Method for treating kitchen waste through linkage of hermetia illucens age identification and intelligent feeding

Abstract

The invention relates to the field of biological waste treatment, and discloses a hermetia illucens age identification and intelligent feeding linkage kitchen waste treatment system, which comprises a data acquisition module, a data processing and activity index calculation module, a data processing and activity index calculation module, a data processing and activity index calculation module, a data processing and activity index calculation module, a data processing and activity index calculation module, and a data processing and activity index calculation module, the activity index is used for representing the activity state of the insect population and the adaptive capacity to the environment; the insect age identification module is based on a Transform model detected in real time and introduces the activity index to serve as an attention guiding signal so as to output insect age distribution, individual counting and health state labels; a feeding linkage control module; and a cloud management and self-learning module. By introducing multiple environmental factors such as temperature, humidity, gas concentration and the like and insect group activity information and combining with the improved deep learning model, the insect age can be identified more accurately, so that identification errors caused by factors such as illumination and insect body overlapping in a traditional method are effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of biological waste treatment technology, specifically a method for treating kitchen waste by linking black soldier fly age identification with intelligent feeding. Background Technology

[0002] Food waste treatment is a crucial aspect of urban solid waste management. Currently, mainstream treatment methods include incineration, landfill, and anaerobic digestion; however, these methods generally suffer from high energy consumption, secondary pollution, or low resource utilization rates. Therefore, finding a more efficient and environmentally friendly treatment method has become an urgent priority. In recent years, black soldier flies (Hermetia illucens), as an emerging biological treatment technology, have received widespread attention due to their high organic matter conversion rate and environmental friendliness. In the treatment of food waste, black soldier flies can not only rapidly digest organic waste but also provide reusable resources through their biotransformation process.

[0003] In black soldier fly farming, the age of the larvae significantly impacts their digestion efficiency, feed requirements, and overall health. However, existing factory farming methods often employ fixed frequencies and feeding amounts, failing to consider the dynamic changes in larval age and population activity during the farming process, leading to several problems. First, the accuracy of larval age identification is low. Current technologies rely on manual observation or single image recognition, which is significantly affected by changes in lighting and larval overlap, resulting in insufficient accuracy. Second, the influence of environmental factors on population activity is neglected. Environmental conditions such as temperature, humidity, and gas concentration significantly affect population growth and activity, but existing identification methods fail to incorporate these factors into the age determination model, leading to inaccurate age identification. Finally, current automation levels are low. The identification system lacks a closed-loop linkage with the feeding system, making it impossible to adjust feeding strategies in real time based on larval age changes, thus affecting farming efficiency and resource utilization. Furthermore, current age identification largely relies on single-frame image analysis, failing to fully utilize the dynamic characteristics of the population over time for state assessment, resulting in a lack of temporal dynamic modeling support for the identification process. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for the integrated treatment of kitchen waste using black soldier fly larvae age identification and intelligent feeding. This method solves the problem that using a fixed frequency and fixed feeding amount does not take into account the dynamic changes in larvae age and colony activity during the breeding process.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a system for the integrated treatment of kitchen waste using black soldier fly larvae age recognition and intelligent feeding, comprising:

[0006] The data acquisition module includes an image acquisition unit and an environmental factor acquisition unit. The image acquisition unit is used to acquire image or video data of the insect tank area, and the environmental factor acquisition unit is used to acquire environmental data such as temperature, humidity, light intensity, oxygen, carbon dioxide, ammonia, and nitrogen.

[0007] The data processing and activity index calculation module is used to normalize environmental data and extract temporal features, and fuse them with motion features obtained from video data to obtain an activity index, which is used to characterize the activity status and environmental adaptability of insect colonies.

[0008] The insect age recognition module is based on a real-time detection Transformer model and introduces the activity index as an attention guidance signal to output insect age distribution, individual count and health status label;

[0009] The feeding linkage control module queries the feeding strategy table based on the insect age distribution and activity index, individual count and health status label, and drives the feeding equipment to perform solid or liquid feeding.

[0010] The cloud management and self-learning module is used to aggregate batch data, retrain the model and perform strategy inference, and distribute feeding parameters.

[0011] The modules are coupled together via a closed-loop data and control system.

[0012] Preferably, the environmental factor acquisition unit includes a temperature sensor, a humidity sensor, a light intensity detector, and a gas detection node for detecting oxygen, carbon dioxide, ammonia, and nitrogen. The environmental factor acquisition unit supports periodic calibration and transmits data to the data processing and activity index calculation module through a standardized communication protocol.

[0013] Preferably, the data processing and activity index calculation module performs normalization on environmental data based on mean and standard deviation to achieve comparison of multi-source data on a unified scale, and performs time synchronization and missing data compensation on data streams from different sources to ensure the stability of subsequent modeling.

[0014] Preferably, the temporal feature extraction uses an improved long short-term memory network to model multiple environmental factor sequences and introduces a Transformer encoder to model global dependencies, thereby obtaining environmental temporal feature vectors.

[0015] Preferably, the motion features are obtained by analyzing the optical flow and density change rate of video data to form a motion feature vector, and the activity index is formed by weighted fusion or neural network fusion of the environmental temporal feature vector and the motion feature vector to quantify the activity level of the insect population.

[0016] Preferably, the insect age recognition module includes a real-time detection network based on RT-DETR and an improved HiDETR model. The activity index is used as an additional attention guidance signal to optimize the attention distribution and improve the accuracy of insect detection and insect age classification in complex scenarios. The improved HiDETR model includes a classification network for outputting insect age categories and a regression network for outputting continuous insect age prediction values.

[0017] Preferably, the feeding strategy table provides the correspondence between feed type and feeding amount based on the insect age stage and activity index, and supports the linkage adjustment of the ratio of solid feed and liquid nutrients according to the distribution of insect age. The feeding linkage control module collects response data while executing feeding and uses it as closed-loop feedback input.

[0018] Preferably, the cloud management and self-learning module aggregates historical data on insect age, activity, feeding amount and environmental factors over a long period of time, and retrains and updates the weights of the HiEnv-Transformer and HiDETR models to form an adaptive insect age identification and feeding strategy. The cloud management and self-learning module supports cross-regional and cross-batch strategy deduction and remote parameter distribution.

[0019] Preferably, the feeding linkage control module includes a screw feeder for solid feeding and a liquid feed pump for liquid feeding. The feeding linkage control module, the data acquisition module, and the insect age recognition module form a closed-loop control to dynamically adjust the feeding intensity, timing, and ratio based on real-time recognition results and activity index during the feeding process.

[0020] A method for treating kitchen waste by linking black soldier fly larvae age identification with intelligent feeding, characterized by the following steps:

[0021] S1. Hardware installation and configuration: Install the image acquisition unit, environmental factor acquisition unit and related sensors to ensure that the image acquisition unit can cover the insect tank area and regularly capture the movement and density distribution data of the insect population, and that the environmental factor acquisition unit can accurately acquire multi-dimensional environmental data such as temperature, humidity, light intensity and gas concentration.

[0022] S2. Data Acquisition and Real-time Transmission: Video data from the image acquisition unit and environmental data from the environmental factor acquisition unit are acquired periodically. The acquired data is transmitted in real time to the data processing and activity index calculation module through a standardized protocol and uploaded to the cloud database for subsequent storage and analysis.

[0023] S3. Data Preprocessing and Analysis: The collected environmental data is normalized and a time series model is used to provide a basis for subsequent insect age identification and feeding strategies.

[0024] S4. Insect Age Identification and Feeding Control: Based on real-time calculated activity index and image data, the improved HiDETR model is used to identify the insect age of the insect population. The optimal feeding strategy is queried from the feeding strategy table according to the insect age distribution and activity index. The operation of the feeding equipment is controlled through the feeding linkage control module to achieve precise feeding.

[0025] S5. Data Feedback and Optimization: After feeding is completed, feedback data of the insect swarm is collected in real time and uploaded to the cloud for analysis. The model is retrained and optimized using historical data. The cloud supports cross-regional and cross-batch insect swarm management and strategy simulation, and dynamically adjusts feeding parameters.

[0026] This invention provides a method for the integrated treatment of kitchen waste using black soldier fly larvae age identification and intelligent feeding. It has the following beneficial effects:

[0027] 1. This invention introduces multiple environmental factors, such as temperature, humidity, and gas concentration, along with insect activity information, and combines them with an improved deep learning model to more accurately identify insect age, thereby effectively reducing identification errors caused by factors such as light and insect overlap in traditional methods.

[0028] 2. This invention incorporates multi-dimensional environmental data such as temperature, humidity, and gas concentration into the insect age recognition model, making insect age determination more comprehensive and accurate, thereby improving data utilization efficiency and optimizing the breeding environment during the breeding process;

[0029] 3. Through closed-loop control of the system, the insect age identification and feeding system can be linked in real time. The feeding strategy can be dynamically adjusted according to the insect age and activity index, which effectively reduces human intervention and improves breeding efficiency and resource utilization.

[0030] 4. This invention introduces an LSTM model for time series data analysis, and utilizes the dynamic changes of insect populations over time to more accurately assess the activity and state changes of insect populations, thereby improving the predictive ability of insect population growth and development. Attached Figure Description

[0031] Figure 1 This is a system framework diagram of the present invention. Detailed Implementation

[0032] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Example:

[0034] Please see the appendix Figure 1 This invention provides a black soldier fly larvae age identification and intelligent feeding linkage system for treating kitchen waste, comprising a data acquisition module, a data processing and activity index calculation module, an larvae age identification module, a feeding linkage control module, and a cloud management and self-learning module. The modules are coupled through a data and control closed loop to form an integrated dynamic intelligent management and control system.

[0035] The data acquisition module includes an image acquisition unit and an environmental factor acquisition unit;

[0036] The image acquisition unit uses a high-definition industrial camera or depth camera, configured with appropriate lenses and installation angles, to periodically acquire high-definition images or videos of the insect enclosure area. This image data is used not only for insect identification but also to analyze the insect swarm's movement, density distribution, and physical characteristics. It is recommended to set the image acquisition frequency to 5-10 frames per minute to ensure the movement trajectory of the insect swarm can be captured.

[0037] The environmental factor acquisition unit consists of a temperature sensor, a humidity sensor, a light intensity detector, and a gas detection node. It collects various environmental indicators such as oxygen, carbon dioxide, ammonia, and nitrogen in real time, forming multi-dimensional input parameters.

[0038] In one embodiment,

[0039] Temperature sensor: A high-precision temperature sensor (such as DS18B20) is used to ensure the reliability of temperature data.

[0040] Humidity sensor: Uses DHT22 or similar model to measure changes in ambient humidity.

[0041] Light intensity detector: The light intensity in the insect tank is detected by a photoresistor or photodiode.

[0042] Gas detection nodes include oxygen sensors (such as MH-Z19), carbon dioxide sensors (such as MH-Z14), and ammonia and nitrogen sensors (such as MiCS-5524) to monitor changes in the concentration of ambient gases in real time. All sensors should be calibrated regularly and data should be transmitted to the data processing module through standardized protocols to ensure the accuracy and stability of the data.

[0043] The data processing and activity index calculation module first normalizes the environmental data, extracts time-series features, and calculates the activity index (Acti). To ensure the accuracy of data processing and the real-time response capability of the system, in one embodiment, it specifically includes:

[0044] Environmental data normalization: Data from various sensors is normalized to ensure that data from different sources can be compared under the same standard, guaranteeing consistency of input data during model training. The normalization formula is:

[0045]

[0046] in, For sensor data, and These are the mean and standard deviation of the data, respectively.

[0047] Temporal Feature Extraction: An improved Long Short-Term Memory (LSTM) network was used to process temporal variation data of multiple environmental factors. LSTM can capture time dependencies and adapt to changes in environmental factors during the growth of the insect population. To cope with the high volatility of the black soldier fly's growth environment, the gate structure of the LSTM model was optimized to ensure its response speed and prediction accuracy to environmental data.

[0048] Multidimensional Feature Modeling: A Transformer encoder is introduced to perform global dependency modeling on multidimensional environmental data. This encoder can effectively extract global information from the data and generate environmental temporal feature vectors. In addition, video data undergoes optical flow calculation and density change rate analysis to generate motion feature vectors (MotionVec).

[0049] Activity Index (Acti) Calculation: The activity index (Acti) is calculated by combining environmental characteristics (EnvVec) and movement characteristics (MotionVec). This index comprehensively reflects the activity status and environmental adaptability of the insect population. The formula is:

[0050]

[0051] in, For combined functions, weighted averaging, neural networks, and other methods can be used depending on the specific circumstances.

[0052] The insect age identification module introduces the activity index Acti as an additional attention guidance signal based on the real-time detection Transformer model RT-DETR, optimizing the attention distribution of the detection model and thus improving the identification accuracy of insects in complex environments. The improved HiDETR model outputs the insect age distribution (e.g., the proportion of 1st to 5th instars), individual counts, and health status labels, specifically including:

[0053] Real-time detection and model optimization: Insect detection based on the Transformer model RT-DETR. RT-DETR is a real-time target detection model that can efficiently extract the spatial features of insects in complex environments. Building upon this, the activity index (Acti) is introduced as an additional attention-guiding signal to optimize the model's attention distribution and improve the accuracy of insect recognition.

[0054] Improved HiDETR Model: The HiDETR model further enhances the accuracy of insect detection. This model is optimized based on the traditional DETR model and can better handle diverse features in dynamic scenes. The insect population age distribution (e.g., the proportion of 1st to 5th instars), individual count, and health status labels are output and will be passed to the subsequent control module.

[0055] Model performance optimization: Data augmentation techniques, such as mirroring, rotation, and color transformation, are used to increase the diversity of training data and improve the robustness and generalization ability of the model.

[0056] The improved HiDETR model classifies worms by age using a classification network. The network outputs a corresponding age category for each worm (e.g., 1st instar, 2nd instar, etc.). These categories are based on the worm's morphological features and learned patterns. In addition to the classification head, the model may also use a regression network to output the worm's age (e.g., the proportion between 1st and 5th instars or consecutive age values), which is necessary for more accurate age determination. The output of the regression model is processed through a linear regression layer to output a specific predicted age value, including:

[0057] The input image is preprocessed, and a pre-trained convolutional neural network (e.g., ResNet) is used to extract low-level features such as texture and edges to obtain feature maps. These feature maps are then fed into a Transformer for global dependency modeling, capturing complex features of the insect, such as its shape, density, and size. Through the HiDETR model's classification network, each insect is assigned an age category. The output of the classification head is processed using a Softmax activation function to obtain the probability distribution of the insect belonging to each age group.

[0058]

[0059] in, It is the classification weight. It is a characteristic feature of the insect body. This is the bias term; the Softmax function outputs the probability of each insect age category.

[0060] At the same time, the regression network outputs a predicted age value for each worm, representing the worm's specific age (e.g., a continuous value from 1 to 5):

[0061] in, It is a regression weight. It is a regression bias;

[0062] Loss function and training:

[0063] Classification loss: The loss for the classification task is calculated using the cross-entropy loss function.

[0064]

[0065] in, It is the true category label of the insect body. It is the probability predicted by the model;

[0066] Regression Loss: Calculate the loss for the regression task using smoothed L1 loss.

[0067]

[0068] in, It is the insect age predicted by the model. It is the actual insect age label;

[0069] Total loss:

[0070]

[0071] in, These are weighting coefficients used to balance classification loss and regression loss;

[0072] The input training image is processed by a convolutional neural network and a Transformer module to extract features. The loss is calculated based on the output of the classification and regression tasks. The backpropagation algorithm is used to optimize the model weights. These prediction results are then passed to the subsequent feeding control module to provide the swarm with a precise feeding strategy.

[0073] The feeding linkage control module, based on the insect age prediction results and activity index, queries a feeding strategy table stored locally or in the cloud, and automatically controls the operation of the feeding equipment. Specifically, this includes: driving a screw feeder to precisely add solid feed, or adjusting the supply of liquid nutrients via a liquid feed pump. This module has a feedback loop; after feeding is completed, it uploads the actual insect swarm response and output data to the cloud to correct the strategy. The specific implementation method is as follows:

[0074] Feeding equipment control: Based on insect age prediction results and activity index, the system automatically selects a suitable feeding strategy and adjusts the operation of the feeding equipment through the control system. It drives the screw feeder to precisely add solid feed or regulates the supply of liquid nutrients through a liquid feed pump. A closed-loop control system is formed between the control module and sensors to ensure the accuracy of the feeding process.

[0075] Feedback mechanism and strategy adjustment: After feeding, real-time response data of the insect population (such as growth rate, health status, etc.) is collected and uploaded to the cloud. By analyzing this data, the feeding strategy is adjusted and optimized in real time.

[0076] The cloud-based management and self-learning module aggregates data on insect age, activity, feeding amount, and environmental factors from each batch to the cloud, forming a historical database. This long-term data is used to retrain and update the weights of the HiEnv-Transformer and HiDETR models, thereby achieving adaptive optimization. The cloud also allows for strategy deduction, remotely distributing optimal feeding parameters, and supports cross-regional and cross-batch collaborative management, specifically including:

[0077] Data aggregation and analysis: Data on insect age, activity, feeding amount, and environmental factors from each batch are aggregated to the cloud to form a historical database. Through big data analysis, insect population growth patterns are extracted, providing a basis for future insect age prediction and feeding strategy optimization.

[0078] Adaptive optimization: Based on historical data, the HiEnv-Transformer and HiDETR models are retrained using long-term learning to update weights and improve prediction accuracy. The system can self-adjust based on historical data of the insect swarm, adapting to the needs of the swarm under different environmental conditions.

[0079] Cross-regional collaborative management: The cloud system supports cross-regional and cross-batch insect swarm management. It can deduce the optimal feeding strategy based on the environmental characteristics of different regions and achieve collaborative optimization between regions through remote distribution.

[0080] Through the organic integration of these modules, this system enables a fully automated and dynamically adjustable black soldier fly biological treatment process, improving the efficiency of food waste treatment and effectively reducing human intervention. The system is highly adaptable and can intelligently optimize itself according to different environmental and insect population needs.

[0081] A method for treating kitchen waste by linking black soldier fly age identification with intelligent feeding includes the following steps:

[0082] S1. Hardware Installation. Install the image acquisition unit (high-definition industrial camera or depth camera) in the insect tank area, ensuring the camera can cover the entire tank and periodically capture information such as the movement and density distribution of the insect swarm. Simultaneously, install temperature and humidity sensors, light intensity detectors, gas sensors, etc., in appropriate locations to ensure accurate collection of environmental factor data such as temperature, humidity, oxygen, and carbon dioxide.

[0083] S2. Data Acquisition and Transmission. Video data of the insect tank area is acquired periodically from the image acquisition unit to record information such as the movement status, density distribution, and physical characteristics of the insect swarm. The environmental factor acquisition unit will collect multiple environmental data such as temperature, humidity, and gas concentration in real time and transmit them to the data processing module through a standardized protocol. All collected data will be transmitted to the data processing and activity index calculation module in real time and aggregated to the cloud database through the data transmission channel. The cloud will receive data from various sensors for further analysis and storage.

[0084] S3. Data Preprocessing and Normalization. The data processing module normalizes the collected environmental factor data to ensure consistency across different data sources. The processed data is then analyzed using an LSTM model to extract environmental change trends. The LSTM network calculates the activity index (Acti) based on the environmental and motion data (optical flow and density change rate extracted from video data) of the insect population. By combining the environmental data (EnvVec) and motion data (MotionVec), the activity index is calculated to form an activity score for insect population growth.

[0085] S4. Insect Age Identification and Feeding Control. Based on the improved HiDETR model, the system analyzes the collected insect swarm videos and combines them with the real-time calculated activity index (Acti) to identify the insect age distribution in real time. The system outputs insect age distribution information (such as the proportion of 1st to 5th instars), insect counts, and health status tags. At the same time, based on the insect age prediction results and activity index, the system queries the optimal feeding strategy from the feeding strategy table stored locally or in the cloud. The strategy table contains the feed types and feeding amounts corresponding to different insect age stages. The feeding linkage control module automatically controls the feeding equipment according to the insect age distribution and activity index, driving the screw feeder to accurately add solid feed, or adjusting the supply of liquid nutrients through the liquid feed pump. At the same time, the control system provides real-time feedback data to monitor the feeding effect and ensure uniform nutrient intake of the insect swarm.

[0086] S5. Data Feedback and Cloud Optimization. After feeding is completed, the system will automatically collect feedback data of the swarm (such as health status, growth rate, output, etc.). This data will be uploaded to the cloud management system for subsequent data analysis and strategy optimization. The cloud management module summarizes historical data and uses long-term data to retrain and optimize the HiEnv-Transformer and HiDETR models. At the same time, based on long-term data feedback, the cloud will deduce and issue the optimal feeding strategy, supporting swarm management across regions and batches. Managers can view the system's operating status, swarm health status, and feeding effect in real time through the cloud platform.

[0087] Through the above steps, the black soldier fly larvae age identification and intelligent feeding linkage system for food waste treatment can automatically and efficiently monitor the larvae population, identify their age, control feeding, and optimize data, achieving efficient management and continuous optimization in the food waste treatment process.

[0088] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A system for the integrated treatment of kitchen waste using black soldier fly larvae age recognition and intelligent feeding, characterized in that: include: The data acquisition module includes an image acquisition unit and an environmental factor acquisition unit. The image acquisition unit is used to acquire image or video data of the insect tank area, and the environmental factor acquisition unit is used to acquire environmental data such as temperature, humidity, light intensity, oxygen, carbon dioxide, ammonia, and nitrogen. The data processing and activity index calculation module is used to normalize environmental data and extract temporal features, and fuse them with motion features obtained from video data to obtain an activity index, which is used to characterize the activity status and environmental adaptability of insect colonies. The insect age recognition module is based on a real-time detection Transformer model and introduces the activity index as an attention guidance signal to output insect age distribution, individual count and health status label; The feeding linkage control module queries the feeding strategy table based on the insect age distribution and activity index, individual count and health status label, and drives the feeding equipment to perform solid or liquid feeding. The cloud management and self-learning module is used to aggregate batch data, retrain the model and perform strategy inference, and distribute feeding parameters. The modules are coupled together via a closed-loop data and control system.

2. The system for linking black soldier fly larvae age identification and intelligent feeding in the treatment of kitchen waste according to claim 1, characterized in that: The environmental factor acquisition unit includes a temperature sensor, a humidity sensor, a light intensity detector, and a gas detection node for detecting oxygen, carbon dioxide, ammonia, and nitrogen. The environmental factor acquisition unit supports periodic calibration and transmits data to the data processing and activity index calculation module through a standardized communication protocol.

3. The system for linking black soldier fly larvae age identification and intelligent feeding in the treatment of kitchen waste according to claim 1, characterized in that: The data processing and activity index calculation module performs normalization on environmental data based on mean and standard deviation to enable comparison of multi-source data on a unified scale, and performs time synchronization and missing data compensation on data streams from different sources to ensure the stability of subsequent modeling.

4. The system for linking black soldier fly larvae age identification and intelligent feeding in the treatment of kitchen waste according to claim 1, characterized in that: The temporal feature extraction employs an improved long short-term memory network to model multiple environmental factor sequences and introduces a Transformer encoder to model global dependencies, thereby obtaining environmental temporal feature vectors.

5. The system for linking black soldier fly larvae age identification and intelligent feeding in the treatment of kitchen waste according to claim 1, characterized in that: The motion features are obtained by analyzing the optical flow and density change rate of video data to form a motion feature vector. The activity index is formed by weighted fusion or neural network fusion of the environmental time-series feature vector and the motion feature vector to quantify the activity level of the insect population.

6. The system for linking black soldier fly larvae age identification and intelligent feeding in the treatment of kitchen waste according to claim 1, characterized in that: The insect age recognition module includes a real-time detection network based on RT-DETR and an improved HiDETR model. The activity index is used as an additional attention guidance signal to optimize the attention distribution and improve the accuracy of insect detection and insect age classification in complex scenarios. The improved HiDETR model includes a classification network for outputting insect age categories and a regression network for outputting continuous insect age prediction values.

7. The system for linking black soldier fly larvae age identification and intelligent feeding in the treatment of kitchen waste according to claim 1, characterized in that: The feeding strategy table provides the correspondence between feed types and feeding amounts based on the insect age stage and activity index, and supports the linkage adjustment of the ratio of solid feed and liquid nutrients according to the distribution of insect age. The feeding linkage control module collects response data while executing feeding and uses it as a closed-loop feedback input.

8. The black soldier fly larvae age identification and intelligent feeding linkage system for processing kitchen waste according to claim 1, characterized in that: The cloud management and self-learning module aggregates historical data on insect age, activity, feeding amount, and environmental factors over a long period of time, and retrains and updates the weights of the HiEnv-Transformer and HiDETR models to form an adaptive insect age identification and feeding strategy. The cloud management and self-learning module supports cross-regional and cross-batch strategy deduction and remote parameter distribution.

9. A system for the integrated treatment of kitchen waste by black soldier fly larvae age identification and intelligent feeding as described in claim 1, characterized in that: The feeding linkage control module includes a screw feeder for solid feeding and a liquid feed pump for liquid feeding. The feeding linkage control module, the data acquisition module and the insect age recognition module form a closed-loop control to dynamically adjust the feeding intensity, timing and ratio based on real-time recognition results and activity index during the feeding process.

10. A method for the integrated treatment of kitchen waste using black soldier fly larvae age identification and intelligent feeding, applied to the system described in any one of claims 1 to 9, characterized in that: Includes the following steps: S1. Hardware installation and configuration: Install the image acquisition unit, environmental factor acquisition unit and related sensors to ensure that the image acquisition unit can cover the insect tank area and regularly capture the movement and density distribution data of the insect population, and that the environmental factor acquisition unit can accurately acquire multi-dimensional environmental data such as temperature, humidity, light intensity and gas concentration. S2. Data Acquisition and Real-time Transmission: Video data from the image acquisition unit and environmental data from the environmental factor acquisition unit are acquired periodically. The acquired data is transmitted in real time to the data processing and activity index calculation module through a standardized protocol and uploaded to the cloud database for subsequent storage and analysis. S3. Data Preprocessing and Analysis: The collected environmental data is normalized and a time series model is used to provide a basis for subsequent insect age identification and feeding strategies. S4. Insect Age Identification and Feeding Control: Based on real-time calculated activity index and image data, the improved HiDETR model is used to identify the insect age of the insect population. The optimal feeding strategy is queried from the feeding strategy table according to the insect age distribution and activity index. The operation of the feeding equipment is controlled through the feeding linkage control module to achieve precise feeding. S5. Data Feedback and Optimization: After feeding is completed, feedback data of the insect swarm is collected in real time and uploaded to the cloud for analysis. The model is retrained and optimized using historical data. The cloud supports cross-regional and cross-batch insect swarm management and strategy simulation, and dynamically adjusts feeding parameters.

Images