Smart city garbage intelligent classification monitoring and management method and device

Abstract

The invention relates to the technical field of smart city management, and particularly provides a smart city garbage intelligent classification monitoring and management method and device, and the method comprises the following steps: S1, image collection and recognition and data processing; s2, large model analysis data processing and classification optimization; s3, garbage separation control design and operation; and S4, constructing and operating intelligent monitoring management. Compared with the prior art, the system can effectively solve the problems of low efficiency, poor accuracy, insufficient monitoring, management lag and the like of traditional manual classification, has the characteristics of high automation degree, accurate classification, intelligent management, high adaptability and the like, and provides an efficient solution for a smart city garbage treatment system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart city management, and specifically provides a smart city garbage intelligent classification monitoring and management method and device. BACKGROUND

[0002] With the acceleration of urbanization, the amount of urban garbage is growing explosively. The traditional garbage disposal mode, which takes manual classification as the core, has been difficult to meet the efficient, environmentally friendly and intelligent management needs of smart cities. The problems existing in the traditional garbage disposal mode mainly lie in the following aspects.

[0003] (1) Low efficiency and high cost of manual classification: Large-scale garbage disposal plants can handle hundreds to thousands of tons of garbage per day. Manual classification requires a large amount of manpower, and each person can only handle about 0.5 tons of garbage per hour. Taking a medium-sized plant that handles 500 tons of garbage per day as an example, 1000 people are needed per day, which not only has high labor costs, but also requires workers to work in a harsh environment of odor, dust, etc. for a long time, with great labor intensity and high personnel mobility.

[0004] (2) Difficulty in ensuring classification accuracy: The types of garbage are diverse, and some composite materials (such as plastic and metal composite packaging, coated paper products, etc.) are difficult to accurately classify by the naked eye. At the same time, workers are prone to fatigue after long periods of repetitive work, resulting in a high error rate in classification. According to industry statistics, the error rate of manual classification is generally 15%-20%, and a large amount of misclassified garbage enters the non-corresponding processing channel, reducing resource recycling efficiency and increasing the risk of environmental pollution.

[0005] (3) Lack of full-process monitoring capability: The traditional mode lacks real-time monitoring means for garbage source classification, transfer and transportation, and terminal processing: garbage collection points cannot monitor the fullness of garbage cans in real time, leading to delayed cleaning and transportation; it is difficult to track the completeness of garbage classification during transfer; and processing plants cannot accurately control the classification effect at the front end, making it difficult to adjust subsequent processing, forming a vicious cycle of "source out of control-process out of control-low efficiency at the terminal".

[0006] (4) Low level of intelligent management: unable to analyze and predict trends based on garbage production and classification data, making it difficult to optimize resource allocation: factors such as seasonal and holiday fluctuations in garbage volume are not taken into account, making it difficult to adjust processing equipment parameters in advance, which can lead to resource waste such as "insufficient processing capacity during peak periods and idle equipment during off-peak periods"; at the same time, there is a lack of cross-link data linkage, making it impossible to achieve intelligent scheduling of the entire garbage "generation-classification-transportation-processing" chain.

[0007] In summary, how to solve the problems of low efficiency, poor accuracy, lack of full-process monitoring and insufficient intelligent management of traditional manual garbage classification in the prior art is a problem that needs to be solved by those skilled in the art. SUMMARY

[0008] The present application is aimed at the deficiencies of the prior art, and provides a smart city garbage intelligent classification monitoring and management method with strong practicability.

[0009] The further technical task of the present application is to provide a smart city garbage intelligent classification monitoring and management device with reasonable design and safety.

[0010] The technical solution adopted by the present application to solve its technical problems is:

[0011] A smart city garbage intelligent classification monitoring and management method has the following steps:

[0012] S1, image acquisition, recognition and data processing;

[0013] S2, large model analysis data processing and classification optimization;

[0014] S3, garbage channel control design and operation;

[0015] S4, construction and operation of intelligent monitoring and management.

[0016] Further, in step S1, intelligent cameras are deployed at garbage collection points, garbage transfer stations and garbage treatment plants, the collection equipment is connected with edge computing equipment NVIDIA Jetson Xavier NX, image data is transmitted to the edge end for preprocessing in real time, the image quality in low light environment is optimized through the image enhancement algorithm Retinex, the garbage area is segmented by using the target detection algorithm YOLOv8, and the background interference is removed;

[0017] The garbage collection point barrel bottom is provided with a weight sensor for real-time collection of garbage weight data, and an infrared overflow sensor is installed on the barrel body, which automatically triggers an overflow alarm when the garbage height reaches 80% of the barrel opening; the data is transmitted to the management platform through LoRa wireless transmission, and the transmission interval can be set to 3-7 minutes / time;

[0018] The garbage transfer vehicle transporting garbage to the garbage transfer station is provided with a GPS positioning module, a load sensor and a video monitoring device, which uploads the vehicle position, the garbage weight carried and the car compartment sealing state data in real time; each vehicle is provided with a 4G / 5G router to ensure stable data transmission during transportation.

[0019] The garbage treatment plant is provided with a flow counter at the outlet of each classification channel and temperature, pressure and power sensors on the treatment equipment for real-time monitoring of equipment operating parameters; the data is aggregated through the industrial bus Modbus protocol to the local server of the treatment plant, and then synchronized to the cloud management platform.

[0020] Further, based on the trained garbage classification image recognition model, the shape contour, material characteristics, and color parameters of the garbage are extracted and compared with the preset garbage feature database to preliminarily determine the category of the garbage and generate the associated data of "garbage image-preliminary category-collection time-place";

[0021] The newly added garbage category information of the city garbage classification management department is connected, and the sample data is automatically supplemented every month to optimize the model recognition accuracy and adapt to the classification needs of new types of garbage.

[0022] Further, in step S2, the edge computing device transmits the step S1 preliminary classification data to the cloud server through 5G or optical fiber network, and adopts AES-256 encryption algorithm to ensure data security during transmission;

[0023] The cloud server cleans the received data, removes fuzzy images and repeated data, and completes the missing information to form a standardized data set.

[0024] Further, the cloud deploys a large model optimized for garbage classification, fine-tunes based on the Transform architecture, integrates garbage processing knowledge graph, inputs the standardized data set, and conducts deep analysis combined with multi-dimensional information:

[0025] (1) Spatial dimension: associate the geographical information of the garbage collection point to determine the characteristics of the garbage source;

[0026] (2) Time dimension: analyze the garbage generation rule combined with the collection time;

[0027] (3) Feature dimension: for garbage that is difficult to classify in the preliminary classification, the large model conducts deep analysis on the microscopic texture and material composition to correct the preliminary classification result;

[0028] The large model receives terminal processing feedback data in real time and performs incremental training once a week: by comparing the difference between the predicted category and the actual category, adjusting the model parameters, and continuously improving the classification accuracy in complex scenarios.

[0029] Further, in step S3, 4-6 classification processing channels are arranged along the downstream direction of the conveyor belt in the sorting workshop of the garbage treatment plant, and an intelligent shunting device is installed at the entrance of each channel. The intelligent shunting device is composed of an electric push rod, a rotating baffle, and a position sensor, and the baffle material is made of wear-resistant polyethylene;

[0030] The shunting device is connected with the cloud server through industrial Ethernet to receive the garbage category signal output by the large model; at the same time, laser sensors are installed on both sides of the conveyor belt to detect the position of the garbage in real time and feed back the data to the shunting device controller.

[0031] Further, when the garbage moves along the conveyor belt to the diversion area, the laser sensor detects the position of the garbage and sends a trigger signal to the controller;

[0032] The controller calls the garbage category data issued by the cloud, and according to the preset "category-channel" corresponding rule, sends an action instruction to the electric push rod of the corresponding channel;

[0033] The electric push rod pushes the rotating baffle to rotate to a specified angle, guiding the garbage to deviate from the original conveyor belt track and enter the corresponding processing channel; after the baffle action is completed, the position sensor is reset and waits for the next instruction.

[0034] For large-volume garbage, automatically start the double-channel linkage: measure the size of the garbage by the laser sensor, if it exceeds the width of the single channel, the controller synchronously triggers the baffles of the adjacent two channels to form a combined channel, ensuring the smooth passage of the garbage.

[0035] Further, if the diversion device fails, the position sensor feeds back an abnormal signal, and the controller immediately suspends the operation of the conveyor belt in the corresponding area and sends an alarm information to the management platform;

[0036] The garbage that has not been classified is temporarily stored in the temporary buffer area, and after the fault is eliminated, the classification data of this part of garbage is called again for secondary diversion.

[0037] Further, in step S4, the management platform uses a Web + mobile terminal architecture to display the garbage generation amount, classification accuracy and equipment running status in real time in the form of column chart, line chart and heat map;

[0038] Based on the garbage generation trend data analyzed by the large model, the cleaning and transportation route is automatically optimized;

[0039] When the equipment operating parameters exceed the normal range and the classification accuracy is less than 90%, the platform automatically triggers an audible and light alarm and pushes the warning information to the management personnel; at the same time, based on historical fault data, the large model gives fault troubleshooting suggestions to assist the management personnel to quickly handle problems;

[0040] Daily report, weekly report and monthly report are automatically generated, covering total garbage processing amount, proportion of each type of garbage, classification accuracy and equipment failure rate, and data export is performed.

[0041] A smart city garbage intelligent classification monitoring and management device, comprising: at least one memory and at least one processor;

[0042] The at least one memory is used to store machine readable programs;

[0043] The at least one processor is used to call the machine readable programs and execute a smart city garbage intelligent classification monitoring and management method.

[0044] Compared with the prior art, the smart city garbage intelligent classification monitoring and management method and device of the present application has the following outstanding beneficial effects:

[0045] (I) greatly improving garbage classification efficiency and accuracy;

[0046] Efficiency improvement: traditional manual classification can process 0.5 tons of garbage per hour per person, while the system can process 5-8 tons of garbage per hour per single sorting line through high-speed image acquisition (frame rate 30 fps), edge fast processing (delay < 500 ms) and automatic lane control (response time < 100 ms), efficiency improvement 10-16 times; the classification preprocessing time of garbage at the collection point and transfer station is shortened by more than 60%, avoiding a large amount of garbage accumulation waiting for classification.

[0047] Accuracy improvement: the error rate of manual classification is 15%-20%, and the error rate of existing single image recognition classification equipment is about 8%-10%, the system can reduce the error rate to below 5% through "image recognition + large model multi-dimensional analysis" combined with garbage source, time, environment and other information to correct the classification results; the recognition accuracy of complex composite garbage (such as plastic containers with metal fittings, paper packaging contaminated with oil) is improved to more than 92%, greatly reducing the waste of resources and environmental pollution caused by misclassification.

[0048] (II) reducing garbage disposal cost and improving working environment

[0049] Labor cost reduction: traditional mode of medium-sized garbage disposal plant requires 100-150 frontline classification workers, after using the system, only 10-15 people are needed to be responsible for equipment monitoring and abnormal handling, labor cost is reduced by 85%-90%; at the same time, the frequency of direct contact with garbage is reduced, avoiding the influence of garbage odor and bacteria on the health of workers, and the working environment is significantly improved.

[0050] Energy consumption and operation cost optimization: through large model prediction of garbage generation trend, adjust the running parameters of processing equipment in advance (such as reducing the power of incinerator during off-peak period, optimizing the route of cleaning vehicle), the overall energy consumption of garbage disposal plant is reduced by 15%-20%; due to route optimization, the empty running rate of cleaning vehicle is reduced by more than 25%, which can save fuel cost by 30%-40% per year.

[0051] (III) realizing transparent monitoring and intelligent management of the whole process;

[0052] Whole-link traceability: From garbage disposal (collection point image recording), transfer (vehicle positioning and load data) to processing (channel recording and equipment operating parameters), the whole process data is uploaded to the management platform in real time to form a "garbage identity file". The processing path of each batch of garbage can be traced through time, location, and category dimensions, solving the problem of "black box operation" in traditional mode, and facilitating supervision departments to check.

[0053] Intelligent scheduling and risk early warning: Based on the prediction of garbage generation (accuracy rate of more than 85%) by the large model, the garbage collection vehicles and processing equipment are allocated in advance to avoid environmental pollution caused by garbage overflow (the collection point overflow rate is reduced from 20%-30% in the traditional mode to less than 5%); the response time of equipment failure warning is shortened to 5-10 minutes, and the downtime of equipment failure is reduced by 70%, ensuring the stable operation of the garbage treatment system.

[0054] (Four) Promote the upgrading of smart city environmental governance;

[0055] Data support decision-making: The data of garbage generation, category proportion, and regional distribution accumulated by the system provide the basis for urban planning, such as: according to the data analysis of the high proportion of kitchen waste (>50%) in a certain area, it is suggested to add a kitchen waste on-site treatment station; combined with the distribution data of recyclable materials, the layout of waste recycling points is optimized to improve the resource recycling rate (the recycling rate of recyclable materials is improved from 20%-30% in the traditional mode to 45%-55%).

[0056] Through the management platform, the garbage classification accuracy rate of each community and street is displayed to provide quantitative basis for garbage classification assessment; at the same time, the garbage classification ranking and optimization suggestions of the area are pushed to the mobile terminal of residents to guide residents to standardize disposal and help achieve the goal of "green and low carbon" of smart city.

[0057] (Five) Strong adaptability and scalability;

[0058] Wide scene adaptation: The system can flexibly adjust the equipment layout (such as adding industrial waste special identification module in industrial park) and model parameters according to the garbage generation characteristics of different scale areas (such as residential communities, commercial complexes, and industrial parks) to adapt to diversified garbage treatment needs; at the same time, it supports the connection with existing city smart management platforms (such as smart city management and smart environmental protection systems) to realize data sharing and collaborative management.

[0059] Convenient technology iteration: The image recognition model and large model adopt modular design, which can realize technology upgrade through online incremental training without the need for large-scale modification of hardware equipment; in the future, functions such as AI voice guidance (such as reminding residents to correctly dispose garbage) and unmanned aerial vehicle inspection (such as investigating the problem of garbage dumping in remote areas) can be expanded to continuously improve the intelligent level of the system. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to make the technical scheme of the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and all other drawings obtained by those of ordinary skill in the art without creative work belong to the protection scope of the present application.

[0061] Figure 1 It is a flowchart of a smart city garbage intelligent classification monitoring and management method.

[0062] Figure 2 It is an image acquisition identification and data processing flowchart of a smart city garbage intelligent classification monitoring and management method.

[0063] Figure 3 It is a garbage channel control flowchart of a smart city garbage intelligent classification monitoring and management method.

[0064] Figure 4 It is an intelligent monitoring management data flowchart of a smart city garbage intelligent classification monitoring and management method. DETAILED DESCRIPTION

[0065] In order to make the technical scheme of the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and all other drawings obtained by those of ordinary skill in the art without creative work belong to the protection scope of the present application.

[0066] A preferred embodiment is given below:

[0067] As shown in Figure 1 , a smart city garbage intelligent classification monitoring and management method in the embodiment has the following steps:

[0068] S1, image acquisition identification and data processing;

[0069] As shown in Figure 2 , the garbage collection point: beside the community garbage can, street drop station and other positions, install high-definition intelligent camera with wide-angle shooting (view angle 120°-150°), infrared light supplement function (resolution not less than 4K), ensure that the garbage shape, color, texture and other characteristics can be clearly captured in strong light during the day and low illumination environment at night; each collection point is equipped with 1-2 cameras, covering the drop port and the surrounding area of the garbage can, avoiding the blind area of shooting.

[0070] Garbage transfer station: At the garbage unloading platform and the entrance of the conveyor belt, deploy a mobile track image acquisition device (support 360° rotation shooting), matched with a laser ranging sensor, to synchronously collect garbage image information and volume data; the device height can be adjusted according to the model of the transfer vehicle to ensure that the shooting range covers the entire unloading process.

[0071] Garbage treatment plant: Install industrial-grade high-speed cameras (frame rate not less than 30 fps) at key positions such as above the garbage sorting line and the pretreatment workshop to continuously capture images of the flowing garbage on the conveyor belt, ensuring no missed collection; the camera is linked with the running speed of the conveyor belt to avoid image blurring caused by garbage movement.

[0072] Connect the acquisition device with the edge computing device NVIDIA Jetson Xavier NX, and transmit the image data to the edge end for real-time preprocessing: use the image enhancement algorithm Retinex to optimize the image quality in low light environments, and use the target detection algorithm YOLOv8 to segment the garbage area and remove background interference.

[0073] Based on the trained garbage classification image recognition model (optimized through 1 million+ garbage samples of various types (including common household garbage, industrial scraps, and special waste)), extract the shape contour (such as circular, square), material characteristics (such as the smoothness of plastic, the reflectivity of metal), and color parameters (such as the yellow-brown color of kitchen waste, the red color of hazardous waste), and compare them with the pre-set garbage feature database (classified by recyclables, kitchen waste, hazardous waste, and other waste, and further classified into subcategories such as plastic bottles, waste paper, and waste batteries), to preliminarily determine the category of the garbage, generate associated data of "garbage image - preliminary category - collection time - location", and control the processing delay within 500ms.

[0074] The feature database supports online updates: by interfacing with the city's garbage classification management department to add new garbage category information, automatically supplement sample data every month, optimize model recognition accuracy, and adapt to the classification needs of new types of garbage (such as biodegradable plastic, electronic cigarette waste, etc.).

[0075] S2, large model analysis data processing and classification optimization;

[0076] The edge computing device transmits the preliminary classification data (including image features, preliminary categories, and collection information) to the cloud server (using Ali Cloud, Huawei Cloud, etc. with high concurrent processing capability) through 5G / optical fiber network, and uses AES-256 encryption algorithm to ensure data security during transmission.

[0077] Cloud server cleans the received data: removes ambiguous images (clarity less than 80%), repeated data (the same garbage is photographed multiple times), and fills in missing information (such as unmarked collection location) by associating with pre-set location information through device ID, forming a standardized data set.

[0078] Cloud deployment of large model optimized for garbage classification (based on Transform architecture fine-tuning, incorporating garbage processing domain knowledge graph), input standardized data set, combined with multi-dimensional information for deep analysis:

[0079] Spatial dimension: associate garbage collection point geographic information (such as community occupancy rate, surrounding business types (catering area / office area)), determine garbage source characteristics (such as high proportion of kitchen waste around catering area, more waste paper around office area);

[0080] Time dimension: combined with collection time (such as morning rush hour, holidays, seasonal changes), analyze garbage generation patterns (such as packaging waste surge during Spring Festival, kitchen waste prone to spoilage in summer);

[0081] Feature dimension: for garbage that is difficult to classify in preliminary classification (such as plastic meal box with oil stains, paper packaging with metal labels), through deep analysis of micro texture (such as plastic meal box surface oil stain distribution), material composition (estimated through image spectrum analysis), the preliminary classification results are corrected.

[0082] Example: a "white sheet object" is collected at a certain community collection point, and the preliminary classification is "waste paper"; the large model combines the surrounding tea shops in the community, the collection time is the afternoon tea period, and the image object has wrinkles and transparent film on the surface, and determines it as "noodle tea cup outer paper packaging (belongs to recyclable, but needs to remove plastic film)", and outputs classification correction suggestions.

[0083] Large model receives terminal processing feedback data (such as manually reviewed misclassification cases, processing plant actual classification results) in real time, and performs incremental training once a week: by comparing the difference between predicted and actual categories, adjusting model parameters (such as feature weights, classification thresholds), continuously improving classification accuracy in complex scenarios, ensuring that the model can maintain an accuracy rate of more than 95% when garbage types are updated.

[0084] S3, garbage channel control design and operation;

[0085] For example Figure 3As shown, in the sorting workshop of the waste treatment plant, 4-6 classification processing channels (corresponding to recyclables, kitchen waste, hazardous waste, other waste, etc.) are arranged along the downstream direction of the conveyor belt, and an intelligent diversion device is installed at the entrance of each channel. The intelligent diversion device is composed of an electric push rod, a rotating baffle and a position sensor. The baffle is made of wear-resistant polyethylene and can withstand an impact of waste of less than 50 kg.

[0086] The diversion device is connected with the cloud server through industrial Ethernet to receive the waste category signal output by the large model. At the same time, laser sensors are installed on both sides of the conveyor belt to detect the position of the waste in real time (accuracy ± 5 cm) and feed the data to the diversion device controller (controlled by PLC programming).

[0087] When the waste moves along the conveyor belt to the diversion area, the laser sensor detects the position of the waste and sends a trigger signal to the controller.

[0088] The controller calls the waste category data issued by the cloud and sends an action instruction (response time ≤100 ms) to the electric push rod of the corresponding channel according to the preset "category-channel" corresponding rule (e.g. recyclables→ No. 1 channel, hazardous waste→ No. 3 channel).

[0089] The electric push rod pushes the rotating baffle to rotate to a specified angle (0°-90° adjustable), guides the waste to deviate from the original conveyor belt track and enters the corresponding processing channel. After the baffle action is completed, the position sensor is reset and waits for the next instruction.

[0090] For large-volume waste (such as furniture and electrical appliances), the system automatically starts the double-channel linkage: the laser sensor measures the size of the waste, and if it exceeds the width of a single channel, the controller simultaneously triggers the baffles of the adjacent two channels to form a combined channel, ensuring the smooth passage of the waste.

[0091] If the diversion device fails (such as baffle jam), the position sensor feeds back an abnormal signal, and the controller immediately suspends the operation of the conveyor belt in the corresponding area and sends an alarm information (including fault position and fault type) to the management platform.

[0092] The waste that has not been classified is temporarily stored in the temporary buffer area. After the fault is eliminated, the system re-calls the classification data of this part of waste for secondary diversion to avoid waste accumulation.

[0093] S4, construction and operation of intelligent monitoring and management;

[0094] For example Figure 4As shown, a weight sensor (precision ±0.1 kg) is installed at the bottom of the garbage can to collect real-time garbage weight data; an infrared overflow sensor is installed on the barrel to automatically trigger an overflow alarm when the garbage height reaches 80% of the barrel opening; data is transmitted to the management platform through LoRa wireless transmission, and the transmission interval can be set to 5 minutes / time (real-time transmission when full).

[0095] A GPS positioning module, load sensor, and video monitoring equipment are installed on the garbage transfer vehicle to upload real-time vehicle location, garbage weight, and vehicle compartment sealing status (leakage prevention) data; each vehicle is equipped with a 4G / 5G router to ensure stable data transmission during transportation.

[0096] A flow counter (statistical unit of garbage throughput per unit time) is installed at the outlet of each classification channel, and temperature, pressure, and power sensors are installed on processing equipment (such as incinerators and crushers) to monitor equipment operating parameters in real time; data is aggregated through the Modbus protocol to the local server of the processing plant and then synchronized to the cloud management platform.

[0097] The management platform uses a Web + mobile (APP / mini program) architecture to display real-time data such as garbage generation, classification accuracy, and equipment operating status in the form of column charts, line charts, and heat maps: the "garbage classification accuracy heat map" visually presents the classification results of different neighborhoods, and the "garbage generation trend line chart" predicts changes in garbage volume in each region over the next 7 days.

[0098] Based on the analysis of garbage generation trend data, the system automatically optimizes the collection and transportation route: for example, in response to the concentrated garbage generation in small communities during the morning rush hour (7:00-9:00), the system adjusts the frequency of garbage collection vehicles and prioritizes covering small communities with high occupancy rates; for collection points with overflow alarms, the system automatically generates dispatch information and pushes it to nearby collection personnel on their mobile devices.

[0099] When equipment operating parameters exceed normal ranges (e.g., incinerator temperature > 1200°C) or classification accuracy is less than 90%, the platform automatically triggers an audible and visual alarm and sends warning information to management personnel; at the same time, based on historical fault data, the large model provides fault troubleshooting suggestions (e.g., "temperature anomaly may be due to excessive feed speed, suggest reducing feed rate to 2 t / h") to assist management personnel in quickly addressing issues.

[0100] Daily, weekly, and monthly reports are automatically generated, covering key indicators such as total garbage processing volume, proportion of each type of garbage, classification accuracy, and equipment failure rate; data export (formats include Excel, PDF, etc.) is supported to provide decision-making basis for urban garbage classification management departments.

[0101] Based on the above method, the intelligent city garbage intelligent classification monitoring and management device in the embodiment comprises at least one memory and at least one processor.

[0102] The at least one memory is used for storing a machine readable program.

[0103] The at least one processor is used for calling the machine readable program and executing an intelligent city garbage intelligent classification monitoring and management method.

[0104] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A smart city garbage intelligent classification monitoring and management method, characterized in that, Having the following steps: S1, image acquisition, recognition and data processing; S2, large model analysis data processing and classification optimization; S3, garbage channel control design and operation; S4, construction and operation of intelligent monitoring management. 2.The smart city garbage intelligent classification monitoring and management method according to claim 1, characterized in that, In step S1, intelligent cameras are deployed at garbage collection points, garbage transfer stations and garbage treatment plants. The collection equipment is connected with edge computing equipment NVIDIA Jetson Xavier NX. Image data is transmitted to the edge in real time for preprocessing. Image quality in low light environment is optimized through image enhancement algorithm Retinex. Target detection algorithm YOLOv8 is used to segment the garbage area and remove background interference. The bottom of the garbage collection point barrel is equipped with a weight sensor to collect real-time garbage weight data. An infrared overflow sensor is installed on the barrel body. When the garbage height reaches 80% of the barrel opening, an overflow alarm is automatically triggered. Data is transmitted to the management platform through LoRa wireless transmission with a transmission interval of 3-7 minutes per time. The garbage transfer vehicle transporting garbage to the garbage transfer station is equipped with a GPS positioning module, a load sensor and a video monitoring device to upload real-time vehicle location, garbage weight and vehicle compartment sealing state data. Each vehicle is equipped with a 4G / 5G router to ensure stable data transmission during transportation. The garbage treatment plant is equipped with a flow counter at the outlet of each classification channel and temperature, pressure and power sensors on the treatment equipment to monitor equipment operating parameters in real time. Data is aggregated through the industrial bus Modbus protocol to the local server of the treatment plant and then synchronized to the cloud management platform. 3.The smart city garbage intelligent classification monitoring and management method according to claim 2, characterized in that, Based on the trained garbage classification image recognition model, the shape contour, material characteristics and color parameters of the garbage are extracted and compared with the pre-set garbage feature database to preliminarily determine the category of the garbage and generate associated data of "garbage image-preliminary category-collection time-location". The new garbage category information is connected with the city garbage classification management department. Sample data is automatically supplemented every month to optimize the model recognition accuracy and adapt to the classification needs of new types of garbage. 4.The smart city garbage intelligent classification monitoring and management method according to claim 3, characterized in that, In step S2, the edge computing equipment transmits the preliminary classification data of step S1 to the cloud server through 5G or optical fiber network. AES-256 encryption algorithm is used to ensure data security during transmission. The cloud server cleanses the received data, removes ambiguous images and duplicate data, and completes missing information to form a standardized data set. 5.The smart city garbage intelligent classification monitoring and management method of claim 4, wherein, The cloud deploys a large model optimized for garbage classification, fine-tunes based on the Transform architecture, integrates garbage treatment knowledge graph, inputs standardized data set, and conducts deep analysis combined with multi-dimensional information: (1) Spatial dimension: correlate with the geographical information of garbage collection points to determine the characteristics of garbage sources; (2) Time dimension: analyze garbage generation patterns based on collection time; (3) Feature dimension: for garbage that is difficult to classify, the large model analyzes the micro texture and material composition to correct the preliminary classification results; The large model receives real-time terminal processing feedback data and performs incremental training once a week: by comparing the difference between predicted categories and actual categories, the model parameters are adjusted to continuously improve the classification accuracy in complex scenarios. 6.The smart city garbage intelligent classification monitoring and management method of claim 5, wherein, In step S3, in the sorting workshop of the waste treatment plant, 4-6 classification processing channels are arranged along the downstream direction of the conveyor belt, and an intelligent shunting device is installed at the entrance of each channel. The intelligent shunting device is composed of an electric push rod, a rotating baffle and a position sensor, and the baffle is made of wear-resistant polyethylene; The shunting device is connected with the cloud server through industrial Ethernet, and receives the garbage category signal output by the large model; at the same time, laser sensors are installed on both sides of the conveyor belt to detect the position of the garbage in real time and feed back the data to the shunting device controller. 7.The smart city garbage intelligent classification monitoring and management method of claim 6, wherein, When the garbage moves along the conveyor belt to the shunting area, the laser sensor detects the position of the garbage and sends a trigger signal to the controller; The controller calls the garbage category data issued by the cloud, and according to the preset "category-channel" corresponding rule, sends an action instruction to the electric push rod of the corresponding channel; The electric push rod pushes the rotating baffle to rotate to a specified angle, guiding the garbage to deviate from the original conveyor belt track and enter the corresponding processing channel; after the baffle action is completed, the position sensor confirms the reset and waits for the next instruction. For large-volume garbage, automatically start double-channel linkage: measure the size of the garbage by laser sensor, if it exceeds the width of single-channel, the controller synchronously triggers the baffles of the adjacent two channels to form a combined channel, ensuring the smooth passing of the garbage. 8.The smart city garbage intelligent classification monitoring and management method of claim 7, wherein, If the shunting device fails, the position sensor feeds back an abnormal signal, and the controller immediately suspends the operation of the conveyor belt in the corresponding area and sends an alarm information to the management platform; The garbage that has not completed classification is temporarily stored in the temporary buffer area, and after the fault is eliminated, the classification data of this part of garbage is called again for secondary shunting. 9.The smart city garbage intelligent classification monitoring and management method of claim 8, wherein, In step S4, the management platform adopts Web + mobile terminal architecture, and displays the garbage generation amount, classification accuracy and equipment running state of each area in the form of column chart, line chart and heat map in real time; Based on the garbage generation trend data analyzed by the large model, the cleaning and transportation route is automatically optimized; When the equipment running parameters exceed the normal range and the classification accuracy is less than 90%, the platform automatically triggers sound and light alarm and pushes the warning information to the management personnel; at the same time, based on the historical fault data, the large model gives fault troubleshooting suggestions to assist the management personnel to quickly handle the problem; Daily report, weekly report and monthly report are automatically generated, covering total garbage processing amount, proportion of each type of garbage, classification accuracy and equipment failure rate, and data export is performed.

10. A smart city garbage intelligent classification monitoring and management device, characterized in that, Comprise: At least one memory and at least one processor; The at least one memory is used to store a machine readable program; The at least one processor is used to call the machine readable program and execute the method of any one of claims 1 to 9.

Images