Intelligent classification treatment system and method for building solid waste

Abstract

The invention discloses an intelligent classification processing system and method for building solid waste, and relates to the technical field of intelligent classification, and the method comprises the steps: traversing a target region, collecting multi-dimensional data of the building solid waste, and determining a to-be-classified data set; determining a plurality of classification labels to construct an intelligent classification model, and synchronizing the to-be-classified data set to the intelligent classification model to obtain a classification decision; the classification decision is simulated and executed to conduct classification processing monitoring on the building solid waste, and operation feedback data is generated; optimizing the intelligent classification model based on the operation feedback data to obtain an intelligent optimization classification model, and synchronizing the to-be-classified data set to the intelligent optimization classification model to obtain a classification optimization decision; and intelligent classification treatment is conducted on the building solid waste. The technical problems that in the prior art, when the building solid waste is classified and treated, efficiency is low, and accuracy is not high are solved, and the technical effect of improving the accuracy and efficiency of building solid waste classification is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent classification, and particularly relates to an intelligent classification and treatment system and method for construction solid waste. Background Art

[0002] In the modern construction industry, with the acceleration of the urbanization process and the continuous expansion of the construction scale, the generation amount of construction solid waste is also increasing rapidly. These construction solid wastes mainly include concrete, bricks, metals, plastics, wood, etc., with a wide variety of types and complex components, usually mixed together. If not properly treated, it will not only occupy a large amount of land resources, but also cause serious pollution to the environment. The traditional treatment methods for construction solid waste mainly rely on manual classification and simple mechanical sorting, with low efficiency and low accuracy, and it is difficult to meet the needs of large-scale waste treatment. Summary of the Invention

[0003] This application provides an intelligent classification and treatment system and method for construction solid waste, which is used to solve the technical problems of low efficiency and low accuracy existing in the prior art when classifying and treating construction solid waste.

[0004] In view of the above problems, this application provides an intelligent classification and treatment system and method for construction solid waste.

[0005] In the first aspect of this application, an intelligent classification and treatment system for construction solid waste is provided. The system includes: A data traversal module, which traverses the target area to collect multi-dimensional data of construction solid waste and determines a data set to be classified; a classification label determination module, which determines multiple classification labels, constructs an intelligent classification model using the multiple classification labels, synchronizes the data set to be classified to the intelligent classification model to obtain a classification decision; a classification processing simulation module, which simulates the execution of the classification decision to monitor the classification and treatment of construction solid waste and generates operation feedback data; a classification optimization module, which optimizes the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronizes the data set to be classified to the intelligent optimized classification model, and obtains a classification optimization decision; a classification processing module, which executes the classification optimization decision to perform intelligent classification and treatment on construction solid waste.

[0006] In the second aspect of this application, an intelligent classification and treatment method for construction solid waste is provided. The method includes: Traverse the target area to collect multi-dimensional data of construction solid waste, and determine the dataset to be classified; determine multiple classification labels, use the multiple classification labels to construct an intelligent classification model, synchronize the dataset to be classified to the intelligent classification model to obtain a classification decision; simulate the execution of the classification decision to monitor the classification process of construction solid waste, and generate operation feedback data; optimize the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronize the dataset to be classified to the intelligent optimized classification model to obtain a classification optimization decision; execute the classification optimization decision to perform intelligent classification processing on construction solid waste.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application traverses the target area to collect multi-dimensional data of construction solid waste, determines the dataset to be classified; determines multiple classification labels, uses the multiple classification labels to construct an intelligent classification model, synchronizes the dataset to be classified to the intelligent classification model to obtain a classification decision; simulates the execution of the classification decision to monitor the classification process of construction solid waste, and generates operation feedback data; optimizes the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronizes the dataset to be classified to the intelligent optimized classification model to obtain a classification optimization decision; executes the classification optimization decision to perform intelligent classification processing on construction solid waste. This invention solves the technical problems of low efficiency and low accuracy in the classification process of construction solid waste in the prior art. Through multi-dimensional data collection, classification label construction, intelligent classification model training, simulation execution monitoring, feedback optimization, and finally executing the classification optimization decision, the technical effect of improving the accuracy and efficiency of the classification of construction solid waste is achieved. Description of the Drawings

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0009] Figure 1 It is a schematic structural diagram of an intelligent classification processing system for construction solid waste provided by an embodiment of this application; Figure 2 It is a schematic flowchart of an intelligent classification processing method for construction solid waste provided by an embodiment of this application.

[0010] Description of the reference numerals: data traversal module 11, classification label determination module 12, classification processing simulation module 13, classification optimization module 14, classification processing module 15. Detailed Embodiments

[0011] This application provides an intelligent classification and processing system and method for construction solid waste, aiming to solve the technical problems of low efficiency and low accuracy in the classification and processing of construction solid waste in the prior art. Through multi-dimensional data collection, classification label construction, intelligent classification model training, simulation execution monitoring, feedback optimization, and final execution classification optimization decision-making, the technical effect of improving the accuracy and efficiency of the classification of construction solid waste is achieved.

[0012] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0013] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0014] Embodiment 1, as Figure 1 shown, the embodiment of the present application provides an intelligent classification and processing system for construction solid waste, and the system includes: A data traversal module 11, and the data traversal module 11 traverses the target area to collect multi-dimensional data of construction solid waste and determines the data set to be classified.

[0015] In the embodiment of the present application, the data traversal module traverses the target area, that is, uses sensors or other detection devices to systematically scan the target area to obtain multi-dimensional data including the spatial position, material type, volume, density, shape, color, etc. of the construction solid waste. The target area refers to the specific physical area where solid waste classification needs to be carried out, such as a building demolition site or a construction waste storage yard.

[0016] Through the above process, the target area collects multi-dimensional data of construction solid waste, and the obtained multi-dimensional data is used as the data set to be classified.

[0017] A classification label determination module 12, and the classification label determination module 12 determines a plurality of classification labels, constructs an intelligent classification model using the plurality of classification labels, and synchronizes the data set to be classified to the intelligent classification model to obtain a classification decision.

[0018] In the embodiment of the present application, first, the dataset to be classified is normalized to ensure the comparability between data of each dimension. Then, multi-dimensional feature extraction is performed on each piece of data in the dataset to be classified. After the multi-dimensional feature extraction is completed, the classification label determination module performs feature annotation and clustering analysis. Labels are assigned to each feature dimension through an annotation algorithm, and the labeled data is clustered through a clustering analysis algorithm to generate multiple classification labels.

[0019] After that, an intelligent classification model is constructed using multiple classification labels. During the construction of the intelligent classification model, each label corresponds to a reinforcement learning agent. Through continuous learning and training, each classification label is verified and optimized, and finally an intelligent classification model capable of accurately classifying according to multi-dimensional features is generated. Finally, the dataset to be classified is synchronized to the model for classification decision-making.

[0020] Further, in the system provided by the embodiment of the application, the classification label determination module 12 is further configured to: Perform data normalization processing based on the dataset to be classified to obtain multiple standard data to be classified; traverse the multiple standard data to be classified for multi-dimensional feature extraction to determine multi-dimensional data features; perform feature annotation on the multiple standard data to be classified according to the multi-dimensional data features to construct a multi-feature network; traverse the multi-feature network for clustering analysis to determine the multiple classification labels.

[0021] In the embodiment of the present application, first, through data normalization processing, the dataset to be classified is preprocessed. Using the Min-Max normalization method, the numerical value of each feature is scaled to the range of [0, 1] to eliminate the dimensional difference between features and ensure that each feature has a consistent numerical scale in subsequent processing. Multiple standard data to be classified are obtained through this process.

[0022] Next, multi-dimensional feature extraction is performed on the multiple standard data to be classified. Using the principal component analysis method, the covariance matrix of the data is calculated, and several principal components that can explain the largest variance of the data are extracted. After completion, a multi-dimensional feature dataset composed of multiple principal components is obtained.

[0023] After that, in the feature annotation stage, the support vector machine algorithm is adopted. The goal of this stage is to assign corresponding classification labels to each data sample based on the extracted multi-dimensional features. The SVM is pre-trained. By learning the classification boundaries of the samples, new samples are classified into the corresponding categories. For example, samples are labeled as "metal waste" or "concrete waste" according to features such as volume, weight, and material. After this stage is completed, an annotated data set with clear classification labels is generated. Then, in the stage of constructing a multi-feature network, a feature node network is constructed according to the correlation between features. Each multi-dimensional feature is regarded as a node, and these nodes are connected to each other through their correlations, forming a multi-feature network. Specifically, each extracted feature is regarded as a node, and the connections between nodes are based on the correlation between features, such as measured by the Pearson correlation coefficient.

[0024] Finally, in the clustering analysis stage, the K-Means clustering algorithm is adopted. Based on the data in the multi-feature network, the K-Means algorithm iteratively adjusts the centroids of the clusters to cluster samples with similar features into the same cluster. Each cluster represents a specific classification label, such as "metal category" or "plastic category". Through this clustering process, multiple classification labels are generated.

[0025] Furthermore, in the system provided by the application embodiment, the classification label determination module 12 is further configured to: Set a multi-label task environment. Based on the multi-label task environment, traverse the multiple classification labels to perform integrated reinforcement learning to obtain multiple reinforcement learners. The multiple reinforcement learners have a corresponding relationship with the multiple classification labels; perform supervised training based on the multiple reinforcement learners to generate a first training result, and verify the first training result to generate a data verification token; integrate the multiple reinforcement learners according to the data verification token to construct the intelligent classification model.

[0026] In the embodiment of the present application, first, in the stage of setting a multi-label task environment, a multi-label classification problem framework, such as ML-KNN, is adopted to handle complex multi-label classification tasks. Each construction waste sample may belong to multiple categories at the same time, such as "metal" and "recyclable". Therefore, first, the classification task is decomposed into multiple binary classification or multi-classification problems, and a multi-label task environment is constructed. In this environment, all label combinations are traversed to form multiple independent sub-tasks, which allow the system to process different labels separately.

[0027] Next, in the reinforcement learning phase, the Q-Learning algorithm is used to train the classifier. Q-Learning is a value function-based reinforcement learning method, and its goal is to maximize the cumulative reward by learning the optimal policy. In this phase, a reinforcement learner is assigned to each label task, and these learners use Q-Learning to learn how to make classification decisions in a multi-label task environment. Specifically, Q-Learning updates the Q-value by repeatedly balancing exploration and exploitation. The Q-value represents the maximum expected reward that can be obtained by taking a certain action in a given state. In multiple iterations, the classification strategies of the learners are continuously adjusted to optimize their performance in different label tasks. Finally, multiple reinforcement learners that can execute the optimal classification strategy are trained.

[0028] After completing the reinforcement learning, the supervised training phase is entered. This phase aims to further optimize the learners trained by Q-Learning through supervised learning methods. Using supervised learning algorithms such as support vector machines or random forests, the already labeled data is compared with the preliminary classification results of the learners, and the classification accuracy is improved by adjusting the model parameters. In this phase, supervised learning uses the already labeled training data to further optimize the output of Q-Learning, enabling the learners to perform well in higher-precision classification tasks. After completing this phase, the first training result is generated.

[0029] Then, the validation and data validation token generation phase is entered. In this phase, cross-validation techniques are used to verify the performance of the first training result. Cross-validation divides the dataset into several folds, and each time a part of it is used as the validation set, and the rest is used as the training set for multiple training and validations to evaluate the generalization ability and robustness of the model. By calculating performance metrics such as the accuracy and F1 score of the model, the performance of the model is evaluated. If the model passes the preset performance criteria, a data validation token is generated for it, indicating that the model has passed the validation and is eligible to participate in the integration. After completing this phase, multiple data validation tokens are generated, verifying the effectiveness of the reinforcement learners and confirming that they can participate in the subsequent integration steps.

[0030] Finally, in the integrated model construction phase, ensemble learning techniques such as Bagging are used to integrate multiple verified reinforcement learners. Bagging trains multiple independent learners and averages their prediction results. During the integration process, the outputs are weighted and averaged according to the performance of each learner, thereby improving the robustness and accuracy of the final decision. Through these integration methods, the construction of the intelligent classification model is finally completed.

[0031] Furthermore, in the system provided by the application embodiment, the classification label determination module 12 is further used for: Based on the multiple reinforcement learners, sequentially determine whether the dataset to be classified conforms to the target classification label; extract the data to be classified that conforms to the target classification label, generate a positive reward instruction, and identify the data to be classified that conforms to the target classification label through the positive reward instruction to obtain a first classification result; extract the data to be classified that does not conform to the target classification label, generate a negative reward instruction, and identify the data to be classified that does not conform to the target classification label through the negative reward instruction to obtain a second classification result; perform weighted calculation based on the first classification result and the second classification result to generate multiple weight coefficients; formulate the classification decision according to the multiple weight coefficients in combination with the first classification result and the second classification result.

[0032] In the embodiment of the present application, first, based on multiple reinforcement learners, determine whether each sample in the dataset to be classified conforms to the target classification label. Each reinforcement learner focuses on a specific label and makes a classification decision according to the characteristics of the sample. By sequentially judging whether the sample conforms to a specific target label through multiple reinforcement learners, after completing this process, a preliminary classification judgment result is generated.

[0033] Next, in the stage of generating the positive reward instruction and identifying the first classification result, generate a positive reward instruction for the sample that conforms to the target classification label. The positive reward is set to +1, indicating that the sample with correct classification should reinforce this classification action. Identify the correctly classified samples as the first classification result and record them as positive classification samples that conform to the target classification label. At the same time, in the stage of generating the negative reward instruction and identifying the second classification result, generate a negative reward instruction for the sample that does not conform to the target classification label. The negative reward, as a feedback mechanism for correcting the classification strategy, is set to -1, informing the system that this classification action is incorrect and the classification strategy needs to be adjusted in the future. Identify these misclassified samples as the second classification result, and these samples are recorded as negative classification samples that do not conform to the target label.

[0034] After that, perform weighted calculation based on the first classification result and the second classification result. Specifically, perform weighted calculation on the first classification result and the second classification result according to the classification results. The determination of the weights depends on the importance in the task requirements and the difficulty of the classification task. When the importance of some classification labels in the overall task is relatively high, such as some high-value waste categories, allocate higher weights to the results of these classification labels according to the task requirements. If some classification labels are difficult to be accurately classified, allocate higher weights to these more difficult-to-classify labels. Based on the above factors, generate multiple weight coefficients through the weighted average algorithm.

[0035] Finally, in the stage of making classification decisions by combining weight coefficients, a weighted voting mechanism is used to integrate the first classification result and the second classification result. According to the weight coefficients generated previously, weighted voting is performed on the positive and negative classification results. The weighted voting mechanism allows the system to balance the impacts of different classification results in the final classification decision, ensuring a more accurate classification decision. Through this weighted voting, the final classification decision is generated.

[0036] The classification processing simulation module 13 simulates the execution of the classification decision to monitor the classification of construction solid waste and generates operation feedback data.

[0037] In the embodiment of the present application, first, the classification decision is simulated and executed, and then the classification processing monitoring stage is carried out. During this process, the classification processing simulation module monitors the classification process in real time, records the performance of the classification processing, and captures various data during the classification process, such as classification speed, classification accuracy, waste allocation efficiency for each category, etc. Operation feedback data is generated through this process.

[0038] Furthermore, in the system provided by the embodiment of the application, the classification processing simulation module 13 is further configured to: Simulate the execution of the classification decision based on the dataset to be classified to generate classification control instructions; activate the sorting machinery according to the classification control instructions to determine multiple areas to be allocated; perform classification processing monitoring on the construction solid waste according to the multiple areas to be allocated to obtain a classification monitoring dataset, where the classification monitoring dataset includes classification speed monitoring data, classification correct monitoring data, and classification error monitoring data; calculate the classification efficiency of the classification decision based on the classification speed monitoring data, the classification correct monitoring data, and the classification error monitoring data, and construct a classification operation log; add the classification operation log to the operation feedback data.

[0039] In the embodiment of the present application, first, classification control instructions are generated by simulating the execution of the classification decision based on the dataset to be classified. Specifically, first, a virtual classification environment is created, which is used to simulate the classification process of construction solid waste in actual operation. The classification processing simulation module simulates the execution of the classification decision based on the virtual data in the dataset to be classified. The intelligent classification model generates classification control instructions in the simulation environment according to the characteristics in the dataset to be classified, such as volume, weight, material, etc. These instructions specify the category attribution of the waste and simulate the execution effects of the instructions.

[0040] Next, according to the classification control instructions generated by the simulation execution, the sorting machinery is further activated in the virtual environment. These sorting machinery models are used to simulate the operation of automated sorting equipment, and classify the virtual construction solid waste according to the classification control instructions. The virtual waste is assigned to multiple areas to be assigned, which are the placement locations after the construction waste is classified. Each area to be assigned corresponds to a specific classification result. For example, metal waste is assigned to the metal area, and plastic waste is assigned to the plastic area. Then, in the simulation environment, the classification processing simulation module monitors the entire classification process through virtual sensors and records various data to form a classification monitoring data set. The classification monitoring data set includes classification speed monitoring data, classification correct monitoring data, and classification error monitoring data. Among them, the classification speed monitoring data records the operation speed of the virtual sorting machinery; the classification correct monitoring data monitors the accuracy of the classification; and the classification error monitoring data records the errors in the classification.

[0041] Based on the classification speed, classification accuracy rate, and classification error rate in the classification monitoring data set, the classification efficiency is calculated. The calculation of the classification efficiency takes into account the rapidity and accuracy of the classification, and reflects the comprehensive performance of the intelligent classification model in the virtual environment. For example, if the classification speed is fast and the classification accuracy rate is high, the classification efficiency will be high; while when the classification error rate is high, the classification efficiency will be affected. After completing the efficiency calculation, a classification operation log is constructed to record the execution of the classification control instructions, the performance of the sorting machinery, and the results of the classification efficiency. Finally, the constructed classification operation log is added to the operation feedback data.

[0042] The classification optimization module 14, based on the operation feedback data, optimizes the intelligent classification model to obtain an intelligent optimized classification model, synchronizes the data set to be classified to the intelligent optimized classification model, and obtains a classification optimization decision.

[0043] In the embodiment of the present application, first, the intelligent classification model is optimized through the operation feedback data. Specifically, the operation feedback data is analyzed to identify the deficiencies of the intelligent classification model in actual operations, such as a relatively high misclassification rate for certain categories or insufficient sorting speed. Based on these feedbacks, the classification optimization module adopts an adaptive learning algorithm to automatically adjust the weights, parameters, and strategies of the intelligent classification model to obtain an intelligent optimized classification model.

[0044] Then, the data set to be classified is synchronized to the intelligent optimized classification model, and the optimized model is used to infer the data set and generate a classification optimization decision.

[0045] Furthermore, in the system provided by the embodiment of the application, the classification optimization module 14 is further used for: Perform data mining on the operation feedback data based on the classified operation logs to obtain multiple classification association rules; analyze the operation feedback data according to the multiple classification association rules to obtain an operation analysis result; perform classification evaluation according to the operation analysis result to obtain an operation classification evaluation result; traverse the operation classification evaluation result to capture anomalies and generate anomaly classification information; formulate a model adjustment strategy based on the anomaly classification information, execute the model adjustment strategy to retrain the intelligent classification model, and obtain a second training result; verify the data validity of the second training result, and generate the intelligent optimized classification model when the verification passes.

[0046] In the embodiment of the present application, first, an association rule mining algorithm, such as the Apriori algorithm, is used to perform data mining on the classified operation logs. The classified operation logs record various data in the classification process, such as classification accuracy, classification error rate, classification speed, etc. Through the association rule mining algorithm, multiple classification association rules are extracted from the logs, and these rules reveal the potential relationships between different classification behaviors. For example, a pattern where the classification error rate of a certain type of waste is relatively high under specific conditions is identified.

[0047] Next, the extracted classification association rules are used to analyze the operation feedback data. The operation feedback data includes the classification performance after the classification task is completed, such as classification error rate, classification accuracy, and classification efficiency. By combining these classification association rules with the feedback data, the bottlenecks or problem points that may occur in the classification task of the system are identified. For example, it is found that under certain rules, the classification efficiency of a specific category is relatively low or the classification error rate is relatively high. Through this association rule-based analysis, the weaknesses and improvement directions of the intelligent classification model are understood. After the analysis is completed, an operation analysis result is generated, revealing the performance of the intelligent classification model in actual applications and the areas that need to be optimized.

[0048] According to the operation analysis result, a comprehensive evaluation of the current intelligent classification model is performed through classification performance evaluation tools, such as ROC curve analysis or confusion matrix analysis. These evaluation methods can detail the accuracy and error rate of the intelligent classification model in different classification tasks. For example, the confusion matrix shows the specific performance of the model when classifying different waste categories, including the number of correct and incorrect classifications. Through the evaluation, the advantages and disadvantages of the intelligent classification model are identified, and the specific parts that need further optimization are clarified. After the evaluation is completed, an operation classification evaluation result is generated, providing a basis for subsequent model optimization.

[0049] During the classification evaluation process, the running classification evaluation results are traversed through anomaly detection algorithms such as Isolation Forest or Local Outlier Factor to capture abnormal situations that occur during the classification process. These anomalies include cases where the classification error rate significantly increases or the classification speed significantly decreases under specific conditions. These anomalies are captured and anomaly classification information is generated, which details the occurrence conditions of the anomalies and their potential impacts.

[0050] Next, targeted model adjustment strategies are formulated based on the anomaly classification information. These adjustment strategies involve optimizing the model's parameters, adjusting the learning rate, balancing class weights, or increasing the diversity of the training dataset. Then, the intelligent classification model is retrained through an adaptive learning algorithm such as Stochastic Gradient Descent. The adaptive learning algorithm dynamically adjusts the model's parameters to better handle complex classification tasks, reducing the error rate and improving the classification efficiency. After completing the retraining, a second training result is generated, showing that the performance of the model in the classification task has been improved after adjustment.

[0051] Finally, cross-validation and test set evaluation are used to verify the effectiveness of the retrained model. Cross-validation tests the performance of the model under different situations by splitting multiple datasets to ensure that the model can maintain good generalization ability when dealing with new data. At the same time, the test set is used to evaluate the performance of the model in actual applications to verify whether it meets the expected classification accuracy and stability. Through these verification steps, an optimized intelligent classification model is finally confirmed and generated.

[0052] Furthermore, in the system provided by the application embodiment, the classification optimization module 14 is further configured to: Perform anomaly tracing based on the anomaly classification information to determine the anomaly data source, where the anomaly data source has an anomaly source identifier; determine the anomaly pattern according to the anomaly source identifier, perform adjustment analysis according to the anomaly pattern to determine the parameter adjustment direction; and formulate the model adjustment strategy of the intelligent classification model based on the parameter adjustment direction in combination with the anomaly data source.

[0053] In the embodiment of the present application, first, anomaly tracing is performed based on the anomaly classification information, and a causal inference algorithm is used to identify the specific data source that causes the classification error. The causal inference algorithm analyzes the causal relationships between different features and variables in the model, gradually traces the paths of these abnormal data being processed during the classification process, and thus determines which input data and features cause the classification anomaly. The input data that causes the problem is called the anomaly data source and is marked with a specific anomaly source identifier, which clearly marks the reason for the anomaly, such as a specific combination of input variables, data features under abnormal conditions, etc.

[0054] Next, according to the anomaly source identifier, the DBSCAN clustering algorithm is used for clustering analysis. By using DBSCAN, high-density regions in the anomaly data source are identified, that is, those anomaly data with similar characteristics and behaviors. The anomaly data are classified into the same anomaly pattern, which represents the anomaly behavior jointly caused by multiple data points under specific environments or specific input conditions. For example, it is identified that the classification error rate of construction waste significantly increases under certain temperature or humidity conditions, and the data with these common characteristics are clustered into an anomaly pattern.

[0055] After identifying the anomaly pattern, it enters the adjustment analysis stage, and LASSO regression analysis is used to analyze the anomaly pattern. LASSO regression selects and adjusts the weights of the model's features by introducing L1 regularization. Through the analysis of the anomaly pattern, it is determined which features play a major role in causing classification errors, and the adjustment direction of the parameters to be adjusted is further clarified. For example, the weights of certain features may need to be increased to ensure that the model can better process the anomaly data source; while the weights of other unimportant features need to be reduced to reduce their interference with the model's classification results.

[0056] Finally, combining the previously determined parameter adjustment direction and the anomaly data source, a model adjustment strategy is formulated. The core of this strategy is to adjust the hyperparameters of the model through optimization algorithms such as genetic algorithms or random search, so that the model can better process these anomaly data sources. For example, the genetic algorithm simulates the natural evolution process and screens out the optimal solution during the continuous parameter adjustment process. Through multiple iterations, the parameter combination that can maximize the improvement of the classification effect is found, thereby optimizing the overall performance of the model. Through the above process, the formulation of the model adjustment strategy for the intelligent classification model is completed.

[0057] The classification processing module 15, and the classification processing module 15 performs the classification optimization decision to perform intelligent classification processing on construction solid waste.

[0058] In the embodiment of the present application, the classification processing module receives the classification optimization decision and applies it to the actual waste classification task. At this time, the data of construction solid waste has been preprocessed and classified standardized according to the optimization decision.

[0059] Then the classification processing module uses intelligent sorting equipment, such as automated sorting machinery, to physically execute these decisions. The automated equipment distributes different types of waste to the corresponding processing areas according to the preset classification criteria and optimized parameters. For example, metal waste is identified and conveyed to the metal recycling area, while plastic waste is transported to the plastic classification area. Through this process, the intelligent classification processing of construction solid waste is completed.

[0060] In the embodiment of the present application, in summary, the embodiment of the present application has at least the following technical effects: This application traverses the target area to collect multi-dimensional data of construction solid waste, and determines the data set to be classified; determines multiple classification labels, constructs an intelligent classification model using the multiple classification labels, synchronizes the data set to be classified to the intelligent classification model to obtain a classification decision; simulates the execution of the classification decision to monitor the classification process of construction solid waste, and generates operation feedback data; optimizes the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronizes the data set to be classified to the intelligent optimized classification model, and obtains a classification optimization decision; executes the classification optimization decision to perform intelligent classification processing on construction solid waste. The present invention solves the technical problems of low efficiency and low accuracy in the prior art when classifying and processing construction solid waste. Through multi-dimensional data collection, classification label construction, intelligent classification model training, simulated execution monitoring, feedback optimization, and finally executing the classification optimization decision, the technical effect of improving the accuracy and efficiency of classifying construction solid waste is achieved.

[0061] Embodiment 2, based on the same inventive concept as an intelligent classification processing system for construction solid waste in the foregoing embodiment, as Figure 2 shown, the embodiment of the present application provides an intelligent classification processing method for construction solid waste, and the method includes: Traverse the target area to collect multi-dimensional data of construction solid waste, and determine the data set to be classified; determine multiple classification labels, construct an intelligent classification model using the multiple classification labels, synchronize the data set to be classified to the intelligent classification model to obtain a classification decision; simulate the execution of the classification decision to monitor the classification process of construction solid waste, and generate operation feedback data; optimize the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronize the data set to be classified to the intelligent optimized classification model, and obtain a classification optimization decision; execute the classification optimization decision to perform intelligent classification processing on construction solid waste.

[0062] Further, for the multiple classification labels, the method further includes: Perform data standardization processing on the data set to be classified to obtain multiple standard data to be classified; traverse the multiple standard data to be classified for multi-dimensional feature extraction to determine multi-dimensional data features; perform feature annotation on the multiple standard data to be classified according to the multi-dimensional data features to construct a multi-feature network; traverse the multi-feature network for clustering analysis to determine the multiple classification labels.

[0063] Further, when constructing the intelligent classification model using the multiple classification labels, the method further includes: Set a multi-label task environment. Based on the multi-label task environment, traverse the multiple classification labels to perform integrated reinforcement learning, and obtain multiple reinforcement learners. There is a corresponding relationship between the multiple reinforcement learners and the multiple classification labels; perform supervised training based on the multiple reinforcement learners to generate a first training result, and verify the first training result to generate a data verification token; integrate the multiple reinforcement learners according to the data verification token to construct the intelligent classification model.

[0064] Further, synchronize the dataset to be classified to the intelligent classification model to obtain a classification decision. The method further includes: Based on the multiple reinforcement learners, sequentially determine whether the dataset to be classified conforms to the target classification label; extract the data to be classified that conforms to the target classification label, generate a positive reward instruction, and identify the data to be classified that conforms to the target classification label through the positive reward instruction to obtain a first classification result; extract the data to be classified that does not conform to the target classification label, generate a negative reward instruction, and identify the data to be classified that does not conform to the target classification label through the negative reward instruction to obtain a second classification result; perform weighted calculation based on the first classification result and the second classification result to generate multiple weight coefficients; formulate the classification decision according to the multiple weight coefficients in combination with the first classification result and the second classification result.

[0065] Further, simulate the execution of the classification decision to classify and process and monitor construction solid waste, and generate operation feedback data. The method further includes: Simulate the execution of the classification decision based on the dataset to be classified to generate a classification control instruction; activate the sorting machine according to the classification control instruction to determine multiple areas to be allocated; classify and process and monitor the construction solid waste according to the multiple areas to be allocated to obtain a classification monitoring dataset, where the classification monitoring dataset includes classification speed monitoring data, classification correct monitoring data, and classification error monitoring data; calculate the classification efficiency of the classification decision based on the classification speed monitoring data, the classification correct monitoring data, and the classification error monitoring data, and construct a classification operation log; add the classification operation log to the operation feedback data.

[0066] Further, optimize the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model. The method further includes: Perform data mining on the operation feedback data based on the classified operation logs to obtain multiple classification association rules; analyze the operation feedback data according to the multiple classification association rules to obtain an operation analysis result; perform classification evaluation based on the operation analysis result to obtain an operation classification evaluation result; traverse the operation classification evaluation result to capture anomalies and generate anomaly classification information; formulate a model adjustment strategy based on the anomaly classification information, execute the model adjustment strategy to retrain the intelligent classification model, and obtain a second training result; verify the data validity of the second training result, and generate the intelligent optimized classification model when the verification passes.

[0067] Further, when formulating a model adjustment strategy based on the anomaly classification information, the method further includes: Perform anomaly tracing based on the anomaly classification information to determine an anomaly data source, where the anomaly data source has an anomaly source identifier; determine an anomaly pattern according to the anomaly source identifier, perform adjustment analysis according to the anomaly pattern, and determine a parameter adjustment direction; formulate the model adjustment strategy of the intelligent classification model based on the parameter adjustment direction in combination with the anomaly data source.

[0068] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0070] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An intelligent classification and treatment system for construction solid waste, characterized in that, The system includes: A data traversal module that traverses a target area to collect multi-dimensional data of construction solid waste and determines a dataset to be classified; A classification label determination module that determines multiple classification labels, constructs an intelligent classification model using the multiple classification labels, and synchronizes the dataset to be classified to the intelligent classification model to obtain a classification decision; A classification processing simulation module that simulates the execution of the classification decision to monitor the classification processing of construction solid waste and generates operation feedback data; A classification optimization module that optimizes the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronizes the dataset to be classified to the intelligent optimized classification model, and obtains a classification optimization decision; A classification processing module that executes the classification optimization decision to perform intelligent classification processing on construction solid waste.

2. The intelligent classification and treatment system for construction solid waste according to claim 1, characterized in that For the multiple classification labels, the system includes: Performing data standardization processing on the dataset to be classified to obtain multiple standard data to be classified; Traversing the multiple standard data to be classified to extract multi-dimensional features and determining multi-dimensional data features; Performing feature annotation on the multiple standard data to be classified according to the multi-dimensional data features to construct a multi-feature network; Traversing the multi-feature network to perform clustering analysis to determine the multiple classification labels.

3. An intelligent classification and treatment system for construction solid waste according to claim 1, characterized in that, Using the multiple classification labels to construct the intelligent classification model, the system includes: Setting a multi-label task environment, and based on the multi-label task environment, traversing the multiple classification labels to perform integrated reinforcement learning to obtain multiple reinforcement learners, where there is a corresponding relationship between the multiple reinforcement learners and the multiple classification labels; Performing supervised training based on the multiple reinforcement learners to generate a first training result, and verifying the first training result to generate a data verification token; Integrating the multiple reinforcement learners according to the data verification token to construct the intelligent classification model.

4. The intelligent classification and treatment system for building solid waste according to claim 3, characterized in that, Synchronizing the dataset to be classified to the intelligent classification model to obtain a classification decision, the system includes: Based on the multiple reinforcement learners, sequentially determining whether the dataset to be classified conforms to the target classification label; Extracting the data to be classified that conforms to the target classification label, generating a positive reward instruction, and identifying the data to be classified that conforms to the target classification label through the positive reward instruction to obtain a first classification result; Extracting the data to be classified that does not conform to the target classification label, generating a negative reward instruction, and identifying the data to be classified that does not conform to the target classification label through the negative reward instruction to obtain a second classification result; Performing weighted calculation based on the first classification result and the second classification result to generate multiple weight coefficients; Formulating the classification decision according to the multiple weight coefficients in combination with the first classification result and the second classification result.

5. An intelligent classification and treatment system for construction solid waste according to claim 1, characterized in that, Simulating the execution of the classification decision to monitor the classification processing of construction solid waste and generating operation feedback data, the system includes: Simulating the execution of the classification decision based on the dataset to be classified to generate a classification control instruction; Activate the sorting machinery according to the classification control instruction and determine multiple areas to be allocated; Classify and process and monitor the construction solid waste according to the multiple areas to be allocated, and obtain a classification monitoring data set, where the classification monitoring data set includes classification speed monitoring data, classification correct monitoring data, and classification error monitoring data; Calculate the classification efficiency of the classification decision based on the classification speed monitoring data, the classification correct monitoring data, and the classification error monitoring data, and construct a classification operation log; Add the classification operation log to the operation feedback data.

6. The intelligent classification and treatment system for building solid waste according to claim 5, wherein, Optimize the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model. The system includes: Perform data mining on the operation feedback data based on the classification operation log to obtain multiple classification association rules; Analyze the operation feedback data according to the multiple classification association rules to obtain an operation analysis result; Conduct a classification evaluation according to the operation analysis result to obtain an operation classification evaluation result; Traverse the operation classification evaluation result to capture anomalies and generate anomaly classification information; Formulate a model adjustment strategy based on the anomaly classification information, execute the model adjustment strategy to retrain the intelligent classification model, and obtain a second training result; Verify the data validity of the second training result, and generate the intelligent optimized classification model when the verification passes.

7. The intelligent classification and treatment system for construction solid waste according to claim 6, characterized in that, Formulate a model adjustment strategy based on the anomaly classification information. The system includes: Perform anomaly tracing based on the anomaly classification information to determine the anomaly data source, and the anomaly data source has an anomaly source identifier; Determine the anomaly mode according to the anomaly source identifier, perform adjustment analysis according to the anomaly mode, and determine the parameter adjustment direction; Formulate the model adjustment strategy of the intelligent classification model based on the parameter adjustment direction in combination with the anomaly data source.

8. An intelligent classification and treatment method for construction solid waste, characterized in that, The method is executed by an intelligent classification processing system for construction solid waste according to any one of claims 1 to 7, and includes: Traverse the target area to collect multi-dimensional data of construction solid waste and determine the data set to be classified; Determine multiple classification labels, construct an intelligent classification model using the multiple classification labels, and synchronize the data set to be classified to the intelligent classification model to obtain a classification decision; Simulate the execution of the classification decision to classify and process and monitor the construction solid waste, and generate operation feedback data; Optimize the intelligent classification model based on the operation feedback data to obtain an intelligent optimized classification model, synchronize the data set to be classified to the intelligent optimized classification model, and obtain a classification optimization decision; Execute the classification optimization decision to perform intelligent classification processing on the construction solid waste.

Images