining the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph include: Determine whether the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base; If the hazardous waste to be processed can match the hazardous waste information in the preset hazardous waste knowledge base, determine the type of hazardous waste to which the hazardous waste to be processed belongs according to the preset hazardous waste knowledge base; If the hazardous waste to be processed cannot be matched with the hazardous waste information in the preset hazardous waste knowledge base, the type of hazardous waste to which the hazardous waste to be processed belongs is determined based on the hazardous waste knowledge graph and the preset hazardous waste knowledge base.

4. The method according to claim 3, characterized in that: The intelligent recommendation processing algorithm is constructed, and the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process, and the feature vectors corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm. Before obtaining the intelligent recommendation processing results of the hazardous waste, the following is also included: Constructing a feature vector extraction model, wherein the feature vector extraction model includes an input layer, a convolution layer, a normalization layer, a pooling layer, and a fully connected layer; The output end of the input layer is connected to the input end of the convolution layer through an information transmission channel; the output end of the convolution layer is connected to the input end of the normalization layer through an information transmission channel; the output end of the normalization layer is connected to the input end of the pooling layer through an information transmission channel; the output end of the pooling layer is connected to the input end of the fully connected layer through an information transmission channel; Obtain basic data of hazardous waste corresponding to the type of hazardous waste; transmit the basic data of hazardous waste corresponding to the type of hazardous waste as input to the input layer, and finally obtain the feature vector of the hazardous waste to be processed after feature extraction through the fully connected layer.

5. The method according to claim 4, characterized in that: The hazardous waste classification results, the hazardous waste form labels to be processed, the preset hazardous waste processing process and the feature vectors corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm to obtain the hazardous waste intelligent recommendation processing results, including: Step 1: Input the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the feature vectors of the hazardous waste to be processed into the graph encoder and the graph decoder, and initialize the graph encoder, the graph decoder and the model parameters; Step 2: Calculate the feature representation corresponding to any node in the graph encoder; Step 3: Calculate the output and loss of the softmax layer for node classification in the graph encoder according to the feature representation; Step 4: Calculate the softmax layer output and loss of link prediction in the graph decoder according to the feature representation; Step 5: Calculate the total target loss; Step 6: Use the gradient descent algorithm to update the model parameters and optimize the total target loss; Step 7: Output the optimal parameter vectors of the graph encoder and the graph decoder to obtain a classification prediction network for node classification and link prediction; Step 8: Import the classification prediction network into the graph neural network model to obtain a more standard node embedding, and use the machine learning algorithm in the downstream task to classify and predict the corresponding relationship between nodes and edges; thereby obtaining the intelligent recommended treatment results of hazardous waste.

6. A hazardous waste intelligent treatment system based on knowledge graph, characterized by: It includes graph construction module, data classification module, morphology recognition and annotation module, intelligent prediction and recommendation module and data transmission module; The graph construction module is used to obtain the hazardous waste knowledge data involved in the target waste-producing unit and construct a hazardous waste knowledge graph; The data classification module is used to obtain basic data of the hazardous waste to be processed, and determine the hazardous waste classification result of the hazardous waste to be processed according to a preset hazardous waste knowledge base and a hazardous waste knowledge graph; The morphological recognition and labeling module is used to obtain image data corresponding to the hazardous waste to be processed, perform morphological recognition on the image data through machine vision technology to obtain the morphological recognition result of the hazardous waste to be processed, and perform machine labeling on the hazardous waste to be processed based on the morphological recognition result to obtain the morphological label of the hazardous waste to be processed, including: Receive image data corresponding to the hazardous waste to be processed, extract, eliminate, and color cluster the image target morphology on the image data to obtain the two-dimensional coordinates, angle, and time of the detection target on the image in the image coordinate system; Focus on tracking and displaying the detected target on the image data to determine whether the target information is consistent with The target information identified in the previous frame is the same target. If it is not the same target, the processed image and the coordinates, angles and time information of the detected target on the image are manually corrected to obtain the corrected coordinate information; The error correction coordinate information is converted into an image coordinate system to obtain the coordinates of the error target or the newly added target, and the information of the error target is removed from the coordinates, angles and time information of the detection target in each frame of the received image in the image coordinate system to obtain the coordinates, angles and time information of the identified target in the image coordinate system, and at the same time, the emphasis on the error target is removed from the received visual feedback image to complete the morphological recognition and obtain the morphological recognition result; Based on the morphological recognition result, the hazardous chemical waste to be processed is machine labeled to obtain a morphological label of the hazardous chemical waste to be processed; The intelligent prediction and recommendation module is used to construct an intelligent recommendation processing algorithm, including constructing an input relationship graph based on the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the characteristic vector of the hazardous waste to be processed as basic elements, wherein the hazardous waste types in the hazardous waste classification results are used as nodes, the attributes of each node correspond to the characteristic vector of the hazardous waste to be processed, the morphological labels of the hazardous waste to be processed are used as labels on the nodes, and the preset hazardous waste processing process is used as the edge between the nodes; if two hazardous waste types have the same label, it is considered that the two hazardous waste types are applicable to the same hazardous waste processing process, and an edge is created for the corresponding two nodes; a graph encoder and a graph decoder are formed according to the input relationship graph, and the graph encoder and the graph decoder together constitute the intelligent recommendation processing algorithm; the hazardous waste classification results, the morphological labels of the hazardous waste to be processed, the preset hazardous waste processing process and the characteristic vector corresponding to the hazardous waste to be processed are input into the intelligent recommendation processing algorithm to obtain the intelligent recommendation processing results of hazardous waste; The data transmission module is used to allocate the corresponding hazardous waste treatment process to the hazardous waste to be treated according to the intelligent recommended treatment result of the hazardous waste, and send it to the receiving terminal of the operator, so as to perform corresponding treatment on the hazardous waste.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images