Basic-level governance auxiliary decision-making data processing method based on nonlinear mapping function

Abstract

The invention relates to the technical field of grassroots governance and intelligent decision support systems, in particular to a grassroots governance auxiliary decision data processing method and a data processing platform, comprising event data acquisition and preprocessing, recursive calling and collaborative optimization based on an optimization algorithm, path simulation and optimization through a virtual environment and a multi-dimensional probability network, and data processing. And adaptive decision adjustment based on real-time feedback. Through the method and the system, efficient event processing and global optimization can be realized, the decision-making capability and the response speed for coping with complex events are enhanced, and the overall efficiency of grassroots governance is remarkably improved.

Description

[0001] Cross - reference to related applications This application is a divisional application of a Chinese patent application with the application number 2024112757868, the application date: September 12, 2024, and the invention title "Data Processing Method and Data Processing Platform for Grassroots Governance Auxiliary Decision - making". Technical field

[0002] The present invention relates to the technical fields of grassroots governance and intelligent decision - making support systems, and particularly to a data processing method and a data processing platform for grassroots governance auxiliary decision - making. Background technique

[0003] Grassroots governance involves the handling of a large number of daily problems and the response to emergencies. In this process, how to efficiently collect, analyze, and process various event data is the key to the success of grassroots governance. Traditional manual processing methods are increasingly difficult to meet the growing complex needs of grassroots governance, especially in modern cities where the types of events are diverse and the frequency is high, and timely and accurate decision - making support is required. For this reason, modern grassroots governance gradually relies on intelligent and data - driven technical means to help improve management efficiency and decision - making accuracy.

[0004] Existing grassroots governance technologies (Chinese invention patent, publication number: CN117271847A, title: Grassroots Governance Business Auxiliary Processing Method and System Based on Large Model) mainly rely on business auxiliary processing methods based on large models. These technologies build grassroots governance event models, train the models by combining event data with labeled tags, and classify and analyze events within the range where the models have been trained. Although this technology can improve the efficiency of event processing to a certain extent, there are still some obvious deficiencies: The model training of the existing technology depends on predefined rules and tags, which makes the system perform poorly when facing unforeseen complex events; due to the limitations of the model training cycle and data annotation, when the event types or data exceed the scope of model training, the response accuracy of the system will decrease significantly; Due to security reasons, the existing technology deploys the model in an intranet environment, restricting the model's ability to access Internet content; although this improves data security, it also leads to the lack of flexibility and update ability of the system when dealing with events related to external data; The existing technology mainly relies on the reasoning of static rules and models, lacking the ability of dynamic adjustment and optimization, which is particularly insufficient when events develop rapidly or the situation is complex and changeable. Summary of the invention

[0005] In view of the many problems existing in the above-mentioned prior art, the present invention provides a data processing method and a data processing platform for auxiliary decision-making in grass-roots governance. Through multi-level recursive optimization and real-time feedback mechanisms, combined with virtual environment simulation and multi-dimensional probability networks, the present invention forms a dynamic and highly flexible decision support system. The core lies in using advanced algorithms and data processing technologies to achieve full-process intelligence from the collection, processing, optimization to execution of event data. Finally, this system significantly improves the decision-making efficiency and accuracy in grass-roots governance, ensuring an efficient response to complex events.

[0006] A data processing method for auxiliary decision-making in grass-roots governance includes the following steps: Collect event data, perform preprocessing of the data, and classify and set priorities for the preprocessed event data according to the event type and urgency, generating high-priority event data; Allocate the high-priority event data to multiple distributed intelligent nodes. Each distributed intelligent node classifies the event, extracts features, and makes a preliminary decision on the high-priority event data. Preliminary decision data is generated through a transformed data model and a hybrid inference system, and then the preliminary decision data is optimized through a non-linear decision feedback mechanism to generate optimized preliminary decision data; Distribute the optimized preliminary decision data to models at multiple levels, and recursively call and collaboratively optimize the optimization results of each level model through a cross-level recursive optimization algorithm to generate cross-level recursive optimization data; Based on the cross-level recursive optimization data, create a virtual environment to simulate the event processing path, and optimize the path through a multi-dimensional probability network, finally generating optimal path data; Execute event processing based on the optimal path data, analyze the execution results and make adaptive adjustments, update the optimized decision data to the global knowledge base, and synchronize it to the distributed intelligent nodes.

[0007] Preferably, the data preprocessing includes removing noise from the event data, using a denoising algorithm to filter out environmental noise in the event data; converting the format of the event data to unify data from different sources into a standardized format; and extracting features from the event data, identifying and extracting key features through a feature extraction algorithm.

[0008] Preferably, the classification and priority setting steps are implemented through an algorithm based on the event type and urgency. Among them, the urgency is set by calculating the potential impact and processing time requirements of the event, and the generated high-priority event data contains a comprehensive index of time sensitivity and processing urgency.

[0009] Preferably, the transformed data model dynamically adjusts the event data features through the following calculation expression:

[0010] Among them, represents the transformed feature data; represents the transformation function related to the event type; represents the initial features of the event data; represents the parameters related to the event type features.

[0011] Preferably, the hybrid inference system includes a rule inference module and a statistical inference module. The rule inference module performs logical inference on structured data based on a predefined rule set, and the statistical inference module performs statistical analysis on unstructured data based on a probability model. The two are combined to generate preliminary decision data.

[0012] Preferably, the preliminary decision data is optimized through the following non-linear mapping function:

[0013] Among them, represents the optimized preliminary decision data; represents the non-linear mapping function; represents the preliminary decision data; represents the cumulative weight factor in the feedback loop.

[0014] Preferably, the optimized preliminary decision data is distributed to models at multiple levels, and the optimization results of the models at multiple levels are recursively called and co-optimized through the following recursive optimization expression:

[0015] Among them, represents the th recursively optimized data; represents the optimized data of the department model; represents the optimized data of the event type model; , and respectively represent the weight coefficients in the recursive optimization process.

[0016] Preferably, the virtual environment is created. The virtual environment performs simulation-driven learning based on historical data and real-time data, and performs probability inference and path optimization on the simulation results of the event processing path through a multi-dimensional probability network. Among them, the path optimization is selected through the following expression:

[0017] Among them, represents the probability value of the optimal path; represents the th simulation result probability of the path.

[0018] Preferably, the event processing is performed based on the optimal path data, the execution result data is collected in real time, the execution result is analyzed through an adaptive algorithm to generate calibration decision data, and the calibration decision data is input into the global knowledge base to update the decision rules and models in the global knowledge base, and the updated global knowledge base data is synchronized to all distributed intelligent nodes.

[0019] A system for implementing the grass-roots governance auxiliary decision data processing method, comprising: A data acquisition module, configured to acquire event data, remove noise, convert formats, and extract features from the event data to generate standardized event data; A data classification and priority setting module, configured to classify and set priorities for the standardized event data according to the event type and urgency to generate high-priority event data; A distributed intelligent node module, including a plurality of distributed intelligent nodes, configured to receive high-priority event data, classify events and extract features from the high-priority event data, and generate preliminary decision data through a transformation data model and a hybrid inference system, and then optimize the preliminary decision data through a non-linear decision feedback mechanism to generate optimized preliminary decision data; A cross-level recursive optimization module, configured to distribute the optimized preliminary decision data to models at multiple levels, and recursively call and collaboratively optimize the optimization results of each level model through a cross-level recursive optimization algorithm to generate cross-level recursive optimization data; An event cognition and simulation module, configured to create a virtual environment based on the cross-level recursive optimization data, simulate an event processing path, and optimize the path of the simulation result through a multi-dimensional probability network to generate optimal path data; An execution and feedback module, configured to perform event processing based on the optimal path data, collect and analyze the execution result data, perform adaptive adjustment, generate calibration decision data, and update the calibration decision data to the global knowledge base; A global knowledge base module, configured to store and update optimized decision data, and synchronize the updated data to a plurality of distributed intelligent nodes to ensure that the system can continuously optimize and respond to future event processing requirements.

[0020] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: Through the cross-level recursive optimization algorithm, the present invention realizes the coordination between different level models and the global optimal decision; this technology overcomes the problem of model limitations in the prior art and can flexibly adjust and optimize the decision path when facing complex events; Through a real-time feedback mechanism and an adaptive algorithm, the present invention realizes the dynamic optimization and continuous improvement of decisions; this enables the system to continuously optimize decisions according to the actual execution situation during the event processing, improving the accuracy of decisions and the execution effect. Through virtual environment simulation and a multi-dimensional probability network, the present invention realizes the efficient optimization of the event processing path; this technology not only solves the limitations of data interaction in the prior art, but also greatly improves the system's ability to perform multi-dimensional analysis and optimal path selection in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic flowchart of the method of the present invention; Figure 2 It is a schematic diagram of cross-level recursive optimization in the present invention; Figure 3 It is a schematic diagram of the virtual environment and the multi-dimensional probability network in the present invention; Figure 4 It is a schematic diagram of the synchronization between the global knowledge base and the distributed intelligent nodes in the present invention; Figure 5 It is a structural block diagram of the data processing platform of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present disclosure. In the following detailed description, for the sake of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, obviously, one or more embodiments can also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present disclosure.

[0023] The terms used herein are merely for describing specific embodiments and are not intended to limit the present disclosure. The terms "including", "comprising", etc. used herein indicate the presence of the described features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0024] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0025] As Figure 1 shown, a data processing method for grass-roots governance auxiliary decision-making includes the following steps: Collect event data, perform preprocessing on the data, and classify and set priorities for the preprocessed event data according to the event type and urgency to generate high-priority event data; Preferably, the data preprocessing includes removing noise from the event data, using a denoising algorithm to filter out environmental noise in the event data; performing format conversion on the event data to unify data from different sources into a standardized format; and extracting features from the event data, identifying and extracting key features through a feature extraction algorithm.

[0026] Noise removal is to clean up irrelevant or interfering information in the data to ensure the purity of the data. In the present invention, noise removal mainly deals with environmental noise in the event data. This is usually achieved by using denoising algorithms that can automatically identify and filter out interfering components in the data, such as background noise in surveillance videos or external electromagnetic interference in sensor data. The main effect of using noise removal is to improve the signal-to-noise ratio of the data, ensuring that subsequent processing steps are based on high-quality data. For example, when processing audio data rich in environmental noise, using Fourier transform for frequency domain analysis can effectively remove high-frequency noise and only retain the audio signals related to the event. This processing improves the accuracy of audio data feature extraction and provides a solid foundation for subsequent event classification and decision-making.

[0027] Format conversion is to unify data from different sources into a standardized format. In grass-roots governance, data may come from multiple heterogeneous sources, such as text, audio, video, and sensor data. The formats and structures of various types of data are different, and directly using these data for decision-making processing may lead to compatibility problems and low processing efficiency. Through format conversion, the system can convert all data into a standardized format to adapt to the requirements of subsequent processing modules. For example, sensor data is usually in a time series format, while text data may be in a natural language description. Through format conversion, these different types of data can be converted into a unified structured data format, facilitating unified processing and analysis by the system in subsequent steps.

[0028] Feature extraction is to further process the preprocessed data to extract key features in the data. The feature extraction algorithm analyzes the internal structure of the data and identifies the information most representative of event classification and decision-making. For example, when processing video data, a convolutional neural network (CNN) can be used to automatically extract the object contours and motion trajectories in the image. These features can help the system quickly identify the nature and possible development trends of the event. Through feature extraction, the system can convert a large amount of raw data into a compact feature set, greatly reducing the complexity of data processing while retaining the information crucial for event judgment.

[0029] Preferably, the classification and priority setting steps are implemented through an algorithm based on event type and urgency, wherein the urgency is set by calculating the potential impact of the event and the processing time requirement, and the generated high-priority event data contains a comprehensive indicator of time sensitivity and processing urgency.

[0030] The core of the classification step is to classify events into different categories based on their characteristics. These categories are usually based on predefined event types, such as public security events, environmental events, public health events, etc. Each event type has its own unique processing flow and resource requirements, so the classification algorithm automatically assigns events to corresponding categories based on relevant features of the event, such as text description, geographic location, time of event occurrence, etc. For example, when processing a large amount of municipal alarm data, the system can use natural language processing (NLP) technology to extract keywords from event descriptions, combined with historical data analysis, to automatically identify whether this is a public security incident or a public safety hazard. This classification based on event type not only helps to improve the system's response speed, but also ensures the rational allocation of resources so that different types of events are handled in the most appropriate way.

[0031] After the classification is completed, the priority setting step further distinguishes the order of handling each event by calculating the urgency of the event. The assessment of urgency is mainly based on two core factors: the potential impact of the event and the processing time requirement. The potential impact assessment involves quantifying the possible consequences of the event, such as casualties, property losses, social stability, etc.; the processing time requirement is to estimate how long the event must be responded to in order to avoid adverse consequences based on the nature of the event and historical data. The urgency is usually calculated by a comprehensive algorithm that combines the potential impact and processing time requirements to generate a comprehensive urgency index. For example, for a fire alarm, the system will comprehensively consider the potential danger area of the fire (such as whether it is close to residential areas), the current meteorological conditions (such as wind speed and direction), and the time it takes for rescue forces to reach, and quickly calculate the urgency index of the event. The system will then sort the events according to this index and prioritize those with the highest urgency index.

[0032] The generated high-priority event data contains comprehensive indicators of time sensitivity and processing urgency. Time sensitivity refers to the sensitivity of an event to changes in external conditions within a specific time window, which directly affects the timeliness of event processing. Processing urgency reflects the serious consequences that may arise if an event is not handled in a timely manner. The high-priority event data generated by these two indicators ensures that when facing multiple concurrent events, the system can prioritize those events that require immediate response and reduce the negative impact of delays.

[0033] In one embodiment, the application scenario of the grass-roots governance system in response to emergencies. For example, in an extreme weather warning system, the system receives a large number of alarm messages, including waterlogging, tree collapse, power outage, etc. The system first classifies these events into different types according to the descriptions of the alarm messages, and then calculates the urgency of each event. For example, a waterlogging event may be given a higher emergency index due to the potential risks of traffic paralysis and people being trapped, while the emergency index of a power outage event is set according to the population density of the affected area and the importance of power supply. Finally, the system gives priority to handling the waterlogging event, dispatches emergency rescue forces to quickly reach the scene, and the power department will handle the power outage problem later.

[0034] Allocate high-priority event data to multiple distributed intelligent nodes. Each distributed intelligent node classifies the events, extracts features, and makes a preliminary decision on the high-priority event data. Generate preliminary decision data through a transformed data model and a hybrid inference system, and then optimize the preliminary decision data through a non-linear decision feedback mechanism to generate optimized preliminary decision data; The allocation of distributed intelligent nodes is the starting point of this step. High-priority event data is first distributed to multiple distributed intelligent nodes, which can be located in different physical locations or logically represent different processing units in the system. The advantage of distributed processing is that it can improve the parallel processing ability of the system, avoid the overload of a single node, and thus improve the overall processing speed and the fault tolerance of the system. In a practical application scenario, such as an emergency response system in a city, the distributed intelligent nodes may be distributed in different areas of the city, and each node is responsible for processing the high-priority event data within its area. In this way, when multiple emergency events occur simultaneously, the system can reasonably allocate tasks among the nodes to ensure a quick response.

[0035] On each distributed intelligent node, event classification and feature extraction are first carried out. The classification step further refines the events, which may be reclassified based on more specific local conditions or more detailed event descriptions. Feature extraction is to analyze the event data in a higher dimension and extract the most useful feature information for subsequent decision-making. The process of feature extraction may involve various technical means, such as feature extraction algorithms based on deep learning, which can automatically identify key patterns and relationships from the event data. For example, when processing surveillance video data, the intelligent node can extract the action features of the people in the event through a convolutional neural network (CNN) to better judge the nature of the event and its possible subsequent development.

[0036] After classification and feature extraction are completed, the intelligent node enters the preliminary decision-making stage. The preliminary decision is based on the feature data extracted in the previous steps, and uses a transformed data model and a hybrid inference system to make a preliminary judgment on the event and give response suggestions. The transformed data model refers to dynamically adjusting the parameters of the model according to the characteristics of different types of events to adapt to different event types and situations. This dynamic adjustment can make the model more flexible and adaptable, thus improving the accuracy of the preliminary decision. The hybrid inference system combines the advantages of rule-based reasoning and statistical reasoning. The former can perform logical judgments based on a predefined rule set, while the latter performs uncertainty analysis based on historical data and probability models. This hybrid reasoning method enables the system to quickly process events with relatively high certainty and also make relatively reliable preliminary judgments when facing uncertainties.

[0037] After the preliminary decision data is generated, the system optimizes these preliminary decision data through a non-linear decision feedback mechanism. The principle of the non-linear decision feedback mechanism is to dynamically adjust the weight factors in the feedback loop, combine the deviation between the actual feedback and the model output, and correct and optimize the preliminary decision data. This step realizes the fine-tuning of the decision-making path through a series of non-linear mapping functions to reduce errors and improve the accuracy of the decision. For example, in the case of an emergency evacuation, the preliminary decision may recommend a certain evacuation route. However, after feedback analysis, it is found that the actual evacuation time of this route exceeds the expectation. The system will readjust the weight factors through the non-linear decision feedback mechanism to optimize a new and more effective evacuation route.

[0038] In one embodiment, in a certain complex emergency event management system. When a high-priority fire alarm is triggered, the system quickly distributes the alarm data to several intelligent nodes near the fire location. These nodes further classify and extract features based on information such as the specific location, time, and weather conditions of the fire, such as extracting the speed of fire spread and the density of surrounding buildings. Subsequently, the system combines these features, dynamically adjusts the parameters of the decision model through the transformed data model, and gives preliminary decision suggestions through the hybrid inference system, such as formulating a preliminary personnel evacuation plan. Then, the system will make real-time adjustments to the evacuation plan according to on-site feedback, such as the actual evacuation time and personnel flow, through the non-linear decision feedback mechanism to ensure the safety and efficiency of the final evacuation route.

[0039] Preferably, the transformed data model dynamically adjusts the event data features through the following calculation expression:

[0040] where represents the transformed feature data; represents the transformation function related to the event type; Represents the initial features of event data; Represents the parameters related to the event type features.

[0041] The transformed data model enhances the adaptability of the decision-making model to different types of events by dynamically adjusting data features. First, the initial features of event data are the basic information extracted from the original data. These features usually contain the basic descriptions of the events, such as time, location, number of participants, etc. However, simply relying on these initial features is not sufficient to accurately handle different types of complex events. Therefore, a transformation function is introduced, which weights and adjusts the initial features according to the specific type of the event (such as natural disasters, social security events, public health events, etc.), making these features show higher relevance and importance in the processing of different types of events.

[0042] The transformation function is usually designed based on historical data and expert knowledge, and it can be automatically adjusted during the decision-making process. For example, for a fire event, the transformation function may amplify the feature data related to the spread of the fire (such as wind speed, humidity), while for a traffic accident, the transformation function may pay more attention to features such as road conditions and traffic flow. Through such dynamic adjustment, the system can flexibly adjust the data input according to the different types of events, ensuring that the system receives the most relevant and decision-making valuable data when processing each event.

[0043] In this model, the parameter acts as a correction factor to fine-tune the transformed feature data. The parameter is usually related to the specific features of the event and can further refine the data according to the uniqueness of the event. For example, in a certain event, in addition to the conventional features of the number of infections and the transmission route, the parameter can also reflect the distribution of local medical resources or the protection awareness of the masses, and these features can have an important impact on the final decision.

[0044] The introduction of the transformed data model greatly improves the flexibility and accuracy of the system. By dynamically adjusting the initial features, the system can generate more accurate feature data according to different event types , thus improving the processing effect of the decision-making model on different events. This flexibility ensures that the system can maintain an efficient and accurate response ability when facing complex and changeable events.

[0045] In one embodiment, it is an emergency event processing scenario in an urban management system. Assume that in a situation where multiple disasters occur simultaneously, the system needs to handle different types of events such as earthquakes, fires, and floods at the same time. For an earthquake event, the transformed data model It may increase the weights of geological features, such as earthquake intensity and focal depth. For fire incidents, the system may pay more attention to features related to fire spread, such as wind speed and building density. Through dynamic adjustment, the feature data of these incidents will be more in line with actual needs, ensuring that when the system analyzes and makes decisions, it can be based on the most relevant data and finally formulate the most effective response measures.

[0046] Preferably, the hybrid inference system includes a rule inference module and a statistical inference module. The rule inference module performs logical inference on structured data based on a predefined rule set, and the statistical inference module performs statistical analysis on unstructured data based on a probability model. The two are combined to generate preliminary decision-making data.

[0047] The rule inference module is an integral part of the hybrid inference system, and its function is to perform logical inference on structured data. Structured data usually includes formatted and organized data information, such as records in a database, data read by sensors, or predefined event information. The rule inference module performs inference operations on this data based on a predefined rule set, which is usually formulated by domain experts based on years of experience and historical data and contains the processing logic for specific events. For example, in emergency management, the rules for fire incidents may include: if the fire spread speed exceeds a certain threshold, then prioritize the evacuation of people in the affected area. This rule-based inference method can quickly reach clear decisions, especially suitable for event processing scenarios with clear logic and conditions.

[0048] Different from the rule inference module, the statistical inference module mainly processes unstructured data, such as text, images, videos, etc. This type of data is usually more complex and difficult to process through simple rules. The statistical inference module performs statistical analysis on unstructured data through a probability model. The probability model can infer the occurrence probability or trend of certain events by learning a large amount of historical data. For example, when processing social media data, the system can analyze the reaction trend of the public to an event through the statistical inference module to judge the urgency or possible development direction of the event. The advantage of statistical inference is that it can process data with a high degree of uncertainty and provide a basis for decision-making through probability analysis, enabling the system to still make reasonable judgments when facing fuzzy or unknown situations.

[0049] The core of the hybrid reasoning system lies in combining the advantages of these two modules. It can not only perform precise rule-based reasoning on structured data but also conduct comprehensive statistical analysis on unstructured data, thereby generating more comprehensive preliminary decision-making data. This combination method ensures the flexibility and accuracy of the system when dealing with different types of data. For example, in an emergency response scenario, when the system receives an alarm message about a potential public safety incident, the rule-based reasoning module can quickly determine whether immediate action is required based on existing rules, while the statistical reasoning module can further analyze relevant social media data to evaluate the public's attention and the spread speed of the incident. By combining the results of these two types of reasoning, the system can arrive at a more comprehensive preliminary decision, such as immediately initiating an emergency response or conducting further observation first.

[0050] In one embodiment, it is a scenario of handling major public events in urban management. For example, a certain urban management system receives an alarm message about a subway accident. The rule-based reasoning module can quickly come up with a preliminary response plan, such as closing relevant lines and evacuating passengers, based on the structured data of the accident, such as subway train numbers, the number of affected passengers, and the location of the accident. At the same time, the statistical reasoning module can evaluate the social impact and spread speed of the event by analyzing relevant discussions on social media. If the statistical analysis results show that the event has attracted a large amount of attention in a short period and the public reaction is strong, the system may adjust the preliminary decision and initiate a higher-level emergency response to prevent the event from further escalating.

[0051] Preferably, the following non-linear mapping function is used to optimize the preliminary decision-making data:

[0052] where represents the optimized preliminary decision-making data; represents the non-linear mapping function; represents the preliminary decision-making data; represents the cumulative weight factor in the feedback loop.

[0053] In the present invention, the introduction of the non-linear decision feedback mechanism plays a key role in improving the accuracy and response speed of decision-making. This mechanism optimizes the preliminary decision-making data through a non-linear mapping function to ensure that the final decision can better adapt to the actual situation and environmental changes.

[0054] In this mechanism, the preliminary decision-making data It is generated based on the decision-making module in the early stage of the system and contains preliminary response suggestions for the current event. However, during the actual execution process of this preliminary decision, it is often affected by various dynamic factors, such as changes in the external environment, real-time availability of resources, and further development of the event. Therefore, relying solely on the preliminary decision data for actions may not be able to handle complex and changing real-world scenarios. At this time, the non-linear mapping function plays a crucial role.

[0055] The non-linear mapping function is a complex mathematical model used to perform non-linear transformation on the input data (i.e., the sum of the preliminary decision data and the cumulative weight factor). The purpose of this non-linear transformation is to adjust the decision according to the non-linear relationship of the data to make it more in line with the actual needs. Compared with the linear model, the non-linear mapping function can handle more complex data relationships and can more accurately reflect the changes in the real world. For example, when dealing with a public health emergency, the preliminary decision data may be a linear prediction based on the current number of infections, but the actual spread of infections often shows non-linear characteristics (such as exponential growth). Through the non-linear mapping function, the system can more accurately adjust the prediction of the spread of infections and optimize resource allocation and response strategies.

[0056] The cumulative weight factor is the feedback component in this mechanism. The cumulative weight factor in the feedback loop represents the empirical data accumulated by the system through real-time monitoring and analysis during the decision execution process. These empirical data may include previous decision results, environmental changes, and the actual development of the event. Over time, the cumulative weight factor gradually reflects the importance of these feedback information, thus playing a role in correction and optimization in the non-linear mapping function. For example, in an urban emergency evacuation plan, the initial decision may be based on standard traffic flow data, but as the actual evacuation progresses, the system will continuously collect actual traffic data through the feedback loop and adjust the cumulative weight factor , enabling the decision to be optimized in real time and ultimately obtaining a more reasonable and effective evacuation plan.

[0057] Through the non-linear decision feedback mechanism, the system can perform real-time decision optimization when facing a complex and dynamic environment. This process ensures that the system not only relies on the initial static decision but also can make dynamic adjustments according to the feedback during actual operation. Ultimately, this mechanism can greatly improve the response speed and decision accuracy of the system, avoiding inefficient or even incorrect decisions caused by the inconsistency between the preliminary decision and the actual situation.

[0058] In one embodiment, it is applied in the urban flood control emergency management system. When the system receives a flood warning, the generated preliminary decision data It may include the areas recommended for evacuation and the allocation of required resources. However, with the actual development of the flood, the initial decision may need to be adjusted. Through a non-linear decision feedback mechanism, the system can monitor factors such as the water level change, wind speed, rainfall of the flood in real time, and continuously adjust the cumulative weight factor through the feedback loop . At this time, the non-linear mapping function will re-optimize the evacuation decision according to these latest data, such as adjusting the evacuation route, increasing or decreasing the allocation of resources, etc. Finally, the system uses the optimized decision data to guide flood control emergency operations and ensure that the city can make the most effective response when dealing with sudden floods.

[0059] Distribute the optimized preliminary decision data to models at multiple levels, and recursively call and co-optimize the optimization results of models at each level through a cross-level recursive optimization algorithm to generate cross-level recursive optimization data; Preferably, the optimized preliminary decision data is distributed to models at multiple levels, and the optimization results of models at multiple levels are recursively called and co-optimized through the following recursive optimization expression:

[0060] where, represents the th recursive optimization data; represents the optimization data of the department model; represents the optimization data of the event type model; , and respectively represent the weight coefficients in the recursive optimization process.

[0061] Cross-level recursive optimization is a highly complex and crucial step for integrating and optimizing decision models at multiple levels. The main purpose of this step is to enable models at different levels to coordinate with each other through recursive calls and co-optimization, so as to generate more accurate and globally optimal decision data.

[0062] The cross-level recursive optimization algorithm distributes the optimized preliminary decision data to models at multiple levels, and recursively calls and co-optimizes among these models to gradually improve the overall effect of the decision, as Figure 2 shown. Here, "level" can be understood as different decision dimensions or modules in the system, such as department models and event type models, etc. Each level model independently performs optimization and generates its corresponding optimization data. However, relying solely on the optimization results of a certain level may lead to local optimality rather than global optimality of the decision. Therefore, the cross-level recursive optimization algorithm recursively calls among different level models, repeatedly adjusts and optimizes the results of each level to ensure the global consistency and optimality of the decision.

[0063] In , As the core data for recursive calls, it is continuously adjusted and optimized in each round of recursion. Department model optimization data and event type model optimization data respectively represent the output results of different-level models in the current optimization round. These results are combined through weight coefficients , and to generate new recursive optimization data . The weight coefficients may be dynamically adjusted during the recursive process to adapt to the importance and influence of different-level models. For example, in some cases, the results of the department model may be more important than those of the event type model, and the system will accordingly adjust the weight coefficients and to highlight the influence of the department model.

[0064] The main advantage of cross-level recursive optimization is that it can optimize decisions globally, ensuring that different-level models can work together and thus avoiding the limitations of local optimal solutions. For example, in grass-roots governance, one department may focus on optimizing resource allocation, while another department focuses on event classification and priority setting. Through cross-level recursive optimization, these different-level decisions can be coordinated with each other, and ultimately a global optimal decision that comprehensively considers the effective utilization of resources, time urgency, and event impact is generated.

[0065] In terms of effects, cross-level recursive optimization ensures the global consistency and overall optimization of decisions, which is particularly important when dealing with complex events. The system can not only achieve optimization at the local level but also integrate the optimization results of each level through recursive calls into a more refined and comprehensive decision-making scheme. This recursive optimization process improves the decision-making efficiency and accuracy of the system. Especially when facing complex events with multiple dimensions and variables, it can significantly enhance the response ability and decision-making effects of the system.

[0066] In one embodiment, in a comprehensive emergency management system, when facing a large-scale natural disaster, the system needs to coordinate multiple departments (such as fire, medical, transportation) and event types (such as fire, flood, earthquake) to respond. The preliminary decision-making data is optimized and then distributed to models at each level, such as the resource allocation model of the fire department and the evacuation model of the traffic management department. Each department independently conducts optimization to generate optimization results and Through the cross - level recursive optimization algorithm, the system will gradually adjust these results to make the decisions of all departments coordinated with each other, ensuring that resources and actions reach an optimal allocation globally. As the recursive calls proceed, the system finally generates a comprehensive response plan to ensure that all departments can cooperate efficiently and minimize the losses caused by disasters.

[0067] As Figure 3 shown, based on the cross - level recursive optimization data, a virtual environment is created to simulate the event - handling path, and the path is optimized through a multi - dimensional probability network, finally generating the optimal path data; Preferably, for the creation of the virtual environment, the virtual environment performs simulation - driven learning based on historical data and real - time data, and performs probability inference and path optimization on the simulation results of the event - handling path through a multi - dimensional probability network. Among them, the path optimization is selected through the following expression:

[0068] Among them, represents the probability value of the optimal path; represents the probability of the simulation result of the

[0069] In the present invention, the simulation of the event - handling path and path optimization based on the cross - level recursive optimization data are the key steps for the system to determine the optimal action plan. This process simulates by creating a virtual environment and combines a multi - dimensional probability network to infer and optimize the event - handling path, thereby generating the optimal path data to support the execution of the decision.

[0070] This process first relies on the cross - level recursive optimization data, which are generated by the recursive optimization algorithm in the previous step and represent the optimal combined output of different - level models. Based on these data, the system creates a virtual environment, which is not just a simple simulation space but a dynamic, digital scenario based on real - world conditions. The virtual environment performs simulation - driven learning based on historical data and real - time data, that is, it extracts patterns from the historical processing history of past events and combines the currently obtained real - time data to simulate the possible future development paths of the event.

[0071] The simulation-driven learning conducted in a virtual environment is achieved through advanced machine learning algorithms, which can continuously adjust environmental parameters according to the input of data to more accurately simulate various possible development scenarios of events. For example, when simulating a flood event, the virtual environment will combine historical flood data (such as water level, rainfall, terrain features, etc.) and current real-time meteorological data to predict the possible development path and impact range of the flood. This simulation not only provides foresight into the possible outcomes of events but also offers rich data support for subsequent path optimization.

[0072] The role of the multi-dimensional probability network is crucial in this process. The multi-dimensional probability network is a complex mathematical model used to calculate and infer the success probabilities of different event processing paths. The simulation results of multiple event processing paths generated in the virtual environment will be input into the multi-dimensional probability network, and the system will conduct probability analysis based on the characteristics of these paths to evaluate the possible success rate of each path in actual execution. The core of path optimization lies in selecting the optimal path, that is, the path with the highest success probability, to ensure that the final decision-making plan can maximize the achievement of the expected goal.

[0073] means selecting the path with the highest probability value among all simulation results as the optimal path. This optimization process makes full use of the advantages of probability analysis. By evaluating the feasibility and potential success rate of different paths under actual conditions, the system can select the path that is most likely to succeed. For example, in the decision-making of urban emergency evacuation, the system may generate multiple simulation results of evacuation routes, and path optimization will select the route with the lowest probability of being blocked and the fastest passing speed as the final evacuation path to maximize the safety of personnel.

[0074] Through this path optimization based on cross-level recursive optimization data, virtual environment simulation, and multi-dimensional probability network, the system can greatly improve the accuracy and operability of decision-making. It not only provides a systematic method to predict and select the optimal action plan but also can cope with complex and changeable real-world conditions to ensure a high success rate of decision-making during the execution process.

[0075] In one embodiment, there is an emergency response scenario in a large-scale public emergency. Suppose a city faces an unexpected situation and the system needs to formulate an optimal evacuation plan for people. First, the system generates simulation results of multiple evacuation routes based on cross-level recursive optimization data. The simulation of these routes takes into account the handling experience of similar events in history, as well as the environmental and personnel flow data collected in real time. Subsequently, the system inputs these simulation results into a multi-dimensional probability network for probability analysis of the routes. Finally, through an optimization expression, an evacuation route with the highest success rate is selected, and the system pushes this route as optimal decision-making data to the emergency response team to ensure that the evacuation operation can be carried out quickly and efficiently, minimizing casualties to the greatest extent.

[0076] Execute event processing based on the optimal path data, analyze the execution results and make adaptive adjustments, update the optimized decision-making data to the global knowledge base, and synchronize it to the distributed intelligent nodes.

[0077] As Figure 4 shown, preferably, the event processing is executed based on the optimal path data, the execution result data is collected in real time, the execution results are analyzed through an adaptive algorithm to generate corrected decision-making data, and the corrected decision-making data is input into the global knowledge base to update the decision rules and models in the global knowledge base, and the updated global knowledge base data is synchronized to all distributed intelligent nodes.

[0078] In the framework of the present invention, executing event processing based on the optimal path data and real-time analysis and adaptive adjustment are important steps to ensure the continuous optimization of the system in actual operation. The core of this process lies in using the optimal path data generated by the previous optimization to execute specific event processing, and at the same time, through real-time monitoring and analysis of the execution results, dynamically adjusting the decision-making, and updating the optimized decision-making to the global knowledge base to ensure that the system can continuously learn and improve.

[0079] After the system generates the optimal path data, it first uses this path data to guide the specific execution of event processing. This step means that the system puts the previously derived optimal decision-making plan into actual operation. For example, in an emergency evacuation, people are evacuated according to the optimized route. The execution process at this stage involves multiple real-time variables, such as the movement speed of people, environmental changes, resource consumption, etc. These variables may deviate from the initial expectations, so the system needs to monitor in real time.

[0080] Real-time collection of execution result data is a key part of this process. The system obtains various types of data generated during the event processing in real time through channels such as sensors, monitoring devices, and communication systems. For example, during the evacuation of people, the system will collect information such as crowd density, traffic flow, and evacuation speed in real time. These data provide the basis for subsequent analysis and decision-making adjustment.

[0081] The system analyzes the execution result data collected in real time through an adaptive algorithm. The core of the adaptive algorithm is its ability to dynamically adjust decision parameters according to real-time data, thereby optimizing the current decision-making path. The algorithm adjusts weights and corrects models based on the differences between the actual execution results and the expected results, and then generates corrected decision data. For example, if the system finds that the actual passing time of a certain evacuation route is longer than expected, perhaps due to a sudden traffic jam, the adaptive algorithm will automatically analyze the cause and suggest changing the evacuation route or adjusting the evacuation speed to ensure the smooth progress of the overall evacuation plan.

[0082] After generating the corrected decision data, the system inputs this data into the global knowledge base. The global knowledge base is the core data storage and knowledge management platform of the system. It records all historical decisions, model parameters, rule sets, and related optimization data. By inputting the latest corrected decision data into the global knowledge base, the system can continuously enrich and improve its knowledge system, enabling the system to better utilize past experience data when facing similar events and enhancing the quality and efficiency of decision-making.

[0083] In addition, synchronizing the updated global knowledge base data to all distributed intelligent nodes is a crucial step in ensuring the coordinated operation of the entire system. Distributed intelligent nodes rely on the latest data and models in the global knowledge base when executing tasks. Therefore, each time the knowledge base is updated, the system synchronizes these updates to each intelligent node, enabling them to operate based on the latest optimized decisions when handling similar events subsequently. This synchronization process ensures the overall consistency and coordination of the system, avoiding decision-making errors or inefficiencies caused by data inconsistencies between different nodes.

[0084] Through this process, the system can collect and analyze data in real time during the actual event handling process, make adaptive adjustments, and thus ensure the accuracy and effectiveness of decision-making. This continuous feedback and adjustment mechanism not only improves the success rate of the current event handling but also continuously optimizes the overall performance of the system, enabling the system to handle increasingly complex and variable actual situations during long-term operation.

[0085] In one embodiment, an application scenario of an urban traffic management system in response to sudden natural disasters. Suppose that after an earthquake, the system generates the optimal traffic evacuation route and uses this route data to guide the traffic evacuation in the urban area. As the evacuation progresses, the system monitors the traffic flow, congestion situation, and average vehicle speed on each road in real time. When the system detects that a certain main road is severely congested due to a traffic accident, the adaptive algorithm immediately analyzes and generates correction decision data, suggesting to re-plan the evacuation route or divert part of the traffic flow. Subsequently, these correction data are input into the global knowledge base and synchronized to other distributed intelligent nodes, enabling the entire traffic system to adjust its strategy within a short time to ensure the smooth implementation of the evacuation plan and the safety of personnel.

[0086] As Figure 5 shown, a data processing platform for implementing the data processing method for grass-roots governance auxiliary decision-making includes: A data collection module, which is used to collect event data, remove noise, convert the format, and extract features from the event data to generate standardized event data; this module is responsible for collecting event data from multiple sources and performing preliminary processing on these data. First, interference information is filtered out through noise removal technology to ensure the purity of the data. Then, data from different sources are unified into a standardized format using format conversion technology for subsequent processing. Finally, key features are identified and extracted through feature extraction algorithms to generate standardized event data. Through these steps, the system can ensure the quality and consistency of the input data, laying a foundation for subsequent decision-making processing. A data classification and priority setting module, which is used to classify and set priorities for the standardized event data according to the event type and urgency, generating high-priority event data; this module classifies the standardized event data according to the event type and urgency and sets priorities. The classification process relies on predefined rules and algorithms to classify events into different types, such as natural disasters, social security events, etc. Priority setting generates high-priority event data by calculating the potential impact of the event and the processing time requirements. In this way, the system can give priority to processing those events that have the greatest impact on grass-roots governance, ensuring the reasonable allocation and use of resources.

[0087] Distributed intelligent node module, including multiple distributed intelligent nodes, which is used to receive high-priority event data, classify events and extract features from the high-priority event data, and generate preliminary decision data through a transformed data model and a hybrid inference system. Subsequently, the preliminary decision data is optimized through a non-linear decision feedback mechanism to generate optimized preliminary decision data; this module includes multiple intelligent nodes, which are used to receive and process high-priority event data. Each node independently conducts event classification and feature extraction, and generates preliminary decision data through a transformed data model and a hybrid inference system. Subsequently, the system uses a non-linear decision feedback mechanism to optimize the preliminary decision data to generate more effective optimized preliminary decision data. The core of this module lies in the ability of distributed processing, which can quickly respond and make preliminary decisions in the case of large-scale and multi-event concurrency.

[0088] Cross-level recursive optimization module, which is used to distribute the optimized preliminary decision data to models at multiple levels, and recursively call and collaboratively optimize the optimization results of models at each level through a cross-level recursive optimization algorithm to generate cross-level recursive optimization data; this module distributes the optimized preliminary decision data to models at multiple levels through a recursive optimization algorithm, and recursively calls and collaboratively optimizes among these models. Finally, cross-level recursive optimization data is generated. The key to this process lies in the collaborative work of models at each level to ensure the global optimality of decisions and avoid the limitations of local optimality.

[0089] Event cognition and simulation module, which is used to create a virtual environment based on the cross-level recursive optimization data, simulate the event processing path, and optimize the path of the simulation results through a multi-dimensional probability network to generate optimal path data; this module creates a virtual environment based on the cross-level recursive optimization data and simulates the event processing path in this environment. The path of the simulation results is optimized through a multi-dimensional probability network to generate optimal path data. The virtual environment simulation combines historical data and real-time data, can accurately predict the development trend of events, and selects the processing path with the highest success rate through probability inference, providing strong support for the final decision.

[0090] Execution and feedback module, which is used to execute event processing based on the optimal path data, collect and analyze the execution result data, conduct adaptive adjustment, generate corrected decision data, and update the corrected decision data to the global knowledge base; after obtaining the optimal path data, the system executes event processing based on these data and collects the execution result data in real time. By analyzing the execution results through an adaptive algorithm, the system generates corrected decision data and updates these data to the global knowledge base. The core of this module lies in real-time feedback and adaptive adjustment to ensure that the system can dynamically optimize decisions during the event processing process, improving the accuracy and response speed of decisions.

[0091] The global knowledge base module is used to store and update the optimized decision-making data, and synchronize the updated data to multiple distributed intelligent nodes to ensure that the system can continuously optimize and meet the future event processing requirements. The global knowledge base is used to store and update all the optimized decision-making data and synchronize this data to multiple distributed intelligent nodes. Through this synchronization mechanism, the system can continuously optimize, ensuring that when dealing with future events, each node can use the latest decision-making models and rules, thereby improving the overall response ability and decision-making quality of the system.

[0092] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects.

[0093] The above are only the embodiments of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A data processing method for grass-roots governance auxiliary decision-making based on a non-linear mapping function, characterized in that, Including the following steps: Collect event data, perform preprocessing of the data, and classify and set priorities for the preprocessed event data according to event types and urgency levels to generate high-priority event data; Allocate the high-priority event data to multiple distributed intelligent nodes. Each distributed intelligent node performs event classification, feature extraction, and preliminary decision-making on the high-priority event data, generates preliminary decision data through a transformed data model and a hybrid inference system, and then optimizes the preliminary decision data through a non-linear decision feedback mechanism to generate optimized preliminary decision data; the transformed data model is used to dynamically adjust the parameters of the model according to different types of event characteristics to adapt to different event types and scenarios; The non-linear decision feedback mechanism corrects and optimizes the preliminary decision data by dynamically adjusting the weight factors in the feedback loop and combining the deviation between the actual feedback and the model output; Distribute the optimized preliminary decision data to models at multiple levels, and perform recursive calls and collaborative optimization on the optimization results of models at each level through a cross-level recursive optimization algorithm to generate cross-level recursive optimization data; the cross-level recursive optimization algorithm makes recursive calls between models at different levels and repeatedly adjusts and optimizes the results at each level; Based on the cross-level recursive optimization data, create a virtual environment to simulate the event processing path, and infer and optimize the event processing path through a combined multi-dimensional probability network, and finally generate optimal path data; Execute event processing based on the optimal path data, analyze the execution results and perform adaptive adjustment, update the optimized decision data to the global knowledge base, and synchronize it to the distributed intelligent nodes; The transformed data model dynamically adjusts the event data characteristics through the following calculation expression: Among them, represents the transformed feature data; represents the transformation function related to the event type; represents the initial feature of the event data; represents the parameter related to the event type feature; Optimize the preliminary decision data through the following non-linear mapping function: Among them, represents the optimized preliminary decision data; represents the non-linear mapping function; represents the preliminary decision data; represents the cumulative weight factor in the feedback loop.

2. The method for processing grass-roots governance auxiliary decision-making data based on a non-linear mapping function according to claim 1, wherein, The data preprocessing includes removing noise from the event data and filtering out environmental noise in the event data using a denoising algorithm; performing format conversion on the event data to unify data from different sources into a standardized format; performing feature extraction on the event data and identifying and extracting key features through a feature extraction algorithm.

3. The method for processing grass-roots governance auxiliary decision-making data based on a non-linear mapping function according to claim 1, wherein The classification and priority setting steps are implemented through an algorithm based on event types and urgency levels. Among them, the urgency level is set by calculating the potential impact and processing time requirements of the event, and the generated high-priority event data contains a comprehensive index of time sensitivity and processing urgency.

4. The method for processing grass-roots governance auxiliary decision-making data based on a non-linear mapping function according to claim 1, wherein The hybrid inference system includes a rule inference module and a statistical inference module. The rule inference module performs logical inference on structured data based on a predefined rule set, and the statistical inference module performs statistical analysis on unstructured data based on a probability model. The two are combined to generate preliminary decision data.

5. The method for processing grass-roots governance auxiliary decision-making data based on a non-linear mapping function according to claim 1, characterized in that The optimized preliminary decision data is distributed to models at multiple levels, and the optimization results of models at multiple levels are recursively called and collaboratively optimized through the following recursive optimization expression: Among them, represents the th recursively optimized data; represents the optimized data of the department model; represents the optimized data of the event type model; , and respectively represent the weight coefficients in the recursive optimization process.

6. The method for processing grass-roots governance auxiliary decision-making data based on a non-linear mapping function according to claim 1, wherein, Create the virtual environment. The virtual environment performs simulation-driven learning based on historical data and real-time data, and performs probability inference and path optimization on the simulation results of the event processing path through a multi-dimensional probability network. Among them, the path optimization is selected through the following expression: Among them, represents the probability value of the optimal path; represents the simulation result probability of the 7. The method for processing grass-roots governance auxiliary decision-making data based on a non-linear mapping function according to claim 1, wherein Execute event processing based on the optimal path data, collect execution result data in real time, analyze the execution results through an adaptive algorithm, generate correction decision data, and input the correction decision data into the global knowledge base to update the decision rules and models in the global knowledge base, and synchronize the updated global knowledge base data to all distributed intelligent nodes.

Images