A recycling method and system based on a cloud platform for renewable resources

Abstract

This invention discloses a recycling method and system based on a cloud platform for renewable resources, comprising: acquiring image data based on historical data of renewable resources from the cloud platform; inputting the image data into a lightweight convolutional neural network for training a resource type identification model; acquiring image data and infrared spectral data of waste resources through multi-source sensors, inputting them into the resource type identification model to identify recyclable resources in the waste; classifying resources based on the identification results, storing the waste in separate containers through a compression and sorting device, and adding basic information tags; and dynamically planning the optimal recycling path and assigning tasks to recycling vehicles by monitoring data in real time through the cloud platform. The advantages of this invention are: by monitoring and optimizing the recycling path in real time through the cloud platform, accurate waste classification and efficient recycling are achieved, improving the intelligence and automation of resource recycling.

Description

Technical Field

[0001] This invention relates to resource recycling technology, and in particular to a recycling method and system based on a cloud platform for renewable resources. Background Technology

[0002] Recycling technology refers to the process of transforming waste into reusable resources through a series of physical, chemical, and biological methods. With the increasing severity of global resource shortages and environmental pollution, the sustainable use of resources has become increasingly important. Recycling not only effectively reduces waste accumulation and environmental pollution but also conserves natural resources and promotes the development of a circular economy.

[0003] Current recycling methods typically rely on manual labor or simple mechanical equipment for resource sorting, resulting in low efficiency and a high risk of missorting. Traditional methods lack intelligent identification technology, failing to accurately distinguish between different types of waste, leading to inaccurate sorting and impacting resource recycling rates. Traditional recycling systems often lack real-time data monitoring and dynamic optimization capabilities, unable to adjust recycling routes and task allocation based on actual conditions, frequently resulting in overflowing containers at recycling points or inefficient vehicle routes, causing resource waste and low operational efficiency. Furthermore, the lack of intelligent tag management and information tracking makes effective management of resource traceability and recycling processes difficult. In contrast, cloud-based recycling methods, through intelligent identification, real-time monitoring, and route optimization technologies, not only improve recycling efficiency but also enable accurate sorting and scientific scheduling, significantly enhancing the quality and efficiency of resource recycling. Summary of the Invention

[0004] To improve existing recycling methods and systems, this paper proposes a recycling method and system based on a cloud platform for renewable resources. This method achieves accurate waste classification and efficient recycling by combining lightweight convolutional neural networks and multi-source sensor data. The cloud platform monitors and optimizes recycling paths in real time, thereby improving the intelligence, automation, and operational efficiency of resource recycling.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A recycling method based on a cloud platform for renewable resources includes:

[0007] Based on the historical data of recycled resources in the cloud platform, they are divided into metal, plastic and paper categories, and image data of each category of recycled resources are obtained.

[0008] The acquired historical image data of recycled resources is input into a lightweight convolutional neural network to train a resource type identification model, and the trained resource type identification model is obtained.

[0009] By acquiring image data and infrared spectral data of waste resources through multi-source sensors, the pre-processed data is input into the resource type identification model to identify and obtain recyclable metal, plastic and paper resources in the waste resources.

[0010] Based on the identification results, resources are classified, and waste is sorted and stored in separate containers using a compression and sorting device, with basic information labels added.

[0011] By monitoring the container storage capacity of each recycling point in real time through a cloud platform, and combining geographical location data and traffic conditions, the optimal recycling route is dynamically planned and tasks are assigned to recycling vehicles.

[0012] Preferably, the step of inputting the acquired historical recycled resource image data into a lightweight convolutional neural network for training a resource type identification model, and obtaining the trained resource type identification model specifically includes:

[0013] Based on the acquired historical image data of recycled resources, the data is labeled and divided into metal, plastic and paper categories, and stored in the corresponding folders respectively;

[0014] Based on the number of samples in each category, data balance was achieved through data augmentation and supplementary collection, and the data was divided into training set, validation set, and test set according to the proportions.

[0015] The training set data is input into a lightweight convolutional neural network, and the model is trained by optimizing it through transfer learning.

[0016] The system records the loss and accuracy of the training and validation sets in real time, adjusts the hyperparameters under conditions of underfitting and overfitting, and completes the training.

[0017] The accuracy of the model is calculated using the reserved test set, a confusion matrix is ​​generated to analyze the identification errors of each category, and targeted optimizations are made for misclassified samples of metal, plastic, and paper.

[0018] Export the trained resource type recognition model.

[0019] Preferably, the step of acquiring waste resource image data and infrared spectral data through multi-source sensors, and inputting the preprocessed data into the resource type identification model to identify and acquire recyclable metal, plastic, and paper resources in the waste specifically includes:

[0020] The surface texture, shape, and color image data of waste are acquired using a visible light camera, and the vibrational characteristic data of the molecules of the material are acquired using a near-infrared spectrometer.

[0021] Based on the acquired multi-source data, the timestamps of the image and spectral data are aligned, and the multi-source data of the same waste sample are stored together.

[0022] Image data preprocessing and infrared spectral data preprocessing are performed based on the acquired multi-source data to obtain waste characteristic data;

[0023] Waste characteristic data is input into the resource type identification model, the probability distribution of waste classification results is output, and the final classification result is determined.

[0024] Preferably, the step of classifying resources based on the identification results, storing waste in separate containers using a compression and sorting device, and adding basic information tags specifically includes:

[0025] Based on the model recognition results, priority is defined according to the recycling value, and sorting instructions are generated for waste in priority order.

[0026] Based on the sorted and classified waste, a differentiated compression strategy is adopted according to the type to compress and package various types of waste.

[0027] Each compressed and packaged waste item is labeled with basic information, and the label information is uploaded to the cloud platform system in real time. The information labels include static information: category, weight, and compression time, and dynamic information: traceability code.

[0028] Based on the type of compression and packaging, waste is stored in the corresponding storage container, and the remaining capacity of the container is monitored in real time.

[0029] When storing waste, the waste category is verified a second time by photoelectric sensors, and missorted items are rejected for storage.

[0030] Preferably, the step of monitoring the storage capacity of containers at each recycling point in real time through a cloud platform, combining geographical location data and traffic conditions, dynamically planning the optimal recycling route, and allocating tasks to recycling vehicles specifically includes:

[0031] A storage capacity change function is constructed based on the changes in container storage capacity data at each recycling point, and the storage capacity change data for the next hour is obtained based on the trend of the function.

[0032] The city roads are abstracted as a directed graph, with road intersections and recycling points as nodes, the direction of traffic on the road segment as edges, and the weights being dynamically calculated travel time costs.

[0033] Based on Dijkstra's algorithm, the shortest time-consuming path from the vehicle's current location to that point is calculated independently for each recycling point;

[0034] Weights are added based on the container storage capacity of each recycling point, and multiple recycling point paths are merged according to the nearest neighbor insertion method;

[0035] In the event of an anomaly, freeze the current task queue and rerun the Dijkstra algorithm to generate a new optimized path.

[0036] Furthermore, a recycling system based on a cloud platform for renewable resources is proposed, including:

[0037] Data acquisition module: The data acquisition module is mainly used to collect historical recycled resource data and acquire waste resource image data and infrared spectral data from multi-source sensors;

[0038] Resource identification module: The resource identification module is mainly used to train a resource type identification model based on a lightweight convolutional neural network, and to identify and classify recyclable resources such as metals, plastics, and paper.

[0039] Sorting and storage module: The sorting and storage module is mainly used to classify and store waste into independent containers according to the identification results through a compression sorting device, and add basic information labels;

[0040] Route planning module: The route planning module is mainly used to monitor the storage capacity of containers at each recycling point in real time, and dynamically plan the optimal recycling route by combining geographical location and traffic data;

[0041] Cloud platform monitoring module: The cloud platform monitoring module is mainly used to monitor the container capacity and recycling vehicle status of the recycling point in real time through the cloud platform, and to optimize resource recycling paths and task allocation;

[0042] Processor: The processor is mainly used for the calculation process of each formula and the construction calculation process of each model.

[0043] Compared with the prior art, the advantages of the present invention are:

[0044] By employing lightweight convolutional neural networks for resource type identification and combining multi-source sensor data, the system can accurately identify and classify recyclable resources such as metals, plastics, and paper in waste, achieving efficient resource sorting and recycling. The system significantly improves the operational efficiency of recycling vehicles and reduces unnecessary time waste and energy consumption by real-time monitoring of container storage capacity at recycling points and incorporating traffic data for optimal route planning. Furthermore, after resource sorting, resources are packaged using a compression and sorting device and labeled with basic information tags to ensure the efficiency and traceability of the recycling process. Intelligent scheduling and task allocation through a cloud platform system automate and optimize the entire recycling process, further reducing labor costs and errors. This method can dynamically respond to the needs of different recycling points, adjusting recycling plans in a timely manner to adapt to actual changes, thus improving the system's flexibility and adaptability. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the method proposed in this invention;

[0046] Figure 2 This is a schematic diagram of the model training proposed in this invention;

[0047] Figure 3 This is a schematic diagram of waste identification proposed in this invention;

[0048] Figure 4 This is a schematic diagram of the resource classification proposed in this invention;

[0049] Figure 5 This is a schematic diagram of the task allocation proposed in this invention;

[0050] Figure 6 This is an architecture diagram of the electronic devices in this solution;

[0051] Figure 7 This is a schematic diagram of the computer-readable storage medium structure in this scheme. Detailed Implementation

[0052] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0053] A recycling system based on a cloud platform for renewable resources includes:

[0054] Data acquisition module: The data acquisition module is mainly used to collect historical recycled resource data and acquire waste resource image data and infrared spectral data from multi-source sensors;

[0055] Resource identification module: The resource identification module is mainly used to train a resource type identification model based on a lightweight convolutional neural network, and to identify and classify recyclable resources such as metals, plastics, and paper.

[0056] Sorting and storage module: The sorting and storage module is mainly used to classify and store waste into independent containers according to the identification results through a compression sorting device, and add basic information labels;

[0057] Route planning module: The route planning module is mainly used to monitor the storage capacity of containers at each recycling point in real time, and dynamically plan the optimal recycling route by combining geographical location and traffic data;

[0058] Cloud platform monitoring module: The cloud platform monitoring module is mainly used to monitor the container capacity and recycling vehicle status of the recycling point in real time through the cloud platform, and to optimize resource recycling paths and task allocation;

[0059] Processor: The processor is mainly used for the calculation process of each formula and the construction calculation process of each model.

[0060] See Figure 1As shown, a recycling method based on a renewable resource cloud platform includes:

[0061] Step 1: Based on the historical data of recycled resources in the cloud platform, classify them into metal, plastic and paper categories, and obtain image data of each type of recycled resource;

[0062] Step 2: Input the acquired historical recycled resource image data into a lightweight convolutional neural network to train the resource type identification model and obtain the trained resource type identification model;

[0063] Step 3: Acquire image data and infrared spectral data of waste resources through multi-source sensors, and input the preprocessed data into the resource type identification model to identify and obtain recyclable metal, plastic and paper resources in the waste resources;

[0064] Step 4: Based on the identification results, classify the resources, and use a compression and sorting device to store the waste in separate containers and add basic information labels;

[0065] Step 5: Monitor the container storage capacity of each recycling point in real time through the cloud platform, and dynamically plan the optimal recycling route by combining geographical location data and traffic conditions, and assign tasks to recycling vehicles.

[0066] See Figure 2 As shown, the acquired historical recycled resource image data is input into a lightweight convolutional neural network for training a resource type identification model. The specifically obtained trained resource type identification model includes:

[0067] Based on the acquired historical image data of recycled resources, the data is labeled and divided into metal, plastic and paper categories, and stored in the corresponding folders respectively;

[0068] Based on the number of samples in each category, data balance was achieved through data augmentation and supplementary collection, and the data was divided into training set, validation set, and test set according to the proportions.

[0069] The training set data is input into a lightweight convolutional neural network, and the model is trained by optimizing it through transfer learning.

[0070] The system records the loss and accuracy of the training and validation sets in real time, adjusts the hyperparameters under conditions of underfitting and overfitting, and completes the training.

[0071] The accuracy of the model is calculated using the reserved test set, a confusion matrix is ​​generated to analyze the identification errors of each category, and targeted optimizations are made for misclassified samples of metal, plastic, and paper.

[0072] Export the trained resource type recognition model.

[0073] Specifically, the acquired historical recycled resource image data is labeled and the data is classified and stored according to different resource types. The labeling categories include metals: such as aluminum cans, iron cans, etc.; plastics: such as plastic bottles, plastic bags, etc.; and paper: such as newspapers, cardboard boxes, etc. Each type of image is stored in a separate folder for metals, plastics, and paper.

[0074] Image data is divided into training, validation, and test sets, typically in a ratio of 70% training, 15% validation, and 15% test.

[0075] The pre-trained model is used to load pre-trained weights through transfer learning, and the last layer is replaced with an output layer adapted to the current task, including three categories: metal, plastic, and paper.

[0076] During training, the model is optimized by adjusting hyperparameters such as learning rate, batch size, and training epochs to prevent overfitting or underfitting. Overfitting can be addressed by early stopping, regularization (such as L2), etc., while underfitting can be optimized by increasing model complexity or increasing the amount of training data.

[0077] The final model is evaluated using a test set, and a confusion matrix is ​​generated to analyze the error patterns in each category. The confusion matrix is ​​as follows:

[0078]

[0079] Where CM is the confusion matrix, TP is the true positive instance, FN is the false negative instance, and FP is the false positive instance;

[0080] Analyze misclassified samples of metals, plastics, and paper, and increase the sample size or adjust the model parameters to ensure the model is more accurate.

[0081] See Figure 3 As shown, image data and infrared spectral data of waste resources are acquired through multi-source sensors. The preprocessed data is then input into a resource type identification model to identify recyclable metals, plastics, and paper resources within the waste. Specifically, these include:

[0082] The surface texture, shape, and color image data of waste are acquired using a visible light camera, and the vibrational characteristic data of the molecules of the material are acquired using a near-infrared spectrometer.

[0083] Based on the acquired multi-source data, the timestamps of the image and spectral data are aligned, and the multi-source data of the same waste sample are stored together.

[0084] Image data preprocessing and infrared spectral data preprocessing are performed based on the acquired multi-source data to obtain waste characteristic data;

[0085] The waste characteristic data is input into the resource type identification model, which outputs the probability distribution of waste classification results and determines the final classification result.

[0086] Specifically, image data is obtained by using a visible light camera to capture the surface texture, shape, and color of the waste, while spectral data is obtained by using a near-infrared spectrometer to capture the molecular vibrational characteristics of the waste. The spectral data of each data point contains absorption values ​​at different wavelengths, reflecting the chemical composition of the substance.

[0087] Each image and spectral data is accompanied by a timestamp, and data with the same timestamp are associated. The image and spectral data of each waste sample are stored in an associated data structure.

[0088] The texture, shape, and color features of waste are obtained through image feature extraction algorithms, and features of the near-infrared spectrum, such as principal components and the wavelength of the maximum absorption peak, are extracted.

[0089] The feature data is input into the resource type identification model, and the probability distribution of each category is output through the Softmax activation function of the output layer. The formula is as follows:

[0090]

[0091] Where F represents the feature data, P(y = i | F) is the probability that a resource is identified as the i-th class given the feature data F, and z i These are the raw output values ​​of the last fully connected layer of the model, without normalization. The sum of all category index values ​​is used to normalize the denominator, ensuring that the sum of probabilities equals 1.

[0092] For each input sample, the model outputs a probability distribution representing the probability that the waste belongs to metal, plastic, or paper, and selects the class with the highest probability as the final classification result.

[0093] See Figure 4 As shown, based on the identification results, resources are classified, and waste is sorted and stored in individual containers using a compression and sorting device, with basic information labels added, specifically including:

[0094] Based on the model recognition results, priority is defined according to the recycling value, and sorting instructions are generated for waste in priority order.

[0095] Based on the sorted and classified waste, a differentiated compression strategy is adopted according to the type to compress and package various types of waste.

[0096] Each compressed and packaged waste item is labeled with basic information, and the label information is uploaded to the cloud platform system in real time. The information labels include static information: category, weight, and compression time, and dynamic information: traceability code.

[0097] Based on the type of compression and packaging, waste is stored in the corresponding storage container, and the remaining capacity of the container is monitored in real time.

[0098] When storing waste, the waste category is verified a second time by photoelectric sensors, and missorted items are rejected for storage.

[0099] Specifically, each category is assigned a priority based on its recycling value. For example, metals have the highest priority, followed by plastics, and paper has the lowest priority, generating sorting instructions: for example, for metal waste, the sorting instruction is "prioritize sorting and compressing";

[0100] Differentiated compression based on the type and recycling value of waste:

[0101] Metals: Due to their high recycling value, metals are typically compressed more tightly to reduce space usage;

[0102] Plastics: Plastics have relatively low recycling value, so moderate compression can be chosen to save storage space;

[0103] Paper: Paper waste can be moderately compressed to avoid excessive compression and waste;

[0104] Add static and dynamic information tags to each compressed and packaged waste item:

[0105] Static information includes: category: such as metal, plastic, paper; weight: obtained through a weighing sensor; compression time: timestamps recording the packaging process;

[0106] Dynamic information includes: traceability coding: using a unique traceability code to track the source and treatment process of each batch of waste;

[0107] This information is uploaded to the cloud platform in real time for easy management and retrieval later.

[0108] During waste storage, each piece of waste is verified twice by photoelectric sensors to confirm whether its category is correct. The photoelectric sensors analyze the shape, color and other characteristics of the waste and match them with the predefined category. If the sensor detects that the category of the waste does not match the expected category, it refuses to store the item and issues an alarm.

[0109] See Figure 5As shown, the cloud platform monitors the container storage capacity of each recycling point in real time, and dynamically plans the optimal recycling route based on geographical location data and traffic conditions. The task allocation for recycling vehicles specifically includes:

[0110] A storage capacity change function is constructed based on the changes in container storage capacity data at each recycling point, and the storage capacity change data for the next hour is obtained based on the trend of the function.

[0111] The city roads are abstracted as a directed graph, with road intersections and recycling points as nodes, the traffic direction of road segments as edges, and the weights as dynamically calculated travel time costs.

[0112] Based on Dijkstra's algorithm, the shortest time-consuming path from the vehicle's current location to that point is calculated independently for each recycling point;

[0113] Weights are added based on the container storage capacity of each recycling point, and multiple recycling point paths are merged according to the nearest neighbor insertion method;

[0114] In the event of an anomaly, freeze the current task queue and rerun the Dijkstra algorithm to generate a new optimized path.

[0115] Specifically, based on the historical storage capacity data of each recycling point, regression analysis is used to fit the trend of storage capacity change, and the storage capacity change in the next hour is predicted based on the fitted storage capacity change function.

[0116] The urban road network is abstracted as a directed graph, and road weights and travel time costs ω are calculated. ij (t) represents the travel time from node i to node j in time t, which is affected by factors such as traffic flow and road congestion, and can be dynamically calculated using traffic monitoring data.

[0117] For each point p to be recycled i Dijkstra's algorithm is used to calculate the shortest path from the vehicle's current location to the point, with the following formula:

[0118] d(v)=min(d(v),d(u)+ω uv )

[0119] Where d(v) is the shortest path from the starting point to node v, ω uv Let be the weight of the edge (u,v), where u is the current node;

[0120] Paths are weighted based on the storage capacity of the reclaimed points. If a reclaimed point has a low capacity (i.e., close to full capacity), it has a higher priority and a greater weight. Paths with multiple reclaimed points are merged using the nearest neighbor insertion method. If multiple reclaimed points exist, the following strategy is adopted:

[0121] Starting from the current node, select the nearest recycling point to visit and update the path;

[0122] Each time, select the next nearest recycling point until all recycling points have been visited;

[0123] In the event of an anomaly, the current task queue processing is paused, and the Dijkstra algorithm is rerun based on the new traffic conditions to generate a new optimal path. The shortest time path for all recycling points is recalculated, taking into account changes in road weights.

[0124] Furthermore, the method according to the embodiments of this application can also be achieved by means of... Figure 6 The architecture of the electronic device shown is used to implement this. For example... Figure 6 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store a recycling method and system based on a renewable resource cloud platform provided in this application. The electronic device 500 may also include a terminal interface 508. Of course, Figure 6 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 6 One or more components in the illustrated electronic device.

[0125] Figure 7 This is a schematic diagram of a computer-readable storage medium structure provided in one embodiment of this application. Figure 7 The diagram illustrates a computer-readable storage medium 600 according to one embodiment of this application. The computer-readable storage medium 600 stores computer-readable instructions. When executed by a processor, the computer-readable instructions can perform a recycling method and system based on a cloud platform for renewable resources, as described above with reference to the accompanying drawings, according to an embodiment of this application. The storage medium 600 includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0126] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0127] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0128] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A recycling method based on a cloud platform for renewable resources, characterized in that, include: Based on the historical data of recycled resources in the cloud platform, they are divided into metal, plastic and paper categories, and image data of each category of recycled resources are obtained. The acquired historical image data of recycled resources is input into a lightweight convolutional neural network to train a resource type identification model, and the trained resource type identification model is obtained. By acquiring image data and infrared spectral data of waste resources through multi-source sensors, the pre-processed data is input into the resource type identification model to identify and obtain recyclable metal, plastic and paper resources in the waste resources. Based on the identification results, resources are classified, and waste is sorted and stored in separate containers using a compression and sorting device, with basic information labels added. By monitoring the container storage capacity of each recycling point in real time through the cloud platform, and combining geographical location data and traffic conditions, the optimal recycling route is dynamically planned and tasks are assigned to recycling vehicles. A storage capacity change function is constructed based on the changes in container storage capacity data at each recycling point, and the storage capacity change data for the next hour is obtained based on the trend of the function. The city roads are abstracted as a directed graph, with road intersections and recycling points as nodes, the direction of traffic on the road segment as edges, and the weights being dynamically calculated travel time costs. Based on Dijkstra's algorithm, the shortest time-consuming path from the vehicle's current location to that point is calculated independently for each recycling point; Weights are added based on the container storage capacity of each recycling point, and multiple recycling point paths are merged according to the nearest neighbor insertion method; In the event of an anomaly, freeze the current task queue and rerun the Dijkstra algorithm to generate a new optimized path.

2. The recycling method based on a renewable resource cloud platform according to claim 1, characterized in that, The step of inputting the acquired historical recycled resource image data into a lightweight convolutional neural network for training a resource type identification model, and obtaining the trained resource type identification model specifically includes: Based on the acquired historical image data of recycled resources, the data is labeled and divided into metal, plastic and paper categories, and stored in the corresponding folders respectively; Based on the number of samples in each category, data balance was achieved through data augmentation and supplementary collection, and the data was divided into training set, validation set, and test set according to the proportions. The training set data is input into a lightweight convolutional neural network, and the model is trained by optimizing it through transfer learning. The system records the loss and accuracy of the training and validation sets in real time, adjusts the hyperparameters under conditions of underfitting and overfitting, and completes the training. The accuracy of the model is calculated using the reserved test set, a confusion matrix is ​​generated to analyze the identification errors of each category, and targeted optimizations are made for misclassified samples of metal, plastic, and paper. Export the trained resource type recognition model.

3. The recycling method based on a renewable resource cloud platform according to claim 1, characterized in that, The process of acquiring image data and infrared spectral data of waste resources through multi-source sensors, and inputting the preprocessed data into the resource type identification model to identify and obtain recyclable metal, plastic, and paper resources in the waste resources specifically includes: The surface texture, shape, and color image data of waste are acquired using a visible light camera, and the vibrational characteristic data of the molecules of the material are acquired using a near-infrared spectrometer. Based on the acquired multi-source data, the timestamps of the image and spectral data are aligned, and the multi-source data of the same waste sample are stored together. Image data preprocessing and infrared spectral data preprocessing are performed based on the acquired multi-source data to obtain waste characteristic data; Waste characteristic data is input into the resource type identification model, the probability distribution of waste classification results is output, and the final classification result is determined.

4. The recycling method based on a cloud platform for renewable resources according to claim 1, characterized in that, Based on the identification results, resource classification is performed, and waste is sorted and stored in individual containers using a compression and sorting device, with basic information tags added. Specifically, this includes: Based on the model recognition results, priority is defined according to the recycling value, and sorting instructions are generated for waste in priority order. Based on the sorted and classified waste, a differentiated compression strategy is adopted according to the type to compress and package various types of waste. Each compressed and packaged waste item is labeled with basic information, and the label information is uploaded to the cloud platform system in real time. The information labels include static information: category, weight, and compression time, and dynamic information: traceability code. Based on the type of compression and packaging, waste is stored in the corresponding storage container, and the remaining capacity of the container is monitored in real time. When storing waste, the waste category is verified a second time by photoelectric sensors, and missorted items are rejected for storage.

5. A recycling system based on a renewable resource cloud platform, used to implement the recycling method based on a renewable resource cloud platform as described in any one of claims 1-4, characterized in that, include: Data acquisition module: The data acquisition module is used to collect historical recycled resource data and acquire waste resource image data and infrared spectral data from multi-source sensors; Resource identification module: The resource identification module is used to train a resource type identification model based on a lightweight convolutional neural network, identify and classify recyclable resources, and classify them into metal, plastic and paper. Sorting and storage module: The sorting and storage module is used to classify and store waste into individual containers according to the identification results through a compression sorting device, and add basic information labels; Route planning module: The route planning module is used to monitor the storage capacity of containers at each recycling point in real time, and dynamically plan the optimal recycling route by combining geographical location and traffic data; Cloud platform monitoring module: The cloud platform monitoring module is used to monitor the container capacity and recycling vehicle status of the recycling point in real time through the cloud platform, and to optimize resource recycling paths and task allocation.

6. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed, enable the at least one processor to perform a recycling method based on a renewable resource cloud platform as described in any one of claims 1-4.

7. A computer-readable storage medium storing computer-readable instructions, characterized in that, When the computer-readable instructions are executed by a processor, they implement the recycling method based on a renewable resource cloud platform according to any one of claims 1-4.

Images