AI-based multi-objective dynamic optimization decision method and system

Abstract

The application relates to the technical field of decision management, and discloses an AI-based multi-target dynamic optimization decision method and system. Original monitoring data of each link of a solid waste treatment process is collected, data management is performed on the original monitoring data based on a pre-constructed intelligent analysis framework, unified monitoring information is obtained, the unified monitoring information is analyzed according to decision standards of a plurality of decision targets, a preliminary decision set corresponding to each decision target is obtained, each preliminary decision set is subjected to weighted fusion processing according to the weight allocated to each decision target, a dynamic decision to be executed is obtained, the solid waste treatment process is subjected to treatment parameter adjustment, the execution effect of the dynamic decision is analyzed based on subsequent original monitoring data, and the weight allocated to each decision target is optimized.

Description

Technical Field

[0001] This invention relates to the technical field of decision management, and in particular to an AI-based multi-objective dynamic optimization decision-making method and system. Background Technology

[0002] Solid waste treatment is a complex process, typically involving multiple stages such as collection, transportation, storage, pretreatment, treatment, and disposal. Each stage involves various technologies and equipment, such as crushing, sorting, incineration, landfill, and composting. Different types of solid waste (such as municipal solid waste, industrial waste, and hazardous waste) require different treatment methods. The treatment of hazardous waste requires stricter safety measures and specialized equipment to prevent harm to the environment and human health. The various parameters in the solid waste treatment process are interconnected and influence each other. Parameters such as the operating temperature, pressure, and residence time of the treatment equipment affect the treatment effect and energy consumption. The treatment capacity and efficiency of different treatment stages are also mutually restrictive. Traditional solid waste treatment decision-making methods are often based on experience and simple mathematical models, which are difficult to handle complex multi-objective optimization problems and dynamically changing situations. When faced with large amounts of real-time data and complex parameter relationships, traditional methods are prone to decision-making errors or inefficiencies. Summary of the Invention

[0003] The purpose of this invention is to provide an AI-based multi-objective dynamic optimization decision-making method and system, which aims to solve the problem of decision-making errors that easily occur when facing complex multi-objective optimization problems in the prior art.

[0004] This invention is implemented as follows: Firstly, this invention provides an AI-based multi-objective dynamic optimization decision-making method, comprising: The raw monitoring data of each stage of the solid waste treatment process is collected, and the raw monitoring data is managed based on a pre-built intelligent analysis framework to obtain unified monitoring information. The unified monitoring information is analyzed based on the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective; Based on the weights assigned to each decision objective, the preliminary decision sets are weighted and fused to obtain dynamic decisions to be executed, so as to adjust the processing parameters of the solid waste treatment process. The effectiveness of the dynamic decisions is analyzed based on subsequent raw monitoring data in order to optimize the weights assigned to each decision objective.

[0005] In a second aspect, the present invention provides an AI-based multi-objective dynamic optimization decision-making system for implementing the AI-based multi-objective dynamic optimization decision-making method described in any one of the first aspects, comprising: The data monitoring module is used to collect raw monitoring data from each stage of the solid waste treatment process, and to manage the raw monitoring data based on a pre-built intelligent analysis framework to obtain unified monitoring information. The decision analysis module is used to analyze the unified monitoring information according to the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective. The decision execution module is used to perform weighted fusion processing on each of the pre-decision sets according to the weights assigned to each of the decision objectives, so as to obtain the dynamic decision to be executed, and to adjust the processing parameters of the solid waste treatment process. The weight optimization module is used to analyze the execution effect of the dynamic decision based on subsequent raw monitoring data, so as to optimize the weights assigned to each decision objective.

[0006] This invention provides an AI-based multi-objective dynamic optimization decision-making method, which has the following beneficial effects: This invention, by collecting and managing data from each stage of solid waste treatment, can comprehensively grasp the process status. Based on multi-objective decision-making criteria, a preliminary decision set is obtained through analysis, and then a dynamic decision is determined through weighted fusion. This allows for precise adjustment of treatment parameters, taking into account multiple objective requirements. The effectiveness of the decision is evaluated and the weights are optimized based on subsequent data, enabling the decision to adapt to dynamic changes in the process, continuously improving the efficiency of solid waste treatment, reducing costs, and minimizing pollution, thereby maximizing economic and environmental benefits. Attached Figure Description

[0007] Figure 1 This is a schematic diagram illustrating the steps of an AI-based multi-objective dynamic optimization decision-making method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of an AI-based multi-objective dynamic optimization decision-making system provided in an embodiment of the present invention. Detailed Implementation

[0008] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0009] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0010] Reference Figure 1 , Figure 2 The diagram shows a preferred embodiment of the present invention.

[0011] In a first aspect, the present invention provides an AI-based multi-objective dynamic optimization decision-making method, comprising: S1: Collect raw monitoring data from each stage of the solid waste treatment process, and perform data management on the raw monitoring data based on a pre-built intelligent analysis framework to obtain unified monitoring information; S2: Analyze the unified monitoring information according to the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective; S3: Based on the weights assigned to each decision objective, perform weighted fusion processing on each of the preliminary decision sets to obtain dynamic decisions to be executed, so as to adjust the processing parameters of the solid waste treatment process; S4: Analyze the execution effect of the dynamic decision based on subsequent raw monitoring data, so as to optimize the weights assigned to each decision objective.

[0012] Specifically, in step S1 of the embodiment provided by the present invention, various types of sensing modules are installed at key stages of the solid waste treatment process, such as feeding, crushing, sorting, incineration, and landfill. These sensing modules may include temperature sensors, pressure sensors, flow sensors, component analyzers, etc., to collect data on the solid waste treatment process from different monitoring methods. For example, temperature sensors are used to monitor temperature changes inside the treatment equipment, and component analyzers are used to analyze the chemical composition of solid waste. The sensing modules collect data in real time or periodically and transmit this data to the data acquisition system.

[0013] More specifically, solid waste treatment processes are complex, and the operational status of different stages and aspects can affect the treatment effect. A single monitoring method cannot fully reflect the actual situation of solid waste treatment. By monitoring through sensor modules with multiple monitoring methods, richer and more comprehensive raw monitoring data can be obtained, providing a sufficient information foundation for subsequent decision analysis.

[0014] More specifically, after the data acquisition system receives the raw monitoring data, it adds corresponding tags to the data according to its source. For example, if a piece of data is collected by the temperature sensor in the feeding process at 10:00 AM, then the data is marked with the process link tag of "feeding process", the monitoring path tag of "temperature sensor", and the collection time tag of "10:00 AM". These tags can be added to the record of the raw monitoring data in the form of data fields.

[0015] More specifically, the raw monitoring data is massive in quantity and comes from complex sources. Without effective labeling, it is difficult to classify, query, and analyze the data. By generating process step labels, monitoring path labels, and collection time labels, the raw monitoring data can be clearly identified, facilitating subsequent data management and analysis, and enabling the rapid location and filtering of the required data.

[0016] More specifically, the pre-built intelligent analysis framework is a data model with a specific structure and rules. It sets corresponding data storage locations for different process links, monitoring methods, and collection times. Based on the process link markers, monitoring method markers, and collection time markers of the original monitoring data, the data is accurately placed into the corresponding positions in the intelligent analysis framework. For example, the data collected by the temperature sensor at 10:00 AM during the feeding process is placed into the corresponding positions in the framework for feeding process, temperature monitoring, and 10:00 AM. In this way, the scattered original monitoring data is integrated into a unified framework to form unified monitoring information.

[0017] More specifically, raw monitoring data is scattered and disordered after collection, which is not conducive to overall analysis and decision-making. The intelligent analysis framework provides a standardized structure that integrates raw monitoring data according to certain rules, making the data consistent and standardized, which facilitates comprehensive and systematic analysis and monitoring of the solid waste treatment process and enables a more intuitive understanding of the operation status of the entire solid waste treatment process.

[0018] Specifically, in step S2 of the embodiment provided by the present invention, the intelligent analysis framework pre-sets information selection rules corresponding to different decision objectives. When a specific decision objective is determined, the system will call the corresponding information selection mode from the framework according to the objective type. For example, if the decision objective is to reduce the energy consumption of solid waste treatment, the system will activate the selection mode of energy consumption-related data. This mode specifies the extraction of data such as power and running time of each treatment device from unified monitoring information. These information selection modes can be algorithm-based filtering rules or preset query conditions. The system reads the decision objective, automatically matches and activates the corresponding rules or conditions to perform subsequent information selection operations.

[0019] More specifically, the unified monitoring information includes a large amount of data on all aspects of the solid waste treatment process. Different decision-making objectives focus on different information priorities. By activating the corresponding information selection mode, information relevant to the current decision-making objective can be selectively filtered from the massive amount of data, avoiding interference from irrelevant data and improving the efficiency and accuracy of the analysis.

[0020] More specifically, based on the activated information selection mode, the system traverses and filters the unified monitoring information. For example, according to the energy-saving information selection mode mentioned above, the system will search for the power and running time data of each processing device in the unified monitoring information, extract and organize this data, and further preprocess the extracted data, such as data cleaning to remove invalid or erroneous data; data conversion to unify data of different formats, etc., and finally form an object that can be analyzed.

[0021] More specifically, effective decision analysis can only be carried out by obtaining accurate analytical objects that are relevant to the decision-making objectives. Through information selection operations, useful parts of the unified monitoring information are extracted, providing a suitable data foundation for subsequent analysis based on decision-making criteria, and ensuring that the direction of the analysis is consistent with the decision-making objectives.

[0022] More specifically, there are pre-set decision criteria for different decision objectives. For example, for the decision objective of reducing energy consumption, the decision criteria are to calculate energy consumption costs based on equipment power and operating time, and formulate feasible energy consumption reduction solutions in combination with factors such as equipment processing efficiency. The system calculates and evaluates the analysis object based on these decision criteria. Taking energy consumption as an example, by analyzing the energy consumption of different equipment under different operating parameters, several preliminary decisions that can reduce energy consumption are derived, such as adjusting the operating time of certain equipment or replacing low-energy-consumption equipment. A priority index is calculated for each preliminary decision. The calculation of the priority index can be based on a variety of factors, such as the magnitude of energy consumption reduction, implementation costs, and impact on other aspects of the processing flow. By comprehensively evaluating these factors, a numerical priority index is assigned to each preliminary decision.

[0023] More specifically, decision criteria are the basis for making scientific decisions. By analyzing the object of analysis based on decision criteria, feasible preliminary decisions can be selected from many possible options. The introduction of priority index can quantitatively evaluate these preliminary decisions, making it easier to compare and select among multiple preliminary decisions and providing a clearer reference for the final dynamic decision.

[0024] More specifically, the unified monitoring information is analyzed at predetermined intervals to determine the form and magnitude of information changes, thereby obtaining the information change patterns of the unified monitoring information at various time scales. Based on these information change patterns, an adaptive analysis is performed on the decision criteria for the decision objective to obtain the adaptive characteristics of the unified monitoring information corresponding to the decision objective. This allows for adaptive optimization of the selection range of the information selection pattern. More specifically, the system analyzes unified monitoring information at predetermined time intervals (such as daily, weekly, etc.). Through statistical analysis, it observes the changes (such as linear changes, periodic changes, etc.) and magnitudes of information at different time scales (such as hourly, daily, weekly, etc.). Based on the information change patterns, it evaluates the decision criteria for the decision objectives and determines whether the current decision criteria are still suitable for the current data changes. For example, if new trends are found in the energy consumption data of certain devices, the original energy consumption reduction decision criteria need to be adjusted. Based on the results of adaptability analysis, the system extracts the adaptability characteristics of the unified monitoring information corresponding to the decision objectives, such as key nodes of energy consumption data changes and new factors affecting energy consumption. Then, based on these adaptability characteristics, the selection range of the information selection mode is adjusted to ensure that information related to the decision objectives can be selected more accurately in the future.

[0025] More specifically, the operational status of solid waste treatment processes is dynamic, and the unified monitoring information will also change accordingly. By regularly analyzing information change patterns, the dynamic characteristics of the data can be discovered in a timely manner. Adaptive analysis of decision-making criteria can ensure the effectiveness and accuracy of decisions, enabling them to adapt to constantly changing realities. Furthermore, adaptive optimization of information selection patterns can further improve the accuracy and relevance of information selection, thereby enhancing the quality of the entire decision-making process.

[0026] Specifically, in step S3 of the embodiment provided by the present invention, the weight assigned to each decision objective is first determined. These weights are usually pre-set based on factors such as the importance and priority of the decision objective. The sum of the weights is generally 1. For each preliminary decision in the preliminary decision set, the weight of the decision objective in which it is located is multiplied by the original priority index of the preliminary decision to obtain the weighted priority index. For example, if the weight of decision objective A is 0.3 and the priority index of a certain preliminary decision in its preliminary decision set is 80, then the weighted priority index of the preliminary decision is 0.3 × 80 = 24.

[0027] More specifically, different decision objectives have different degrees of importance in the overall decision-making process. This difference can be reflected by assigning weights. By weighting the priority index of preliminary decisions, preliminary decisions under important decision objectives can occupy a more important position in the final decision, avoiding the equal treatment of preliminary decisions for all decision objectives, and thus making the decision results more in line with actual needs and overall objectives.

[0028] More specifically, iterate through all the preliminary decision sets, find the weighted priority index of the same preliminary decision in different decision target preliminary decision sets, and add these weighted priority indices to obtain the fusion priority index of the preliminary decision. For example, if the weighted priority index of preliminary decision X in the preliminary decision set of decision target A is 24 and the weighted priority index in the preliminary decision set of decision target B is 16, then the fusion priority index of preliminary decision X is 24 + 16 = 40.

[0029] More specifically, the same preliminary decision has an impact on multiple decision objectives. By superimposing its weighted priority index in each preliminary decision set, the performance of the preliminary decision under different decision objectives can be comprehensively considered. The fusion priority index can more comprehensively reflect the comprehensive advantages of the preliminary decision in multi-objective decision-making, and provide a more accurate basis for subsequent selection of the optimal decision.

[0030] More specifically, the integration priority index of all preliminary decisions is compared, and the preliminary decision with the highest integration priority index is selected as the dynamic decision to be executed. After the dynamic decision is determined, the corresponding treatment parameters of the solid waste treatment process are adjusted according to the specific content of the decision. For example, if the dynamic decision is to adjust the operating power of a certain piece of equipment, the power setting parameters of that equipment are modified accordingly.

[0031] More specifically, the integration priority index integrates the impact of various decision objectives. Selecting the preliminary decision with the highest integration priority index can meet the needs of multiple decision objectives to the greatest extent, enabling the solid waste treatment process to achieve a better operating state in multiple aspects. Adjusting the treatment parameters based on dynamic decision-making can optimize the solid waste treatment process in a timely manner, improve treatment efficiency, reduce costs, reduce environmental pollution, and achieve the goal of multi-objective dynamic optimization decision-making.

[0032] Specifically, in step S4 of the embodiment provided by the present invention, when the dynamic decision begins to be executed, the system records the execution time point as a benchmark and determines a specified time range according to the actual situation, such as the next 24 hours, a week, etc. Within this time range, data related to the dynamic decision is filtered from the continuously collected raw monitoring data. For example, if the dynamic decision is to adjust the operating parameters of a certain device, then the relevant monitoring data such as temperature, pressure, and processing volume of the device within the specified time are selected.

[0033] More specifically, selecting data based on the execution time of dynamic decisions ensures that the data obtained is generated after the decision is implemented, accurately reflecting the execution effect of the decision. The setting of a specified time range is to evaluate the effect of the decision within a reasonable time period, avoiding inaccurate results due to excessively long or short time periods.

[0034] More specifically, the effect analysis information is processed and analyzed using methods such as statistical analysis and comparative analysis. For example, the changes in energy consumption of a certain equipment and the changes in solid waste treatment efficiency are compared before and after the implementation of the decision. Key indicators that can reflect the effect of the decision implementation are extracted, such as the energy consumption reduction rate and the treatment volume increase rate. These indicators are combined to form the effect characteristics. The weights assigned to each decision objective corresponding to the dynamic decision are recorded. These weights are represented in a specific form (such as vectors, matrices, etc.) to form weight allocation characteristics.

[0035] More specifically, the performance characteristics can intuitively demonstrate the effectiveness of dynamic decisions in actual implementation, providing a quantitative basis for subsequent evaluation of the merits of decisions, while the weight allocation characteristics record the importance allocation of each objective when making decisions, facilitating subsequent analysis of the relationship between weights and performance characteristics.

[0036] More specifically, as time goes on, the system continuously records the execution effect characteristics and weight allocation characteristics of each dynamic decision, arranges the execution effect characteristics at different times into vectors in a certain order, and combines multiple vectors to form the first feedback feature matrix; similarly, the weight allocation characteristics at different times are arranged into vectors to form the second feedback feature matrix.

[0037] More specifically, vectorized and matrix representations facilitate the storage, processing, and analysis of large amounts of data. Through these two matrices, the execution effect and weight allocation of decisions at different times can be clearly seen, providing a data foundation for subsequent correlation analysis.

[0038] More specifically, data analysis methods, such as correlation analysis and regression analysis, are used to study the relationship between the first feedback feature matrix and the second feedback feature matrix. Based on the analysis results, the weight allocation features are decomposed into smaller allocation form units according to certain rules. For example, the weight allocation vector is decomposed into weight sub-vectors for different decision objectives. For each allocation form unit, its corresponding execution effect features are analyzed to obtain execution feedback information, such as the impact of increasing or decreasing the weight of a certain decision objective on the decision execution effect.

[0039] More specifically, correlation analysis can identify the intrinsic link between weight allocation and decision execution effectiveness. By breaking down the weight allocation characteristics into allocation form units and obtaining execution feedback information, we can gain a more detailed understanding of the impact of changes in the weight of each decision objective on the overall decision effect, providing a concrete basis for weight optimization.

[0040] More specifically, computer simulation technology is used to try various combinations of different allocation units. Based on the execution feedback information, the decision execution effect under each combination is evaluated. By comparing the effects of different combinations, the allocation unit combination that theoretically achieves the optimal decision execution effect is found. Based on this optimal combination, the weights allocated to each decision objective are adjusted.

[0041] More specifically, multi-combination simulations can comprehensively explore the effects of different weight allocation schemes, find the most suitable weight allocation method for the current solid waste treatment process, and continuously optimize the weights to make subsequent dynamic decisions more scientific and reasonable, better meet the needs of multiple decision objectives, and improve the overall efficiency of solid waste treatment.

[0042] This invention provides an AI-based multi-objective dynamic optimization decision-making method, which has the following beneficial effects: This invention, by collecting and managing data from each stage of solid waste treatment, can comprehensively grasp the process status. Based on multi-objective decision-making criteria, a preliminary decision set is obtained through analysis, and then a dynamic decision is determined through weighted fusion. This allows for precise adjustment of treatment parameters, taking into account multiple objective requirements. The effectiveness of the decision is evaluated and the weights are optimized based on subsequent data, enabling the decision to adapt to dynamic changes in the process, continuously improving the efficiency of solid waste treatment, reducing costs, and minimizing pollution, thereby maximizing economic and environmental benefits.

[0043] Preferably, the steps of collecting raw monitoring data from each stage of the solid waste treatment process and managing the raw monitoring data based on a pre-built intelligent analysis framework to obtain unified monitoring information include: S11: The solid waste treatment process is monitored through multiple monitoring channels and sensor modules to obtain the original monitoring data of each stage relative to each monitoring channel. S12: Generate process step markers, monitoring path markers, and collection time markers for the original monitoring data based on the solid waste treatment process steps, monitoring methods, and collection time corresponding to the original monitoring data. S13: Substitute the original monitoring data into the corresponding position of the pre-built intelligent analysis framework according to the process step marker, the monitoring path marker, and the collection time marker to obtain unified monitoring information.

[0044] Specifically, various types of sensing modules are rationally deployed at key stages of the solid waste treatment process, such as feeding, crushing, sorting, incineration, and landfill. For example, a weight sensor is installed at the feed inlet to monitor the feed amount, and a temperature sensor and a gas composition sensor are installed inside the incinerator to monitor the furnace temperature and the composition of the gases produced during combustion, respectively. The sensing modules collect data in real time or at preset time intervals. For example, the temperature sensor can collect temperature data every 5 minutes, and the gas composition sensor can collect gas composition data every minute. The collected data is transmitted to the data acquisition terminal via wired or wireless communication. The data acquisition terminal performs preliminary processing on the received raw data, such as noise removal and data calibration, to improve the data quality.

[0045] More specifically, solid waste treatment processes are complex, and the operational status of different stages and aspects can affect the treatment effect. A single monitoring method can only obtain limited information and cannot fully reflect the actual situation of solid waste treatment. Monitoring through sensor modules with multiple monitoring channels can obtain data from multiple dimensions, ensuring a comprehensive understanding of the solid waste treatment process. Abundant raw monitoring data provides a sufficient information foundation for subsequent decision analysis. Only by mastering detailed data of each stage can we accurately analyze the problems existing in the solid waste treatment process and formulate reasonable decision-making solutions.

[0046] More specifically, clear tag generation rules are established. For example, process stage tags can use abbreviations, such as "JL" for the feeding stage and "PS" for the crushing stage; monitoring path tags can be named according to the type of sensor module, such as "WT" for the temperature sensor and "QL" for the gas composition sensor; and acquisition time tags directly record the specific time of data acquisition, accurate to the second. Programs are written in the data acquisition terminal or data processing system to automatically add corresponding tags to the original monitoring data based on its source. For example, when the system receives data from the feed inlet temperature sensor, it automatically adds the process stage tag "JL", the monitoring path tag "WT", and the current acquisition time tag to the data. The generated tags are stored in the database along with the original monitoring data for subsequent querying and use.

[0047] More specifically, the raw monitoring data is massive in volume and complex in origin. Without effective labeling, it is difficult to classify, query, and analyze the data. By generating process step labels, monitoring path labels, and collection time labels, the raw monitoring data can be clearly identified, facilitating subsequent data management and maintenance. Labeling helps to accurately filter and extract the required data during the data analysis process. For example, when analyzing temperature changes in the feeding process, relevant data can be quickly found based on process step labels and monitoring path labels, improving analysis efficiency and accuracy.

[0048] More specifically, an intelligent analysis framework with a specific structure and rules is pre-constructed. This framework can be a multi-dimensional data model, setting corresponding data storage locations for different process stages, monitoring methods, and collection times. For example, a three-dimensional array can be set in the framework, with the first dimension representing the process stage, the second dimension representing the monitoring method, and the third dimension representing the collection time. Based on the process stage markers, monitoring method markers, and collection time markers of the original monitoring data, the data is accurately placed into the corresponding positions in the intelligent analysis framework. For example, the data collected by the temperature sensor at 10:00 AM in the feeding stage is placed into the array elements corresponding to "JL" (feeding stage), "WT" (temperature sensor), and "10:00 AM" in the framework. By substituting all the original monitoring data into the intelligent analysis framework, a unified and structured dataset is formed, namely unified monitoring information. This information can intuitively display the operating status of the solid waste treatment process at different stages and times.

[0049] More specifically, raw monitoring data is scattered and disordered after collection, which is not conducive to overall analysis and decision-making. The intelligent analysis framework provides a standardized structure to integrate raw monitoring data according to certain rules, making the data consistent and standardized. This facilitates comprehensive and systematic analysis and monitoring of the solid waste treatment process. Unified monitoring information can more intuitively reflect the overall operating status of the solid waste treatment process, providing decision-makers with clear decision-making basis. Decision-makers can quickly understand the operation of each link, discover potential problems, and formulate corresponding decision-making solutions in a timely manner by viewing unified monitoring information.

[0050] Preferably, the step of analyzing the unified monitoring information according to the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective includes: S21: Based on the intelligent analysis framework, activate the corresponding information selection mode for the decision-making objective; S22: Select information from the unified monitoring information according to the information selection mode to obtain the analysis object corresponding to the decision objective; S23: Analyze the object of analysis according to the decision criteria corresponding to the decision objective to obtain several preliminary decisions and priority indices corresponding to each preliminary decision, which serve as a set of preliminary decisions corresponding to the decision objective.

[0051] Specifically, the intelligent analysis framework pre-sets information selection modes corresponding to different decision objectives. These modes can be rule-based logical judgments or machine learning models. For example, if the decision objective is to reduce solid waste treatment costs, the corresponding information selection mode is set to filter data related to raw material procurement, equipment energy consumption, and labor costs from unified monitoring information. Once a specific decision objective is determined, the system automatically searches for and activates the matching information selection mode in the intelligent analysis framework. For example, when the decision objective is to improve solid waste treatment efficiency, the system quickly locates and activates the information selection mode corresponding to that objective. This mode focuses on data selection related to equipment processing speed, material transfer time, etc.

[0052] More specifically, the unified monitoring information includes massive amounts of data from all aspects of the solid waste treatment process. Different decision-making objectives focus on different information priorities. By activating the corresponding information selection mode, information closely related to the current decision-making objective can be selectively filtered from the massive data, avoiding interference from irrelevant data and improving the efficiency and accuracy of subsequent analysis. Accurately selecting information related to the decision-making objective is the foundation for making high-quality decisions. A suitable information selection mode can ensure the acquisition of key data, providing strong support for subsequent analysis and decision-making.

[0053] More specifically, if the information selection mode is rule-based, the system traverses and filters the unified monitoring information according to preset rules. For example, if the information selection mode specifies that data on the operating power of a certain device within a specific time period is greater than a certain threshold, the system will search for and extract data that meets the condition in the unified monitoring information. If the information selection mode is a machine learning model, the system inputs the unified monitoring information into the model, and the model outputs information related to the decision-making objective through calculation and analysis. For example, a trained neural network model is used to extract feature data related to solid waste treatment quality from the unified monitoring information. The filtered or extracted information is integrated and organized to form an object that can be analyzed. For example, data from different sources but related to the same decision-making objective are associated and merged, and duplicate and invalid data are removed.

[0054] More specifically, by selecting information, data directly related to the decision-making objectives can be extracted from unified monitoring information to form analysis objects. This allows subsequent analysis to focus on key information and avoids wasting time and resources on irrelevant data. Accurate analysis objects are a prerequisite for effective analysis. Only by obtaining accurate data that is highly relevant to the decision-making objectives can reliable analysis results be obtained, providing a scientific basis for decision-making.

[0055] More specifically, for each decision objective, there are pre-defined decision criteria. The system substitutes the analysis object into these criteria for analysis. For example, if the decision objective is to optimize the environmental indicators of solid waste treatment processes, the decision criteria include pollutant emission limits and resource recycling rates. The system evaluates the pollutant emission data and resource recycling data of the analysis object according to these criteria. Based on the analysis results, it generates several preliminary decisions that meet the decision objective. For example, if the analysis finds that the pollutant emissions of a certain equipment exceed the standard, the generated preliminary decisions include adjusting the equipment operating parameters and replacing it with equipment with better environmental performance. A priority index is calculated for each preliminary decision. The calculation of the priority index can comprehensively consider multiple factors, such as implementation cost, expected effect, and implementation difficulty. For example, for the preliminary decision to adjust the equipment operating parameters, if the implementation cost is low, the expected effect is good, and the implementation difficulty is low, the priority index is high. Through the quantified priority index, the preliminary decisions can be ranked and compared.

[0056] More specifically, by analyzing the analysis object according to decision criteria, multiple preliminary decisions are generated, providing decision-makers with a variety of feasible options, increasing the flexibility and diversity of decision-making. The introduction of priority index allows decision-makers to intuitively understand the advantages and disadvantages of each preliminary decision, facilitating comparison and selection among multiple preliminary decisions, thereby making more scientific and reasonable decisions.

[0057] Preferably, the unified monitoring information is analyzed at predetermined intervals to determine the information change patterns and magnitudes at each time scale. Based on these information change patterns, an adaptive analysis is performed on the decision criteria for the decision objective to obtain the adaptive characteristics of the unified monitoring information corresponding to the decision objective. This allows for adaptive optimization of the selection range of the information selection mode.

[0058] Specifically, unified monitoring information is grouped according to predetermined time scales (such as hours, days, weeks, and months). For example, data within a day is divided into hourly segments, and data within a month is divided into weekly segments. Statistical methods and data analysis techniques are used to observe the patterns exhibited by each group of data over time, determining the form of information change. Common forms of change include linear changes (data shows a straight upward or downward trend), periodic changes (such as similar fluctuations every day or week), and random changes (data has no obvious pattern). For example, by drawing line graphs of data for each time period, if the graph approximates a straight line, it is judged as a linear change; if periodically repeating peaks and troughs appear, it is a periodic change.

[0059] More specifically, for each set of data, the difference between its maximum and minimum values ​​within the corresponding time scale is calculated, or the magnitude of information change is measured by calculating a statistical measure (such as the mean or standard deviation) of the absolute value of the difference between data at adjacent time points. For example, the operating power of a device is statistically analyzed hourly over a day, and the difference between the highest and lowest power is calculated to obtain the magnitude of power change over a day.

[0060] More specifically, by comprehensively considering the forms and magnitudes of information changes across various time scales, we can summarize the change patterns of unified monitoring information across different time dimensions. For example, if we find that the energy consumption data of a certain type of solid waste treatment process exhibits periodic changes during the daily working hours, and that energy consumption on weekends is significantly lower than on weekdays, this is a specific information change pattern. We store the summarized information change patterns in a database and update them according to predetermined schedules. As new unified monitoring information is continuously added, the original change patterns will change, and timely updates can ensure their accuracy and timeliness.

[0061] More specifically, the summarized information change patterns are compared and analyzed with the decision criteria for each decision objective. For example, if a decision objective is to control solid waste treatment costs within a certain range, and the decision criterion is set as monthly costs not exceeding a certain amount, analysis of unified monitoring information reveals that treatment costs experience periodic and significant increases in certain months of each quarter. This necessitates an assessment of whether the current decision criteria can adapt to such changes. Based on the comparison results, it is determined whether relevant indicators in the decision criteria need to be adjusted. If the information change patterns are severely mismatched with the decision criteria, making it impossible to make effective decisions according to the existing criteria, then adjustments are deemed necessary. For instance, due to seasonal fluctuations in raw material prices leading to periodic changes in solid waste treatment costs, the original cost control decision criteria need to be adjusted accordingly for different seasons.

[0062] More specifically, features related to the decision-making objective and reflecting the adaptation of unified monitoring information to the decision-making objective are extracted from information change patterns and adaptive analysis results. For example, for the decision-making objective of improving solid waste treatment efficiency, it is found that the throughput of the treatment equipment fluctuates significantly at different times of the day, and the fluctuation period is related to the workers' shift schedule. Then, the "temporal fluctuation pattern of throughput" and "correlation with shift schedule" can be used as adaptive features of the decision-making objective. These adaptive features are described in a quantitative way, such as using correlation coefficients to represent the strength of the correlation and fluctuation ranges to represent the degree of change. The quantified features are stored in a database for subsequent analysis and use.

[0063] More specifically, based on the extracted adaptive features, the rules or conditions used in the information selection mode are adjusted. For example, the original information selection mode only selects data from a device during normal working hours. However, analysis reveals that some abnormal data from the device during non-working hours also affect the decision-making objective. Therefore, the time range for information selection can be expanded to include data from non-working hours. The optimization of the information selection mode is a dynamic process. As the unified monitoring information is continuously updated and the information change pattern evolves, the selection range is constantly adjusted and optimized to ensure that the information most relevant to the decision-making objective can be accurately selected.

[0064] More specifically, solid waste treatment is a dynamic process, and unified monitoring information will change with factors such as time, environment, and process conditions. By regularly analyzing the form and magnitude of information changes, these changes can be captured in a timely manner, and their change patterns at different time scales can be summarized, thereby better addressing the dynamism of data and providing accurate basis for scientific decision-making.

[0065] More specifically, the decision-making criteria for decision-making objectives are formulated under certain assumptions and conditions. When the change pattern of unified monitoring information does not match the decision-making criteria, the decision made according to the original criteria cannot achieve the expected results. By conducting an adaptive analysis of the decision-making criteria, standards that do not conform to the actual situation can be identified and adjusted in a timely manner, ensuring the effectiveness and feasibility of the decision.

[0066] More specifically, the scope of information selection mode directly affects the quality and accuracy of subsequent analysis objects. By extracting the adaptive characteristics of decision-making objectives corresponding to unified monitoring information and optimizing the scope of information selection mode accordingly, information selection can be more accurately focused on key information related to decision-making objectives, thereby improving the efficiency and quality of decision analysis.

[0067] More specifically, the entire process forms a closed-loop feedback mechanism. Through continuous analysis of unified monitoring information and dynamic adjustment of decision-making criteria and information selection patterns, adaptive optimization of the decision-making process can be achieved. As solid waste treatment processes continue to change, the system can automatically adjust its decision-making strategies to adapt to new situations, improving the flexibility and adaptability of decision-making.

[0068] Preferably, the step of performing weighted fusion processing on each of the preliminary decision sets according to the weights assigned to each of the decision objectives to obtain dynamic decisions to be executed, in order to adjust the processing parameters of the solid waste treatment process, includes: S31: Assign the weights allocated to the decision objective to each of the pre-decision decisions in the pre-decision set, so as to perform weighted processing on the priority index of each of the pre-decision decisions to obtain the weighted priority index of each of the pre-decision decisions; S32: The weighted priority indices of the same preliminary decision in each of the preliminary decision sets are numerically superimposed to obtain the fusion priority index of the preliminary decision; S33: Select the optimal preliminary decision based on the fusion priority index as the dynamic decision to be executed, and adjust the processing parameters of the solid waste treatment process based on the dynamic decision.

[0069] Specifically, before weighted processing, the weight assigned to each decision objective must first be determined. These weights are usually pre-set based on factors such as the importance of the decision objective, actual needs, and the focus of the current solid waste treatment process. The sum of the weights of all decision objectives is 1. For example, if there are three decision objectives: reducing costs, improving treatment efficiency, and reducing environmental pollution, their assigned weights are 0.3, 0.4, and 0.3, respectively.

[0070] More specifically, for each preliminary decision in the preliminary decision set, the weight of the decision objective to which it belongs is multiplied by the original priority index of the preliminary decision to obtain the weighted priority index. Suppose that in the preliminary decision set for the decision objective of reducing costs, there is a preliminary decision with a priority index of 80 and a weight of 0.3, then the weighted priority index of this preliminary decision is 0.3 × 80 = 24.

[0071] More specifically, different decision objectives have varying degrees of importance in the actual solid waste treatment process. By assigning weights to each decision objective and weighting the priority index of preliminary decisions, preliminary decisions under important decision objectives can occupy a more important position in the final decision evaluation, avoiding treating all decision objectives equally. This makes the decision results more in line with actual needs and overall goals. The weighting quantifies the importance of decision objectives, allowing the influence of each preliminary decision under different decision objectives to be reflected in specific values, facilitating subsequent comprehensive evaluation and comparison.

[0072] More specifically, iterate through all the preliminary decision sets and find the weighted priority index of the same preliminary decision in different decision target preliminary decision sets. It is necessary to uniquely identify each preliminary decision, for example, by assigning a specific number to each preliminary decision for accurate matching. The weighted priority indices of the same preliminary decision in different preliminary decision sets are added together. For example, if a preliminary decision has a weighted priority index of 24 in the cost reduction decision target preliminary decision set, a weighted priority index of 32 in the efficiency improvement decision target preliminary decision set, and a weighted priority index of 18 in the environmental pollution reduction decision target preliminary decision set, then the combined priority index of this preliminary decision is 24 + 32 + 18 = 74.

[0073] More specifically, the same preliminary decision affects multiple decision objectives. Considering the priority index under a single decision objective alone cannot fully reflect the merits of that decision. By superimposing the weighted priority indices of the same preliminary decision in each set of preliminary decisions, a fused priority index is obtained. This allows for a comprehensive consideration of the decision's performance under multiple decision objectives, providing a more complete assessment of its overall advantages in multi-objective decision-making. The fused priority index provides a unified comparison standard for all preliminary decisions, enabling comparison of preliminary decisions under different decision objectives on the same scale, thus facilitating the selection of the optimal decision.

[0074] More specifically, the fusion priority index of all preliminary decisions is compared, and the preliminary decision with the highest fusion priority index is selected as the dynamic decision to be executed. This can be achieved automatically through programming, or the decision can be made manually after reviewing the ranking results. After the dynamic decision is determined, the corresponding treatment parameters of the solid waste treatment process are adjusted according to the specific content of the decision. For example, if the dynamic decision is to adjust the operating power of a certain piece of equipment to achieve the comprehensive goal of reducing costs and improving treatment efficiency, then the power setting parameters of that equipment are modified according to the decision requirements.

[0075] More specifically, selecting the preliminary decision with the highest integration priority index as the dynamic decision can maximize the satisfaction of multiple decision objectives, enabling the solid waste treatment process to achieve better operation in multiple aspects, realizing dynamic optimization of multiple objectives, and adjusting the treatment parameters according to the dynamic decision can optimize the solid waste treatment process in a timely manner, improve treatment efficiency, reduce costs, reduce environmental pollution, and thus improve the overall benefits of solid waste treatment.

[0076] Preferably, the step of analyzing the execution effect of the dynamic decision based on subsequent raw monitoring data to optimize the weights assigned to each decision objective includes: S41: The execution time of the dynamic decision is marked as a benchmark to select the original monitoring data within a subsequent specified time range to obtain the effect analysis information of the dynamic decision; S42: Analyze the execution effect of the dynamic decision based on the effect analysis information to obtain the execution effect characteristics of the dynamic decision, and generate the weight allocation characteristics of the dynamic decision based on the weights allocated to each decision objective corresponding to the dynamic decision. S43: Record the execution effect characteristics and weight allocation characteristics of dynamic decisions at each time point in order to optimize the weights allocated to each decision objective.

[0077] Specifically, when a dynamic decision begins execution, the system automatically records that moment as a baseline time, accurate to the second. This record can be stored in a dedicated log file or a specific field in a database table, and linked to other key information of the dynamic decision. Based on the characteristics of the solid waste treatment process and the nature of the decision, an appropriate time range is pre-set. For example, for some short-term adjustment decisions, the time range can be set to 24 or 48 hours after execution; for long-term strategic decisions, it can be set to a week, a month, or even longer. Based on the set time range, data within that time period is filtered from the continuously collected raw monitoring data repository (such as a database). The filtering criteria can be matched and queried based on the timestamp field to ensure that only data within the specified time range after the execution of the dynamic decision is obtained. This data is the effect analysis information.

[0078] More specifically, selecting data based on the execution time of dynamic decisions ensures that the data is generated after the decision is implemented, eliminating interference from other factors before the decision is implemented. This allows for a more accurate reflection of the actual implementation effect of the decision. By specifying a time range, the effect of the decision can be evaluated within a reasonable period of time. If the time is too short, the long-term impact of the decision cannot be fully reflected; if the time is too long, more interference from external factors will be introduced, making it difficult to accurately judge the true effect of the decision.

[0079] More specifically, various data analysis methods are used to process and analyze the effect analysis information, such as comparative analysis, which compares key indicators (such as cost, treatment volume, pollutant emissions, etc.) before and after the decision is implemented; and trend analysis, which observes the changing trends of relevant indicators over time after the decision is implemented, and extracts key features that can directly reflect the effect of the decision implementation from the analysis results, such as the percentage reduction in cost, the increase in treatment volume, and the degree of reduction in pollutant emissions. These features are then quantified and organized to form an implementation effect feature vector or matrix.

[0080] More specifically, examine and record the weights assigned to each decision objective corresponding to the dynamic decision, and express these weights in a specific form (such as a vector) to form a weight allocation feature. For example, if there are three decision objectives with weights of 0.2, 0.5, and 0.3 respectively, the weight allocation feature can be represented as [0.2, 0.5, 0.3].

[0081] More specifically, the performance characteristics can present the effectiveness of dynamic decision-making in a concrete and quantifiable way, making it easier to intuitively evaluate the merits and effectiveness of decisions and providing a clear basis for subsequent analysis and adjustment. The weight allocation characteristics record the importance distribution of each objective when making decisions. Combining them with the performance characteristics helps to analyze the impact of different weight allocations on the performance of decision-making and provides a basis for weight optimization.

[0082] More specifically, over time, the execution effect characteristics and weight allocation characteristics of each dynamic decision are continuously recorded. A dedicated database table can be established to store these records. Each record contains information such as decision identifier, execution time, execution effect characteristics, and weight allocation characteristics, ensuring data integrity and traceability. The recorded data can be analyzed using statistical analysis methods, machine learning algorithms, etc. For example, correlation analysis can be used to study the degree of correlation between weight allocation characteristics and execution effect characteristics; regression analysis can be used to establish a mathematical model between weights and effects. Based on the analysis results, the optimal weight allocation combination that can achieve the best decision execution effect can be found. By simulating different weight allocation schemes, the corresponding execution effects can be predicted, and the optimal scheme can be selected to adjust and optimize the weights allocated to each decision objective.

[0083] More specifically, recording the execution effect characteristics and weight allocation characteristics of dynamic decisions at each moment to form historical data helps to accumulate experience and knowledge in the decision-making process, providing reference and guidance for future decisions. Solid waste treatment is a dynamic process with constantly changing external environment and internal conditions, and the original weight allocation is no longer applicable. By regularly analyzing the recorded data and optimizing the weights, the decision-making process can be made more scientific and reasonable, continuously improving the quality and effectiveness of decisions to adapt to changing environment and needs.

[0084] Preferably, the step of recording the execution effect characteristics and weight allocation characteristics of dynamic decisions at each time point, in order to optimize the weights allocated to each decision objective, includes: S41: Record the execution effect characteristics and weight allocation characteristics of the dynamic decision at each time point, and express them in vector form to obtain the first feedback feature matrix and the second feedback feature matrix; S42: Based on the correlation analysis of the first feedback feature matrix and the second feedback feature matrix, various weight allocation features are decomposed into several allocation form units, and the execution feedback information of each allocation form unit is obtained; S43: Perform multiple combination simulations on each of the allocation form units based on the execution feedback information of each allocation form unit to obtain the theoretically optimal combination of allocation form units, so as to optimize the weights allocated to each decision objective.

[0085] Specifically, after each dynamic decision is executed, its execution effect characteristics (such as cost reduction rate, percentage improvement in treatment efficiency, pollutant emission reduction, etc.) and weight allocation characteristics (weights assigned to each decision objective) are recorded in detail. These records can be stored in a database. Each record corresponds to a dynamic decision and includes the decision execution time, execution effect characteristic value, and weight allocation characteristic value.

[0086] More specifically, the execution effect characteristics and weight allocation characteristics of each dynamic decision are organized into vector form. For example, assuming there are three decision objectives, the execution effect characteristics include three index values, which are combined into a three-dimensional vector; the weight allocation characteristics are also the weight values ​​of the three objectives, which are also combined into a three-dimensional vector. As time goes by, the execution effect characteristic vectors of multiple dynamic decisions are arranged in chronological order to form the first feedback characteristic matrix, and the weight allocation characteristic vectors are arranged in chronological order to form the second feedback characteristic matrix. For example, after 10 dynamic decisions, the first feedback characteristic matrix is ​​a 10-row, 3-column matrix, and the second feedback characteristic matrix is ​​also a 10-row, 3-column matrix.

[0087] More specifically, vectorization and matrix representation are common methods in data analysis and machine learning, which facilitate various mathematical operations and statistical analyses. By constructing the execution effect features and weight allocation features into matrices, mature matrix operation tools and algorithms can be used for subsequent processing. The first feedback feature matrix and the second feedback feature matrix can present the execution effect and weight allocation of dynamic decision-making at different times as a whole, making it easy to observe the trend of the decision effect changing with the weight allocation, and providing a clear data structure for subsequent correlation analysis.

[0088] More specifically, various methods can be used for correlation analysis, such as correlation analysis (calculating Pearson correlation coefficient, etc.) and regression analysis (linear regression, nonlinear regression). For example, linear regression analysis can be used to explore the linear relationship between weight allocation characteristics and performance characteristics. Based on the results of the correlation analysis, the weight allocation characteristics can be decomposed into several allocation form units. For instance, if there are three decision objectives A, B, and C, the original weight allocation characteristic is a three-dimensional vector (wA, wB, wC), which can be decomposed into three allocation form units: increasing the weight of A alone, increasing the weight of B alone, and increasing the weight of C alone. For each allocation form unit, its impact on performance characteristics can be analyzed to obtain the corresponding performance feedback information. For example, if the analysis shows that increasing the weight of A alone significantly improves the cost reduction rate, then the increase in the cost reduction rate is the performance feedback information of that allocation form unit.

[0089] More specifically, correlation analysis can reveal the intrinsic connection between weight allocation characteristics and execution effect characteristics. By breaking down weight allocation characteristics into allocation form units, we can study the individual impact of each weight factor on the decision execution effect in more detail. This helps to deepen our understanding of the complex relationship between weight and effect. The execution feedback information of each allocation form unit provides a specific basis for subsequent weight optimization. Understanding the impact of each allocation form unit on the decision effect allows us to adjust the weight allocation in a targeted manner to achieve better decision results.

[0090] More specifically, computer simulation technology can be used to combine various allocation units in multiple ways. For example, for the three allocation units mentioned above, various scenarios can be simulated, such as using a single unit, a combination of two units, or a combination of three units. Based on the execution feedback information of each allocation unit, the decision-making effect under each combination can be evaluated. A comprehensive evaluation index can be set, such as comprehensively considering multiple indicators such as cost, efficiency, and environmental protection, to measure the advantages and disadvantages of each combination.

[0091] More specifically, by comparing the evaluation results of various combinations, the combination that achieves the optimal comprehensive evaluation index is selected as the theoretically optimal allocation unit combination. Based on the theoretically optimal allocation unit combination, the weights allocated to each decision objective are adjusted. For example, if the optimal combination is to simultaneously increase the weights of A and B, then the weight values ​​of A and B are increased accordingly, while the weight value of C is decreased, in order to optimize the weights.

[0092] More specifically, a single weight allocation method cannot achieve the best decision-making effect. Through multiple combination simulations, we can comprehensively explore various possible weight allocation schemes and find the combination that theoretically optimizes the decision execution effect. The solid waste treatment process is dynamic, and the importance of each decision objective may change at different times. By continuously conducting combination simulations and weight optimization, the decision-making process can better adapt to this dynamic change and improve the scientificity and effectiveness of the decision.

[0093] Reference Figure 2 As shown, in a second aspect, the present invention provides an AI-based multi-objective dynamic optimization decision-making system for implementing the AI-based multi-objective dynamic optimization decision-making method described in any one of the first aspects, comprising: The data monitoring module is used to collect raw monitoring data from each stage of the solid waste treatment process, and to manage the raw monitoring data based on a pre-built intelligent analysis framework to obtain unified monitoring information. The decision analysis module is used to analyze the unified monitoring information according to the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective. The decision execution module is used to perform weighted fusion processing on each of the pre-decision sets according to the weights assigned to each of the decision objectives, so as to obtain the dynamic decision to be executed, and to adjust the processing parameters of the solid waste treatment process. The weight optimization module is used to analyze the execution effect of the dynamic decision based on subsequent raw monitoring data, so as to optimize the weights assigned to each decision objective.

[0094] In this embodiment, the specific implementation of each module in the above system embodiment is described in the above method embodiment, and will not be repeated here.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-objective dynamic optimization decision-making method based on AI, characterized in that, include: The raw monitoring data of each stage of the solid waste treatment process is collected, and the raw monitoring data is managed based on a pre-built intelligent analysis framework to obtain unified monitoring information. The unified monitoring information is analyzed based on the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective; Based on the weights assigned to each decision objective, the preliminary decision sets are weighted and fused to obtain dynamic decisions to be executed, so as to adjust the processing parameters of the solid waste treatment process. The effectiveness of the dynamic decisions is analyzed based on subsequent raw monitoring data in order to optimize the weights assigned to each decision objective.

2. The AI-based multi-objective dynamic optimization decision-making method as described in claim 1, characterized in that, The steps of collecting raw monitoring data from each stage of the solid waste treatment process and managing the raw monitoring data based on a pre-built intelligent analysis framework to obtain unified monitoring information include: The solid waste treatment process is monitored by sensor modules with multiple monitoring channels to obtain the original monitoring data of each stage relative to each monitoring channel. Based on the solid waste treatment process steps, monitoring methods, and collection times corresponding to the original monitoring data, process step markers, monitoring method markers, and collection time markers are generated for the original monitoring data respectively. Based on the process step markers, monitoring path markers, and data collection time markers, the raw monitoring data is substituted into the corresponding positions of the pre-built intelligent analysis framework to obtain unified monitoring information.

3. The AI-based multi-objective dynamic optimization decision-making method as described in claim 1, characterized in that, The steps of analyzing the unified monitoring information based on the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective include: Based on the intelligent analysis framework, the corresponding information selection mode is activated for the decision-making objective; The unified monitoring information is selected according to the information selection mode to obtain the analysis object corresponding to the decision objective; The analysis object is analyzed according to the decision criteria corresponding to the decision objective to obtain several preliminary decisions and priority indices for each preliminary decision, which serve as a set of preliminary decisions corresponding to the decision objective.

4. The AI-based multi-objective dynamic optimization decision-making method as described in claim 3, characterized in that, The unified monitoring information is analyzed at predetermined intervals to determine the information change patterns and magnitudes at various time scales. Based on these patterns, an adaptive analysis is performed on the decision criteria for the decision objective to obtain the adaptive characteristics of the unified monitoring information corresponding to the decision objective. This allows for adaptive optimization of the selection range of the information selection mode.

5. The AI-based multi-objective dynamic optimization decision-making method as described in claim 1, characterized in that, The steps of weighted fusion processing of each of the preliminary decision sets according to the weights assigned to each decision objective to obtain the dynamic decision to be executed, and adjusting the processing parameters of the solid waste treatment process, include: The weights assigned to the decision objective are distributed to each of the preliminary decisions in the preliminary decision set, and the priority indices of each of the preliminary decisions are weighted to obtain the weighted priority index of each of the preliminary decisions. The weighted priority indices of the same preliminary decision in each of the preliminary decision sets are numerically superimposed to obtain the fusion priority index of the preliminary decision; The optimal preliminary decision is selected based on the fusion priority index as the dynamic decision to be executed, and the processing parameters of the solid waste treatment process are adjusted based on the dynamic decision.

6. The AI-based multi-objective dynamic optimization decision-making method as described in claim 1, characterized in that, The steps of analyzing the execution effect of the dynamic decisions based on subsequent raw monitoring data, and optimizing the weights assigned to each decision objective, include: The execution time of the dynamic decision is marked as a benchmark to select the original monitoring data within a subsequent specified time range, thereby obtaining the effect analysis information of the dynamic decision; The execution effect of the dynamic decision is analyzed based on the effect analysis information to obtain the execution effect characteristics of the dynamic decision, and the weight allocation characteristics of the dynamic decision are generated according to the weights assigned to each decision objective corresponding to the dynamic decision. The execution effect characteristics and weight allocation characteristics of dynamic decisions at each time point are recorded in order to optimize the weights allocated to each decision objective.

7. The AI-based multi-objective dynamic optimization decision-making method as described in claim 6, characterized in that, The steps for recording the performance characteristics and weight allocation characteristics of dynamic decisions at each time point, and optimizing the weights assigned to each decision objective, include: The execution effect characteristics and weight allocation characteristics of dynamic decisions at each time point are recorded and expressed in vector form to obtain the first feedback feature matrix and the second feedback feature matrix. Based on the correlation analysis of the first feedback feature matrix and the second feedback feature matrix, various weight allocation features are decomposed into several allocation form units, and the execution feedback information of each allocation form unit is obtained. Based on the execution feedback information of each allocation unit, multiple combination simulations are performed on each allocation unit to obtain the theoretically optimal combination of allocation units, so as to optimize the weights allocated to each decision objective.

8. An AI-based multi-objective dynamic optimization decision-making system, characterized in that, A method for implementing an AI-based multi-objective dynamic optimization decision-making method according to any one of claims 1-7 includes: The data monitoring module is used to collect raw monitoring data from each stage of the solid waste treatment process, and to manage the raw monitoring data based on a pre-built intelligent analysis framework to obtain unified monitoring information. The decision analysis module is used to analyze the unified monitoring information according to the decision criteria of several decision objectives to obtain a preliminary decision set corresponding to each decision objective. The decision execution module is used to perform weighted fusion processing on each of the pre-decision sets according to the weights assigned to each of the decision objectives, so as to obtain the dynamic decision to be executed, and to adjust the processing parameters of the solid waste treatment process. The weight optimization module is used to analyze the execution effect of the dynamic decision based on subsequent raw monitoring data, so as to optimize the weights assigned to each decision objective.

Images