Solid waste resourceization whole-process intelligent management and control method based on internet of things and big data

Abstract

The present application relates to solid waste treatment and resource technology field, especially for based on internet of things and big data's solid waste resource whole process intelligent management and control method, its method includes the following steps: S1: whole process multi-mode sensing and edge computing;S2: cloud big data convergence and digital main line construction;S3: multi-time and space scale joint prediction model group construction;S4: whole process multi-objective dynamic collaborative optimization scheduling;S5: based on digital twinning's closed loop intelligent control;S6: block chain credible evidence and dynamic credit.The present application has the advantages of comprehensive intergenerational improvement from sensing, decision-making, control to trust system, in the actual use process, source control level, whole process multi-mode sensing network and edge intelligent identification system without dead angle, change the passive supervision mode of relying on artificial sampling, realize the real-time objective supervision to the putting behavior and classification quality, and the front-end classification accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of solid waste treatment and resource utilization technology, specifically to an intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data. Background Technology

[0002] With the continuous acceleration of urbanization, the amount of solid waste generated has exploded. Resource utilization has become the core path to solve the problem of garbage surrounding cities and promote the development of a circular economy. However, upon examining the current solid waste management system, there are still many prominent bottlenecks that restrict its effectiveness. In the front-end sorting stage, the sorting effect is highly dependent on residents' environmental awareness and self-discipline. The lack of real-time, objective supervision and accurate feedback mechanisms leads to drastic fluctuations in sorting accuracy and persistent mixed disposal, which greatly interferes with subsequent resource utilization. In the collection and transportation stage, the extensive operation mode of fixed routes and fixed cycles is generally used, which cannot dynamically respond to the spatiotemporal heterogeneity of garbage generation in different time periods and areas, resulting in a high vehicle empty-run rate. In some areas, waste collection is lagging behind and waste accumulation coexists with transportation capacity waste in other areas. In the transfer and treatment process, the facilities are independent of each other and operate independently. The operating parameters are mostly set by manual experience. There is a lack of proactive perception and adaptive adjustment capabilities for fluctuations in key characteristics such as the composition of incoming materials, lower heating value, moisture content and biodegradability. This directly leads to low resource recycling efficiency, high energy consumption and secondary pollution emissions. At the data level, the data of each link of front-end delivery, mid-end collection and transportation and end-end treatment are fragmented and inconsistent, and it is impossible to form a digital main line that runs through the entire process of "generation-collection-transfer-resource-disposal". It is not only difficult to achieve accurate traceability of the source and destination of solid waste, but also cannot support the collaborative optimization and reliable supervision of the whole chain.

[0003] While current technologies and localized solutions for smart trash cans, vehicle-mounted weighing systems, and vehicle dispatch optimization systems exist, their perception dimensions are generally singular, and their decision-making logic is isolated and closed. They lack the ability to deeply and thoroughly perceive, predict intelligently in multiple time and space, dynamically schedule globally, and precisely control the entire lifecycle of solid waste from "source, collection and transportation network, treatment plant, to resource utilization." Therefore, there is an urgent need to invent a new management and control method that can deeply integrate the Internet of Things, big data, and artificial intelligence to achieve refined perception, multi-dimensional prediction, collaborative scheduling, and intelligent control of the entire solid waste process. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data. It has the advantages of comprehensive generational improvement from perception, decision-making, control to trust system, and solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent management and control of the entire process of solid waste resource utilization based on the Internet of Things and big data, the method comprising the following steps: S1: Full-process multi-mode sensing and edge computing: Construct a full-process multi-mode IoT sensing layer and edge computing framework. Deploy various types of sensors and smart terminals at solid waste generation sources, collection containers, collection vehicles, transfer stations, resource recovery facilities, and final disposal sites to collect multi-source heterogeneous data on the quantification, composition, state, and location of solid waste in real time. Perform data preprocessing, feature extraction, and real-time event recognition at the edge. S2: Cloud big data aggregation and digital master line construction: The cloud big data platform aggregates multi-source data that has undergone edge preprocessing and integrates it with external environmental and socio-economic data. After cleaning, correlation, and storage, it constructs a digital master line and knowledge graph for the entire life cycle of solid waste. S3: Construction of a multi-temporal and spatiotemporal scale joint prediction model group: Based on the aforementioned digital mainline, a joint prediction model group is constructed using a deep spatiotemporal prediction network, a physical information fusion model, and a multispectral visual analysis model to predict the amount of solid waste generated, its dynamic composition, the load of the collection and transportation network, and the operating status of the treatment facilities at multiple spatiotemporal scales; S4: Full-process multi-objective dynamic collaborative optimization scheduling: The joint prediction results are input into the full-process multi-objective dynamic collaborative optimization scheduling engine (wherein the joint prediction results are obtained by real-time collection of multi-source heterogeneous raw data from multi-mode IoT sensing devices deployed in S1 at disposal points, collection vehicles, transfer stations, and treatment facilities, and then processed through cloud big data in S2). After platform cleaning and association, it is integrated with external data to form a digital master line and knowledge graph of the entire life cycle of solid waste. The fused data of the digital master line and knowledge graph is input into the joint prediction model group of S3, thereby generating the joint prediction results required by the S4 scheduling engine, namely the future waste generation of each grid category, the composition parameters of the low calorific value and moisture content of the incoming material, and the operating status trend of the treatment facilities, which are used to drive the dynamic collaborative optimization scheduling of the entire process with multiple objectives. With economic cost, resource recovery rate, environmental emissions and system reliability as optimization objectives, it combines reinforcement learning and evolutionary algorithms to generate dynamic collection and transportation scheduling schemes and treatment facility load allocation schemes that adapt to real-time disturbances. S5: Closed-loop intelligent control based on digital twin: Establish a digital twin of the resource recovery facility, receive upstream scheduling instructions and local sensing data in real time, conduct online simulation through a hybrid mechanism and data-driven model, and use a deep reinforcement learning controller to adjust key process parameters in a closed loop to achieve adaptive optimal control of facility operation; S6: Blockchain trusted storage and dynamic credit: Introduce blockchain technology to store hashed digest data of key links in solid waste disposal, collection, treatment and resource recovery on the blockchain, realize traceability, points incentives and automatic settlement among multiple parties through smart contracts, and build a dynamic credit evaluation model for participating entities based on on-chain data.

[0006] Furthermore, as a preferred embodiment of the present invention, in step S1, multimodal sensing includes deploying an intelligent all-in-one machine integrating weighing sensors, multispectral cameras, RFID readers, and gas sensor arrays at waste disposal points; installing GPS / BeiDou dual-mode positioning, load sensors, RFID readers, and fuel consumption monitoring units on collection vehicles; installing high-temperature infrared thermal imagers, online flue gas analyzers, and near-infrared material composition analyzers in the facilities of incineration plants and anaerobic digestion plants; the edge computing framework running a lightweight instance segmentation network on the multispectral images in real time to identify mixed waste and evaluate classification quality; and performing anomaly detection on the sensor time-series stream and uploading event, attribute, and spatiotemporal tagged data packets.

[0007] Furthermore, as a preferred embodiment of the present invention, in step S2, the method for constructing the digital main thread and knowledge graph is to use the unique identifier of solid waste as a clue to associate the entire life cycle nodes of the disposal event, collection trajectory, transfer compression, and processing batches into a time-series data chain; at the same time, external information such as policies and regulations, facility processing capacity, historical cases, weather, and holidays is constructed into a knowledge graph, and through entity alignment and relation extraction, it is dynamically integrated with the data chain to provide a reasonable context for prediction and scheduling.

[0008] Furthermore, as a preferred embodiment of the present invention, in step S3, the joint prediction model group includes three types of models: a spatiotemporal prediction model for solid waste generation, which uses urban grids as units, constructs a spatiotemporal map using POI distribution, population heat map, and historical generation sequence, and integrates graph convolutional networks and Transformer architecture to output the generation of each grid category for the next 24 hours to 7 days; a real-time inference model for solid waste composition, which integrates near-infrared spectroscopy and machine vision data from transfer stations and treatment plant inlets, extracts component features using a deep convolutional autoencoder, and outputs key parameters such as low calorific value, proportion of biodegradable organic matter, and moisture content in real time; and a treatment facility status prediction model, which is based on physical information neural networks, constructs prediction models for incinerator furnace temperature field, pollutant generation rate, and anaerobic digestion gas production rate, providing a basis for subsequent process control.

[0009] Furthermore, as a preferred embodiment of the present invention, in step S4, the full-process multi-objective dynamic collaborative optimization scheduling engine adopts a two-layer optimization architecture: the upper layer is a multi-objective evolutionary algorithm, which aims to minimize the total economic cost of the entire system, maximize the resource utilization rate, and minimize carbon emissions, and solves the Pareto front solution set for the collection vehicle routes, vehicle allocation, and processing plant task scheduling for a future time period; the lower layer is a dynamic scheduler based on multi-agent deep reinforcement learning, where each collection vehicle and processing plant is regarded as an agent, and online rescheduling is performed near the Pareto solution space based on real-time traffic, faults, and sudden large-item waste disturbances to ensure the robustness of the solution.

[0010] Furthermore, as a preferred embodiment of the present invention, in step S5, the digital twin and the deep reinforcement learning controller specifically involve: establishing a hybrid twin model for the incinerator that integrates a computational fluid dynamics reduced-order model and a long short-term memory network residual correction, and extrapolating the furnace operating conditions at a speed of seconds; the reinforcement learning controller uses furnace temperature, steam flow rate, and flue gas oxygen content as states, and grate speed, primary air volume, and secondary air volume as actions, and designs a reward function that achieves stable steam production, low pollutant emissions, and minimum slag heat loss rate, and realizes autonomous optimization control of the combustion process through online training and deployment of a near-end strategy optimization algorithm.

[0011] Furthermore, as a preferred embodiment of the present invention, in step S6, the blockchain adopts a consortium blockchain architecture, and the smart contracts include a classification quality points contract, a collection service settlement contract, and a processing subsidy contract. The dynamic credit evaluation model calculates the credit scores of each residential community, collection and transportation company, and processing plant based on the indicators of mixed collection rate, collection on-time rate, and processing compliance rate recorded on the chain, using a Bayesian reputation model, and adjusts the incentive coefficient and supervision frequency accordingly.

[0012] Furthermore, as a preferred embodiment of the present invention, the intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data provides a system comprising a perception and execution layer, a network transmission layer, a cloud platform layer, and an application service layer. The cloud platform layer is equipped with a data middleware, a model engine, a collaborative scheduling engine, a digital twin engine, and a blockchain service for executing the above method.

[0013] Beneficial effects: The technical solution of this application has the following advantages: This invention has the advantage of a comprehensive generational improvement from perception, decision-making, control to trust system. In actual use, at the source control level, the multi-modal perception network and edge intelligent identification system without blind spots in the entire process change the passive supervision mode that relies on manual spot checks, realize real-time objective supervision of disposal behavior and classification quality, and improve the accuracy of front-end classification. At the transportation scheduling level, the deep integration of multi-temporal and spatial generation prediction and dual-layer collaborative scheduling engine enables the collection and transportation system to dynamically adapt to the spatiotemporal heterogeneity of waste, reduce collection and transportation costs, and improve the timeliness of collection. Furthermore, by intelligently allocating the load of processing facilities, peak shaving and valley filling are achieved. At the facility operation level, real-time prediction of incoming material composition and high-fidelity digital twins enable incinerators and anaerobic digesters to have feedforward adjustment capabilities. Combined with deep reinforcement learning closed-loop control, incineration power generation efficiency is improved, anaerobic digestion gas production rate is increased, and fluctuations in secondary pollution emissions are significantly narrowed. At the trusted management level, the integration of blockchain and knowledge graphs constructs an immutable traceability chain for the entire solid waste chain. Dynamic credit evaluation provides a trusted data foundation for points rewards, automatic fee settlement, and carbon trading, promoting the transformation of solid waste management towards a refined, market-oriented, and trusted governance paradigm.

[0014] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below can be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. To better understand the technical content of the present invention, specific embodiments are provided and described in conjunction with the accompanying drawings. Various aspects of the present invention are described in this disclosure with reference to the accompanying drawings, which illustrate numerous illustrative embodiments. It should be understood that the various concepts and embodiments described above, as well as those described in more detail below, can be implemented in any of many ways. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0017] As attached Figure 1 As shown: This embodiment provides a method for intelligent management and control of the entire process of solid waste resource utilization based on the Internet of Things and big data. The method includes the following steps: S1: Full-process multi-mode sensing and edge computing: Construct a full-process multi-mode IoT sensing layer and edge computing framework, deploy multiple types of sensors and smart terminals at solid waste generation sources, collection containers, collection and transportation vehicles, transfer stations, resource utilization facilities and final disposal sites, collect multi-source heterogeneous data on the quantification, composition, state and location of solid waste in real time, and perform data preprocessing, feature extraction and real-time event recognition at the edge. S2: Cloud Big Data Convergence and Digital Mainline Construction: The cloud big data platform gathers multi-source data that has been preprocessed at the edge, integrates it with external environmental and socio-economic data, cleans, correlates and stores it, and constructs a digital mainline and knowledge graph for the entire life cycle of solid waste. S3: Construction of a multi-temporal and spatial scale joint prediction model group: Based on the digital main line, a joint prediction model group is constructed by using a deep spatiotemporal prediction network, a physical information fusion model and a multispectral visual analysis model to predict the amount of solid waste generated, the dynamic composition of solid waste, the load of the collection and transportation network and the operating status of the treatment facilities at multiple spatiotemporal scales. S4: Full-process multi-objective dynamic collaborative optimization scheduling: The joint prediction results are input into the full-process multi-objective dynamic collaborative optimization scheduling engine (where the joint prediction results are obtained by multi-mode IoT sensing devices deployed in S1 at collection points, collection vehicles, transfer stations and treatment facilities, which collect multi-source heterogeneous raw data in real time, and then cleaned and correlated by the cloud big data platform in S2, and then fused with external data to form a digital master line and knowledge graph of the entire life cycle of solid waste. The fused data of the digital master line and knowledge graph is input into the joint prediction model group in S3, thereby generating the joint prediction results required by the S4 scheduling engine, namely the future waste generation of each grid category, the composition parameters of the low calorific value and moisture content of the incoming material, and the operating status trend of the treatment facilities, which are used to drive the full-process multi-objective dynamic collaborative optimization scheduling). With economic cost, resource recovery rate, environmental emissions and system reliability as optimization objectives, reinforcement learning and evolutionary algorithms are combined to generate dynamic collection and transportation scheduling schemes and treatment facility load allocation schemes that adapt to real-time disturbances. S5: Closed-loop intelligent control based on digital twin: Establish a digital twin of the resource processing facility, receive upstream scheduling instructions and local sensing data in real time, conduct online simulation through mechanism and data hybrid driving model, and use deep reinforcement learning controller to adjust key process parameters in a closed loop to achieve adaptive optimal control of facility operation. S6: Blockchain Trusted Evidence Storage and Dynamic Credit: Introducing blockchain technology, hash-digested data from key stages of solid waste disposal, collection, treatment, and resource recovery are stored on the blockchain. Through smart contracts, traceability, points incentives, and automatic settlement among multiple parties are achieved. A dynamic credit evaluation model for participating entities is built based on the on-chain data.

[0018] Specifically, in step S1, multi-mode sensing includes deploying an intelligent all-in-one machine at the waste disposal point that integrates weighing sensors, multispectral cameras, RFID readers, and gas sensor arrays; installing GPS / BeiDou dual-mode positioning, load sensors, RFID readers, and fuel consumption monitoring units on collection vehicles; installing high-temperature infrared thermal imagers, online flue gas analyzers, and near-infrared material composition analyzers in the facilities of incineration plants and anaerobic digesters; using an edge computing framework to run a lightweight instance segmentation network on the multispectral images in real time to identify mixed materials and assess classification quality; and performing anomaly detection on the sensor time-series stream and uploading event, attribute, and spatiotemporal tagged data packets.

[0019] Specifically, in step S2, the method for constructing the digital main thread and knowledge graph is to use the unique identifier of solid waste as a clue to associate the entire life cycle nodes of the disposal event, collection trajectory, transfer compression, and processing batches into a time-series data chain; at the same time, external information such as policies and regulations, facility processing capacity, historical cases, weather, and holidays are constructed into a knowledge graph, and through entity alignment and relationship extraction, it is dynamically integrated with the data chain to provide a reasonable context for prediction and scheduling.

[0020] Specifically, in step S3, the joint prediction model group includes three types of models: a spatiotemporal prediction model for solid waste generation, which uses urban grids as units and constructs a spatiotemporal map using POI distribution, population heat map, and historical generation sequences, and integrates graph convolutional networks and Transformer architecture to output the generation of each category in each grid for the next 24 hours to 7 days; a real-time inference model for solid waste composition, which integrates near-infrared spectroscopy and machine vision data from transfer stations and treatment plant inlets, and uses a deep convolutional autoencoder to extract composition features and output key parameters such as low calorific value, proportion of biodegradable organic matter, and moisture content in real time; and a treatment facility status prediction model, which is based on physical information neural networks and constructs prediction models for incinerator temperature field, pollutant generation rate, and anaerobic digestion gas production rate to provide a basis for subsequent process control.

[0021] Specifically, in step S4, the full-process multi-objective dynamic collaborative optimization scheduling engine adopts a two-layer optimization architecture: the upper layer is a multi-objective evolutionary algorithm, which aims to minimize the total economic cost of the entire system, maximize the resource utilization rate, and minimize carbon emissions, and solves the Pareto front solution set for the collection vehicle routes, vehicle allocation, and processing plant task scheduling for a future time period; the lower layer is a dynamic scheduler based on multi-agent deep reinforcement learning, where each collection vehicle and processing plant is regarded as an agent, and online rescheduling is performed near the Pareto solution space based on real-time traffic, faults, and sudden large-item waste disturbances to ensure the robustness of the solution.

[0022] Specifically, in step S5, the digital twin and deep reinforcement learning controller are as follows: a hybrid twin model of the incinerator is established, which integrates a computational fluid dynamics reduced-order model and a long short-term memory network residual correction, to simulate the furnace working conditions at a speed of seconds; the reinforcement learning controller takes the furnace temperature, steam flow rate, and flue gas oxygen content as states, and the grate speed, primary air volume, and secondary air volume as actions, and designs a reward function that achieves stable steam production, low pollutant emissions, and minimum slag heat loss rate. Through online training and deployment of the near-end strategy optimization algorithm, autonomous optimization control of the combustion process is realized.

[0023] Specifically, in step S6, the blockchain adopts a consortium blockchain architecture, and the smart contracts include classification quality points contracts, collection service settlement contracts, and processing subsidy contracts. The dynamic credit evaluation model calculates the credit scores of each residential community, collection and transportation company, and processing plant based on the indicators of mixed collection rate, collection on-time rate, and processing compliance rate recorded on the chain, and adjusts the incentive coefficients and supervision frequency accordingly.

[0024] Specifically, the method for intelligent management and control of the entire process of solid waste resource utilization based on the Internet of Things and big data provides a system that includes a perception and execution layer, a network transmission layer, a cloud platform layer, and an application service layer. The cloud platform layer is equipped with a data middleware, a model engine, a collaborative scheduling engine, a digital twin engine, and blockchain services to execute the above method.

[0025] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0026] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for intelligent management and control of the entire process of solid waste resource utilization based on the Internet of Things and big data, characterized by: The method includes the following steps: S1: Full-process multi-modal sensing and edge computing: Construct a full-process multi-modal IoT sensing layer and edge computing framework. Deploy various types of sensors and smart terminals at solid waste generation sources, collection containers, collection vehicles, transfer stations, resource recovery facilities, and final disposal sites to collect multi-source heterogeneous data on the quantification, composition, state, and location of solid waste in real time. Perform data preprocessing, feature extraction, and real-time event recognition at the edge. S2: Cloud big data aggregation and digital masterline construction: The cloud big data platform aggregates multi-source data that has undergone edge preprocessing. It integrates this data with external environmental and socio-economic data, performs cleaning, correlation, and storage, and constructs a digital masterline and knowledge graph for the entire life cycle of solid waste. S3: Construction of a multi-temporal and spatial scale joint prediction model group: Based on the aforementioned digital masterline, use a deep spatiotemporal prediction network, a physical information fusion model, and a multispectral visual analysis model to construct a joint prediction model group for solid waste generation, composition dynamics, collection network load, and treatment facility operation status at multiple spatiotemporal scales. S4: Full-process multi-objective dynamic collaborative optimization scheduling: The joint prediction results are input into the full-process multi-objective dynamic collaborative optimization scheduling engine. With economic cost, resource recovery rate, environmental emissions and system reliability as optimization objectives, reinforcement learning and evolutionary algorithms are combined to generate dynamic collection and transportation scheduling schemes and processing facility load allocation schemes that adapt to real-time disturbances; S5: Closed-loop intelligent control based on digital twin: A digital twin of the resource-based processing facility is established to receive upstream scheduling instructions and local sensing data in real time. Online simulation is performed through a hybrid mechanism and data-driven model, and key process parameters are adjusted in a closed loop using a deep reinforcement learning controller to achieve adaptive optimal control of facility operation; S6: Blockchain Trusted Evidence Storage and Dynamic Credit: Introducing blockchain technology, hash-digested data from key stages of solid waste disposal, collection, treatment, and resource recovery are stored on the blockchain. Through smart contracts, traceability, points incentives, and automatic settlement among multiple parties are achieved. A dynamic credit evaluation model for participating entities is built based on the on-chain data.

2. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claim 1, characterized in that: In step S1, multi-mode sensing includes deploying an intelligent all-in-one machine at the waste disposal point that integrates weighing sensors, multispectral cameras, RFID readers, and gas sensor arrays; installing GPS / BeiDou dual-mode positioning, load sensors, RFID readers, and fuel consumption monitoring units on collection vehicles; installing high-temperature infrared thermal imagers, online flue gas analyzers, and near-infrared material composition analyzers in the facilities of incineration plants and anaerobic digestion plants; the edge computing framework runs a lightweight instance segmentation network on the multispectral images in real time to identify mixed materials and evaluate classification quality; and performing anomaly detection on the sensor time-series stream, uploading event, attribute, and spatiotemporal tagged data packets.

3. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claim 1, characterized in that: In step S2, the method for constructing the digital mainline and knowledge graph is to use the unique identifier of solid waste as a clue to associate the entire life cycle nodes of the disposal event, collection trajectory, transfer compression, and processing batches into a time-series data chain; at the same time, external information such as policies and regulations, facility processing capacity, historical cases, weather, and holidays are constructed into a knowledge graph, and through entity alignment and relationship extraction, it is dynamically integrated with the data chain to provide a reasonable context for prediction and scheduling.

4. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claim 1, characterized in that: In step S3, the joint prediction model group includes three types of models: a spatiotemporal prediction model for solid waste generation, which uses urban grids as units and constructs a spatiotemporal map using POI distribution, population heat map, and historical generation sequence, and integrates graph convolutional networks and Transformer architecture to output the generation of each category in each grid for the next 24 hours to 7 days; a real-time inference model for solid waste composition, which integrates near-infrared spectroscopy and machine vision data from transfer stations and treatment plant inlets, and uses a deep convolutional autoencoder to extract component features and output key parameters such as low calorific value, proportion of biodegradable organic matter, and moisture content in real time; and a treatment facility status prediction model, which is based on physical information neural networks and constructs prediction models for incinerator temperature field, pollutant generation rate, and anaerobic digestion gas production rate to provide a basis for subsequent process control.

5. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claim 1, characterized in that: In step S4, the full-process multi-objective dynamic collaborative optimization scheduling engine adopts a two-layer optimization architecture: the upper layer is a multi-objective evolutionary algorithm, which aims to minimize the total economic cost of the whole system, maximize the resource output rate, and minimize carbon emissions, and solves the Pareto front solution set of the collection vehicle route, vehicle allocation, and processing plant task scheduling for a future time period. The lower layer is a dynamic scheduler based on multi-agent deep reinforcement learning. Each collection vehicle and processing plant is regarded as an agent. Based on real-time traffic, faults, and disturbances caused by sudden large waste items, online rescheduling is carried out near the Pareto solution space to ensure the robustness of the solution.

6. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claim 1, characterized in that: In step S5, the digital twin and deep reinforcement learning controller specifically involve: establishing a hybrid twin model for the incinerator that integrates a computational fluid dynamics reduced-order model and a long short-term memory network residual correction, and deriving the incinerator's internal operating conditions at a speed of seconds. The reinforcement learning controller uses furnace temperature, steam flow rate, and flue gas oxygen content as states, and grate speed, primary air volume, and secondary air volume as actions. The reward function is designed to achieve stable steam production, low pollutant emissions, and minimum slag heat loss rate. Through online training and deployment of a near-end strategy optimization algorithm, it realizes autonomous optimization control of the combustion process.

7. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claim 1, characterized in that: In step S6, the blockchain adopts a consortium blockchain architecture, and the smart contracts include classification quality points contracts, collection service settlement contracts, and processing subsidy contracts. The dynamic credit evaluation model calculates the credit scores of each residential community, collection and transportation company, and processing plant based on the mixed collection rate, collection on-time rate, and processing compliance rate recorded on the chain, and adjusts the incentive coefficients and supervision frequency accordingly.

8. The intelligent management and control method for the entire process of solid waste resource utilization based on the Internet of Things and big data as described in claims 1-7, characterized in that: The proposed method for intelligent management and control of the entire process of solid waste resource utilization based on the Internet of Things and big data provides a system comprising a perception and execution layer, a network transmission layer, a cloud platform layer, and an application service layer. The cloud platform layer is equipped with a data middleware, a model engine, a collaborative scheduling engine, a digital twin engine, and blockchain services to execute the aforementioned method.

Images