Property management method and system based on artificial intelligence

Abstract

The invention provides a property management method based on artificial intelligence, and the method comprises the steps: collecting local data through employing a property facility sensor network and a resident interaction platform, so as to generate integrated data, generating a problem list and an opportunity point set through employing an artificial intelligence decision-making system constructed through a deep reinforcement learning algorithm, and carrying out the decision-making of the problem list and the opportunity point set. And selecting an optimal plan from a preset emergency response plan library by using a Bayesian optimization algorithm, obtaining information transmission content by applying a natural language generation technology and an emotion analysis model, and searching different management strategy combinations by using a genetic algorithm to generate an optimal management strategy and a specific implementation plan. Meanwhile, a feedback mechanism based on the block chain technology is established, the management scheme is adjusted by using an online learning algorithm, and an optimal property management mode is obtained. According to the technical scheme provided by the invention, the accuracy and response speed of problem identification are improved, the effectiveness and affinity of communication with residents are enhanced, and the high efficiency and adaptability of the property management mode are ensured.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, and in particular, to a property management method based on artificial intelligence. Background Art

[0002] With the acceleration of the urbanization process and the improvement of residents' living quality requirements, property management is facing increasingly complex challenges. Traditional property management methods rely on manual inspections and data collection at fixed time intervals, making it difficult to respond promptly to emergencies such as facility failures and security issues, and there are problems of low efficiency in resource allocation and service optimization. In addition, traditional management methods often lack personalized attention to residents' needs, resulting in poor communication and low service satisfaction. To address these challenges, artificial intelligence (AI) technology has gradually been introduced into property management, real-time monitoring the property status through a smart sensor network, and realizing predictive maintenance and optimizing management processes by combining data analysis and machine learning algorithms.

[0003] Although existing property management systems have been able to use information technology to improve the management level to a certain extent, there are still some significant deficiencies. First, most existing systems focus on post-event processing rather than preventive maintenance, with limited ability to detect and warn of potential problems early, and unable to effectively avoid the occurrence of emergencies. Second, current management systems usually adopt static rules or simple automated processes, making it difficult to adapt to the changing environment and personalized needs of residents, which limits the flexibility and pertinence of services. Finally, the lack of an effective feedback mechanism and dynamic adjustment ability makes it difficult to quickly optimize the management plan according to the actual situation once it is formulated, affecting the overall management efficiency and service quality. Summary of the Invention

[0004] The embodiments of the present invention provide a property management method and system based on artificial intelligence to solve the problems in the prior art, including limited ability to detect and warn of potential problems early, inability to effectively avoid the occurrence of emergencies, difficulty in adapting to the changing environment and personalized needs of residents, lack of an effective feedback mechanism and dynamic adjustment ability, making it difficult to quickly optimize the management plan according to the actual situation once it is formulated, and affecting the overall management efficiency and service quality.

[0005] In a first aspect, the embodiments of the present invention provide a property management method based on artificial intelligence, including:

[0006] Collecting multi-source heterogeneous real-time status information and requests through a property facility sensor network and a resident interaction platform, preliminarily processing local data through an edge computing node, and implementing data routing and integration based on a distributed hash table to generate integrated data;

[0007] An artificial intelligence decision-making system constructed using a deep reinforcement learning algorithm based on the integrated data, combines long short-term memory networks for time series prediction to identify potential problem trends, and graph neural networks to analyze the relationship network between facilities, generating a problem list and a set of opportunity points for optimizing the daily management process;

[0008] Based on the problem list, use the Bayesian optimization algorithm to select the optimal plan from a preset emergency response plan library, and apply natural language generation technology and sentiment analysis models to automatically generate humanized solutions, obtaining information transmission content;

[0009] Use a genetic algorithm to search for different combinations of management strategies for the set of opportunity points, quantify the influence weights of key performance indicators based on the Markov decision process, and generate the optimal management strategy and specific implementation plan;

[0010] According to the optimal management strategy and specific implementation plan, use a personalized recommendation algorithm to customize and design a push mechanism, and at the same time establish a feedback mechanism based on blockchain technology, and use an online learning algorithm to adjust the management plan in real time for the dynamic data collected during the execution process, obtaining the best property management model.

[0011] Optionally, an artificial intelligence decision-making system constructed using a deep reinforcement learning algorithm based on the integrated data, combines long short-term memory networks for time series prediction to identify potential problem trends, and graph neural networks to analyze the relationship network between facilities, generating a problem list and a set of opportunity points for optimizing the daily management process, including:

[0012] Use the integrated multi-source heterogeneous real-time status information to dynamically update the input layer of the artificial intelligence decision-making system, and process data from different sources based on adaptive data normalization technology to obtain optimized input data with a unified scale;

[0013] According to the optimized input data with a unified scale, use long short-term memory networks combined with self-attention mechanisms to perform deep learning and long-term dependency analysis on historical data, obtaining problem trend prediction results;

[0014] Based on the problem trend prediction results, use graph neural networks to model the relationship network within the property, and introduce a random forest algorithm to evaluate the mutual influence intensity between different facilities, generating a relationship network analysis report containing potential risks and optimization opportunities;

[0015] Combine the problem trend prediction results, the relationship network analysis report, and the mutual influence intensity between facilities, and apply a multi-criteria decision analysis method to comprehensively calculate the development trend in the time dimension, the correlation effect in the space dimension, and the interaction between facilities, generating a problem list including emergency problem responses and a set of opportunity points for optimizing the daily management process;

[0016] Optionally, based on the problem trend prediction results, use a graph neural network to model the relationship network within the property, and introduce a random forest algorithm to evaluate the mutual influence intensity between different facilities, generating a relationship network analysis report containing potential risks and optimization opportunities, including:

[0017] Use the problem trend prediction results to perform graph neural network modeling on the physical connections and functional dependencies between the internal facilities of the property, obtaining a relationship network model reflecting the interactions between facilities;

[0018] Based on the relationship network model, combined with the static attributes and dynamic state information of the facilities, use the random forest algorithm to evaluate the mutual influence intensity between different facilities, generating an influence matrix between facilities;

[0019] Use the influence matrix between facilities to identify key facility nodes with mutual influence intensity and the associated paths of the key facility nodes, and quantitatively analyze the synergy effects caused by the key facility nodes based on the multi-criteria decision analysis method, obtaining a list of potential risk points and optimization opportunities;

[0020] According to the list of potential risk points and optimization opportunities, comprehensively analyze the occurrence probability and impact degree of each risk point and the expected benefits and implementation difficulties of each optimization opportunity, generating a relationship network analysis report.

[0021] Optionally, use the influence matrix between facilities to identify key facility nodes with mutual influence intensity and the associated paths of the key facility nodes, and quantitatively analyze the synergy effects caused by the key facility nodes based on the multi-criteria decision analysis method, obtaining a list of potential risk points and optimization opportunities, including:

[0022] Use the influence matrix between facilities to deeply analyze the mutual influence intensity between different facilities, identify key facility nodes with high mutual influence intensity and the associated paths of the key facility nodes, obtaining a key facility node graph;

[0023] Based on the key facility node graph, apply the multi-criteria decision analysis method, comprehensively calculate the impacts of time, cost, and resource consumption, and quantitatively evaluate the synergy effects caused by each key facility node, generating a synergy effect score;

[0024] According to the synergy effect score, analyze the roles and functions of key facility nodes in property management, identify potential risk points and optimization opportunities generated by the synergy effects between facilities, and classify and organize the potential risk points and optimization opportunities, obtaining classified potential risk points and optimization opportunities;

[0025] For the classified potential risk points and optimization opportunities, combined with historical data, evaluate the occurrence probability of each risk point and the impact degree of each risk on the overall property management. At the same time, evaluate the expected benefits and implementation difficulties brought by each optimization opportunity to generate a list of potential risk points and optimization opportunities.

[0026] Optionally, it is characterized in that, based on the problem list, use the Bayesian optimization algorithm to select the optimal plan from the preset emergency response plan library, and apply natural language generation technology and sentiment analysis model to automatically generate humanized solutions for the solution suggestions, and obtain the information transmission content, including:

[0027] According to the problem list, perform priority sorting and classification processing on each identified problem, and based on the urgency and impact scope of the problem, obtain the problem sequence to be processed first;

[0028] According to the problem sequence to be processed first, use the Bayesian optimization algorithm to search in the preset emergency response plan library, calculate the historical success rate, resource consumption and implementation difficulty of different plans, and select the best plan;

[0029] Based on the best plan, combined with natural language generation technology, generate solution suggestions for each problem;

[0030] Optimize the solution suggestions, test the effect of the information transmission content through a simulated dialogue system, and adjust the language expression to enhance the communication effect and resident satisfaction to obtain accurate and popular information transmission content.

[0031] Optionally, it is characterized in that the genetic algorithm is used to search different management strategy combinations for the opportunity point set, and the influence weight of key performance indicators is quantified based on the Markov decision process to generate the optimal management strategy and specific implementation plan, including:

[0032] Use the opportunity point set to perform combined search processing on different management strategies, and explore various possible management strategy combinations through genetic operations such as crossover and mutation to obtain a candidate management strategy set;

[0033] Based on the candidate management strategy set, combined with historical data and current state information, use the Markov decision process to model the execution path of each candidate strategy within a preset time, evaluate the influence weight of the candidate strategy on resource consumption and resident satisfaction, and generate a strategy evaluation report;

[0034] According to the strategy evaluation report, comprehensively calculate the cost-benefit ratio, implementation difficulty and expected benefits of each candidate management strategy, and select the optimal management strategy;

[0035] Refine the optimal management strategy into a specific implementation plan, clarify the responsible person, time node, and expected effect of each task, and formulate corresponding monitoring and adjustment mechanisms to effectively implement the management strategy, obtaining the optimal management strategy and the specific implementation plan.

[0036] Optionally, according to the optimal management strategy and the specific implementation plan, customize and design a push mechanism using a personalized recommendation algorithm, and at the same time establish a feedback mechanism based on blockchain technology. Use an online learning algorithm to adjust the management plan in real time for the dynamic data collected during the execution process to obtain the best property management model, including:

[0037] Utilize the optimal management strategy and the specific implementation plan, combine the historical behavior data and personal preferences of residents, and use a personalized recommendation algorithm to perform customized design processing on the push mechanism to obtain a personalized push plan;

[0038] Based on the personalized push plan, construct a sentiment analysis model for the time arrangement, channel selection, and format optimization of the push content to generate the best personalized push mechanism;

[0039] At the same time, establish a feedback mechanism based on blockchain technology to record and verify the execution status of each management decision and the feedback from residents. Through the immutable feature of the blockchain, the data is made authentic and transparent, enhancing residents' trust to obtain verified feedback data;

[0040] For the dynamic data collected during the execution process, combine the verified feedback data and use an online learning algorithm to perform real-time analysis and processing to obtain the latest management status update;

[0041] According to the latest management status update, adjust and optimize the existing management plan in real time to generate an optimized management plan. According to the optimized management plan, form a closed-loop management system integrating personalized push, blockchain feedback, and online learning adjustment.

[0042] In a second aspect, an embodiment of the present invention provides an artificial intelligence-based property management system, including:

[0043] A collection module for collecting multi-source heterogeneous real-time status information and requests using a property facility sensor network and a resident interaction platform, preliminarily processing local data through an edge computing node, and implementing data routing and integration based on a distributed hash table to generate integrated data;

[0044] A construction module for using an artificial intelligence decision system constructed by a deep reinforcement learning algorithm according to the integrated data, combining a long short-term memory network for time series prediction to identify potential problem trends, and a graph neural network to analyze the relationship network between facilities, generating a problem list and a set of opportunity points for optimizing the daily management process;

[0045] A selection module, configured to select an optimal plan from a preset emergency response plan library based on the problem list by using a Bayesian optimization algorithm, and apply natural language generation technology and a sentiment analysis model to automatically generate a humanized solution recommendation to obtain information transmission content;

[0046] A quantification module, configured to search for different management strategy combinations for the set of opportunity points by using a genetic algorithm, quantify the influence weight of key performance indicators based on a Markov decision process, and generate an optimal management strategy and a specific implementation plan;

[0047] An adjustment module, configured to customize a push mechanism according to the optimal management strategy and the specific implementation plan by using a personalized recommendation algorithm, and simultaneously establish a feedback mechanism based on blockchain technology, and use an online learning algorithm to adjust the management plan in real time for the dynamic data collected during the execution process to obtain an optimal property management mode.

[0048] In a third aspect, an embodiment of the present invention provides a computing device, including a processor and a memory, where a computer program is stored in the memory, and the processor is configured to run the computer program to execute any one of the methods for property management based on artificial intelligence in the first aspect.

[0049] In a fourth aspect, an embodiment of the present invention provides a computer storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the methods for property management based on artificial intelligence described in any one of the first aspects are implemented.

[0050] In the embodiments of the present invention, a property facility sensor network and a resident interaction platform are used to collect multi-source heterogeneous real-time status information and requests. Local data is preliminarily processed by edge computing nodes, and data routing and integration are realized based on a distributed hash table to generate integrated data. According to the integrated data, an artificial intelligence decision-making system constructed by using a deep reinforcement learning algorithm combines a long short-term memory network for time series prediction to identify potential problem trends, and a graph neural network analyzes the relationship network between facilities to generate a set of opportunity points. Based on the problem list, the Bayesian optimization algorithm is used to select the optimal emergency response plan from a preset emergency response plan library, and natural language generation technology and a sentiment analysis model are applied to automatically generate user-friendly solution suggestions to obtain information transmission content. The genetic algorithm is used to search for different management strategy combinations for the set of opportunity points, and the influence weights of key performance indicators are quantified based on the Markov decision process to generate the optimal management strategy and specific implementation plan. According to the optimal management strategy and specific implementation plan, a personalized recommendation algorithm is used to customize and design a push mechanism, and at the same time, a feedback mechanism based on blockchain technology is established. The online learning algorithm is used to adjust the management plan in real time for the dynamic data collected during the execution process to obtain the best property management mode. The technical solution provided by the present invention uses a property facility sensor network and a resident interaction platform to collect multi-source heterogeneous real-time status information, preliminarily processes local data through edge computing nodes, and realizes efficient data routing and integration based on a distributed hash table. This method ensures low latency and high reliability of data processing, improves the accuracy of problem identification and response speed. Combining a long short-term memory network (LSTM) for time series prediction and a graph neural network (GNN) to analyze the relationship network between facilities can identify potential problem trends in advance and generate a set of opportunity points to optimize the daily management process, realizing the transformation from passive response to active prevention. Based on the problem list, the Bayesian optimization algorithm is used to select the optimal emergency response plan, and combined with natural language generation technology and a sentiment analysis model to automatically generate user-friendly solution suggestions, enhancing the effectiveness and affinity of communication with residents and improving residents' satisfaction. The genetic algorithm is used to search for different management strategy combinations for the set of opportunity points, and the influence weights of key performance indicators are quantified based on the Markov decision process to generate the optimal management strategy and its specific implementation plan, ensuring the scientificity and effectiveness of the management plan. According to the optimal management strategy and specific implementation plan, a personalized recommendation algorithm is used to customize and design a push mechanism, and at the same time, a feedback mechanism based on blockchain technology is established. Combining the online learning algorithm to adjust the management plan in real time forms a closed-loop management system, ensuring the efficiency and adaptability of the property management mode.

[0051] Further, by dynamically updating the integrated multi-source heterogeneous real-time status information and processing data from different sources based on adaptive data normalization technology, optimized input data with unified scale is obtained, ensuring the consistency and accuracy of subsequent analysis.

[0052] Deep learning and long-term dependency analysis: By using long short-term memory networks (LSTMs) combined with self-attention mechanisms to perform deep learning and long-term dependency analysis on historical data, not only the accuracy of problem trend prediction is improved, but also more complex time series features can be captured, providing strong support for early warning. Based on the problem trend prediction results, a graph neural network (GNN) is used to model the relationship network within the property, and a random forest algorithm is introduced to evaluate the mutual influence strength between different facilities, generating a relationship network analysis report containing potential risks and optimization opportunities, deepening the understanding of the correlation effects between facilities and providing detailed basis for optimization management. By combining the problem trend prediction results, the relationship network analysis report, and the mutual influence strength between facilities, the multi-criteria decision analysis (MCDA) method is applied to comprehensively calculate the development trend in the time dimension, the correlation effects in the space dimension, and the interactions between facilities, generating a problem list including emergency problem responses and a set of opportunity points for optimizing daily management processes, providing more scientific and reasonable decision-making support.

[0053] These aspects or other aspects of the present invention will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0055] Figure 1 It is a flowchart of a property management method based on artificial intelligence provided by an embodiment of the present invention;

[0056] Figure 2 It is a schematic structural diagram of a property management system based on artificial intelligence provided by an embodiment of the present invention;

[0057] Figure 3 It is a schematic structural diagram of a computing device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] In order to enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention.

[0059] In some of the processes described in the specification, claims, and above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations can be performed not in the order in which they appear herein or in parallel. The operation numbers such as 101, 102, etc. are only used to distinguish different operations, and the numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations can be executed sequentially or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit that "first" and "second" are of different types.

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of protection of the present invention.

[0061] Figure 1 The flowchart of a property management method based on artificial intelligence provided for the embodiments of the present invention is as Figure 1 shown, and the method includes:

[0062] Based on this, the present invention provides a property management method based on artificial intelligence, as Figure 1 , including:

[0063] Step 101: Use the property facility sensor network and the resident interaction platform to collect multi-source heterogeneous real-time status information and requests, preliminarily process local data through edge computing nodes, and implement data routing and integration based on a distributed hash table to generate integrated data;

[0064] In this step, the property facility sensor network refers to various sensors (such as temperature, humidity, smoke alarms, etc.) installed in the property management area, which are used to monitor the environment and facility status in real time.

[0065] The resident interaction platform refers to channels including mobile applications, websites, etc. through which residents can submit requests or feedback problems; the edge computing node refers to a computing device near the data generation location that can preliminarily process local data, reduce transmission delay, and relieve the burden on the central server; the distributed hash table (DHT) refers to a distributed storage structure used to efficiently manage and search for data distributed on different nodes; the integrated data refers to a data set in a unified format after preliminary processing and routing of data from different sources;

[0066] Collect multi-source heterogeneous real-time status information and requests using the property facility sensor network and the resident interaction platform. Initially process this local data through edge computing nodes, and implement efficient data routing and integration based on a distributed hash table, ultimately generating integrated data that can be used for subsequent analysis.

[0067] Step 102: Based on the integrated data, use the artificial intelligence decision-making system constructed by the deep reinforcement learning algorithm, combined with the long short-term memory network for time series prediction to identify potential problem trends, and analyze the relationship network between facilities using a graph neural network, generating a problem list and a set of opportunity points for optimizing the daily management process;

[0068] In this step, the deep reinforcement learning algorithm: a machine learning method that combines the advantages of deep learning and reinforcement learning, used to construct an intelligent decision-making system; the long short-term memory network (LSTM) refers to a special recurrent neural network (RNN) that is good at capturing long-term dependencies in time series; the graph neural network (GNN) refers to a network used to model and analyze the relationships between nodes in a complex network structure; the problem list refers to listing the problems that require immediate response and their priorities; the set of opportunity points for optimizing the daily management process refers to discovering opportunities to improve the existing management process to increase efficiency or reduce costs;

[0069] Based on the integrated data generated in Step 101, use the artificial intelligence decision-making system constructed by the deep reinforcement learning algorithm, combined with LSTM for time series prediction to identify potential problem trends; at the same time, use GNN to analyze the relationship network between facilities, generating a problem list and a set of opportunity points for optimizing the daily management process.

[0070] Step 103: Based on the problem list, use the Bayesian optimization algorithm to select the optimal plan from the preset emergency response plan library, and apply natural language generation technology and a sentiment analysis model to automatically generate a humanized solution suggestion, obtaining the information transmission content;

[0071] In this step, the Bayesian optimization algorithm refers to a global optimization method for finding the maximum or minimum value of a function, suitable for selecting the optimal solution from multiple options; the preset emergency response plan library refers to a series of pre-prepared plans for dealing with emergencies; natural language generation technology (NLG) refers to the technology of automatically generating human language text; the sentiment analysis model: a tool for evaluating the sentiment tendency in text to ensure more humanized information transmission;

[0072] Based on the problem list generated in Step 102, use the Bayesian optimization algorithm to select the optimal plan from the preset emergency response plan library. Then, combined with natural language generation technology and a sentiment analysis model, automatically generate a humanized solution suggestion, obtaining information transmission content that is easy to understand.

[0073] Step 104: Use a genetic algorithm to search for different combinations of management strategies for the set of opportunity points, quantify the influence weights of key performance indicators based on the Markov decision process, and generate an optimal management strategy and a specific implementation plan;

[0074] In this step, the genetic algorithm refers to a search and optimization algorithm that simulates the natural evolution process; the Markov decision process (MDP) refers to a mathematical framework for modeling decision-making processes, especially suitable for environments with uncertainty; the key performance indicator (KPI) refers to an important criterion for measuring management effectiveness;

[0075] Use a genetic algorithm to search for different combinations of management strategies for the set of opportunity points generated in Step 102, and quantify the influence weights of KPIs based on the MDP to generate an optimal management strategy and a specific implementation plan. This process aims to find the best solution that can meet current needs and long-term planning.

[0076] Step 105: According to the optimal management strategy and the specific implementation plan, use a personalized recommendation algorithm to customize and design a push mechanism, and at the same time establish a feedback mechanism based on blockchain technology. Use an online learning algorithm to adjust the management plan in real time for the dynamic data collected during the execution process to obtain the best property management model;

[0077] In this step, the personalized recommendation algorithm refers to a method of customizing and recommending content based on user behavior patterns and preferences; blockchain technology refers to a decentralized ledger technology that ensures the authenticity and immutability of data; the online learning algorithm refers to a learning method that can continuously update model parameters when new data arrives;

[0078] According to the optimal management strategy and its specific implementation plan generated in Step 104, use a personalized recommendation algorithm to customize and design a push mechanism to ensure the relevance and practicality of information. At the same time, establish a feedback mechanism based on blockchain technology to record and verify the execution of each management decision. Use an online learning algorithm to adjust the management plan in real time for the dynamic data collected during the execution process to form an efficient and adaptable closed-loop management system.

[0079] Suppose that in a large residential community, the property management team introduces an artificial intelligence-based property management method aimed at improving service quality, optimizing resource allocation, and enhancing resident satisfaction. The following is the specific implementation process of this method:

[0080] First, the property management company has installed various types of sensors in the community, including temperature, humidity, smoke alarms, etc., forming a comprehensively covered property facility sensor network. At the same time, a resident interaction platform has been developed, allowing residents to submit requests and feedback problems through mobile applications or websites. All these multi-source heterogeneous real-time status information and requests will be transmitted to the edge computing nodes distributed throughout the community for preliminary processing. These edge computing nodes can quickly analyze local data, reduce unnecessary data uploads, and achieve efficient data routing and integration through a distributed hash table (DHT), ultimately generating integrated data in a unified format.

[0081] Next, the artificial intelligence decision-making system constructed using deep reinforcement learning algorithms starts to work. This system combines long short-term memory networks (LSTM) to perform time series prediction on the integrated data to identify potential problem trends, such as the risk of future power load overload or water pipe leakage in a certain area. At the same time, graph neural networks (GNN) are used to analyze the relationship network among various facilities in the community and evaluate the intensity of the mutual influence between different facilities. In this way, the system not only generates a list of problems including emergency problem responses but also discovers multiple sets of opportunity points for optimizing daily management processes, such as improving garbage collection routes or adjusting the lighting time in public areas.

[0082] For the identified problems, the system selects the optimal pre-plan from the preset emergency response pre-plan library according to the Bayesian optimization algorithm. To ensure the effective transmission of information, the system uses natural language generation technology and sentiment analysis models to automatically generate solution suggestions. For example, when it is detected that there may be a fault in the elevator of a certain building, the system will automatically generate a detailed repair notice, accompanied by a warm reminder, informing residents of the possible inconvenience and the estimated repair time. This user-friendly information transmission method improves residents' understanding and acceptance and reduces unnecessary panic.

[0083] For the sets of opportunity points for optimizing daily management processes, the system uses genetic algorithms to search for different combinations of management strategies and quantifies the influence weights of key performance indicators based on the Markov decision process. Through the comprehensive evaluation of multiple possible solutions, the system generates the optimal management strategy and its specific implementation plan. For example, in terms of garbage collection, the system recommends increasing the collection frequency during weekend periods to cope with the increased garbage volume on weekends; in terms of lighting control, it proposes a strategy to adjust the street lamp switch time according to seasonal changes. These optimization measures not only improve resource utilization but also enhance residents' living comfort.

[0084] Finally, based on the generated optimal management strategy and specific implementation plan, the system customizes and designs a push mechanism using a personalized recommendation algorithm. For example, for resident groups that require special attention (such as the elderly or those with limited mobility), the system will prioritize the push of relevant information and service reminders. In addition, a feedback mechanism based on blockchain technology is established to ensure that the implementation status of each management decision can be accurately recorded and verified, enhancing the transparency of the system and the trust of residents. On this basis, the system continuously collects the dynamic data generated during the execution process and uses online learning algorithms for real-time adjustment, continuously optimizing the management plan to form an efficient and highly adaptable closed-loop management system.

[0085] Through the above steps, the property management company has successfully achieved the intelligent upgrade of property management. The entire process not only improves the accuracy of problem recognition and response speed, enhances the personalization of services and resident satisfaction, but also provides scientific and reasonable decision-making support for property management. Ultimately, this innovative property management model has significantly improved the overall management level and service quality of the community, and has received the recognition and support of the majority of residents.

[0086] Based on this, the present invention provides a specific embodiment. In step 102, a dynamic data fusion framework is constructed. The dynamic data fusion framework adjusts the association rules between the structured data and the unstructured data according to real-time analysis requirements, and specifically includes the following steps:

[0087] Step 201: Utilize the integrated multi-source heterogeneous real-time status information to dynamically update the input layer of the artificial intelligence decision-making system, and process the data from different sources based on the adaptive data normalization technology to obtain optimized input data with a unified scale.

[0088] In this step, the multi-source heterogeneous real-time status information refers to real-time data from different sources (such as sensor networks, resident feedback platforms) with different formats.

[0089] The artificial intelligence decision-making system refers to an intelligent system constructed based on machine learning or deep learning algorithms for analysis and decision-making; the dynamic update of the input layer means adjusting the input of the model in real time according to the latest acquired data to ensure that it reflects the current situation; the adaptive data normalization technology refers to a data standardization method that can automatically adapt to different types of data and convert them into a unified scale for subsequent processing; the optimized input data refers to the data that is suitable for direct input into the artificial intelligence decision-making system after preprocessing.

[0090] Dynamically update the input layer of the artificial intelligence decision-making system using the integrated multi-source heterogeneous real-time status information. Meanwhile, adopt the adaptive data normalization technology to process data from different sources to ensure that all data are on the same scale, thereby obtaining optimized input data with a unified scale. This step ensures the consistency and accuracy of subsequent analyses.

[0091] Step 202: According to the optimized input data with a unified scale, use the long short-term memory network combined with the self-attention mechanism to perform deep learning and long-term dependence analysis on historical data, and obtain the problem trend prediction result.

[0092] In this step, the long short-term memory network (LSTM) refers to a type of recurrent neural network (RNN) that is particularly good at processing time series data and capturing long-term dependence relationships; the self-attention mechanism refers to a technology used in deep learning models to enhance the attention to important parts in the sequence, which helps to better understand the context; historical data refers to past records and observations, which are used as the basis for training and prediction; the problem trend prediction result refers to the prediction of possible future problems and their development trends.

[0093] According to the optimized input data with a unified scale generated in Step 201, combine the long short-term memory network (LSTM) and the self-attention mechanism to perform deep learning and long-term dependence analysis on historical data. In this way, the system can not only identify the problem trends that may occur in the future for a period of time, but also more accurately capture complex time series features, and finally obtain accurate problem trend prediction results.

[0094] Step 203: Based on the problem trend prediction result, use the graph neural network to model the relationship network inside the property, and introduce the random forest algorithm to evaluate the mutual influence intensity between different facilities, and generate a relationship network analysis report containing potential risks and optimization opportunities.

[0095] In this step, the graph neural network (GNN) refers to a network used to model and analyze the relationships between nodes in complex network structures, which is suitable for the interactions between facilities; the random forest algorithm refers to an ensemble learning method that improves the stability and accuracy of prediction by constructing multiple decision trees; the relationship network analysis report refers to a document that details the degree of mutual influence between the internal facilities of the property and potential risk points.

[0096] Based on the problem trend prediction result obtained in Step 202, use the graph neural network (GNN) to model the relationship network between the internal facilities of the property and evaluate how these facilities influence each other. Introduce the random forest algorithm to further quantify the mutual influence intensity between different facilities and generate a relationship network analysis report containing potential risks and optimization opportunities. This step helps the property management team deeply understand the correlation effects between facilities and prepare countermeasures in advance.

[0097] Step 204: Combine the problem trend prediction result, the relationship network analysis report, and the mutual influence intensity between facilities, and apply a multi-criteria decision analysis method to comprehensively calculate the development trend in the time dimension, the correlation effect in the space dimension, and the interaction between facilities, so as to generate a problem list including emergency problem responses and a set of opportunity points for optimizing the daily management process;

[0098] In this step, the multi-criteria decision analysis (MCDA) method refers to a decision-making tool for evaluating decisions under the influence of multiple factors, considering multiple criteria or objectives; the development trend in the time dimension refers to the trend that changes over time, such as the equipment aging rate; the correlation effect in the space dimension refers to the mutual influence between different locations or regions; the interaction between facilities refers to the direct or indirect influence method between facilities; the problem list and the set of opportunity points refer to listing the problems that need to be immediately responded to and the opportunities to improve the management process;

[0099] Combine the problem trend prediction result of Step 202, the relationship network analysis report of Step 203, and the mutual influence intensity between facilities, and apply the multi-criteria decision analysis (MCDA) method to comprehensively calculate the development trend in the time dimension, the correlation effect in the space dimension, and the interaction between facilities. Finally, generate a problem list including emergency problem responses and a set of opportunity points for optimizing the daily management process. This step provides comprehensive and scientific decision-making support to help the property management team formulate optimal strategies.

[0100] Based on this, the present invention provides a specific embodiment. In Step 203, based on the problem trend prediction result, use a graph neural network to model the relationship network inside the property, and introduce a random forest algorithm to evaluate the mutual influence intensity between different facilities, so as to generate a relationship network analysis report including potential risks and optimization opportunities, which specifically includes the following steps:

[0101] Step 301: Dynamically update the input layer of the artificial intelligence decision-making system by using the integrated multi-source heterogeneous real-time status information, and process the data from different sources based on the adaptive data normalization technology to obtain optimized input data with a unified scale;

[0102] In this step, the integrated multi-source heterogeneous real-time status information refers to diverse and differently formatted real-time data obtained from different sources (such as sensor networks, resident feedback platforms), which is used for further analysis after preliminary integration; the artificial intelligence decision-making system refers to an intelligent system built based on machine learning or deep learning algorithms for automated analysis and decision support; the dynamic update of the input layer means adjusting the input of the model in real time according to the latest collected data to ensure that it can reflect the current most accurate status; the adaptive data normalization technology refers to a data standardization method that automatically adapts to different types of data and converts them into a unified scale for subsequent processing; the optimized input data with unified scale refers to the standardized data that is suitable for direct input into the artificial intelligence decision-making system after preprocessing;

[0103] Utilize the integrated multi-source heterogeneous real-time status information to dynamically update the input layer of the artificial intelligence decision-making system. At the same time, adopt the adaptive data normalization technology to process the data from different sources to ensure that all data is on the same scale, thereby obtaining the optimized input data with unified scale. This step ensures the consistency and accuracy of subsequent analysis, enabling the system to more effectively capture potential problems and opportunity points.

[0104] Step 302: According to the optimized input data with unified scale, use the long short-term memory network combined with the self-attention mechanism to perform deep learning and long-term dependency analysis on historical data to obtain the problem trend prediction result;

[0105] In this step, according to the optimized input data with unified scale generated in Step 301, combined with the long short-term memory network (LSTM) and the self-attention mechanism, perform deep learning and long-term dependency analysis on historical data. Through this method, the system can not only identify the problem trends that may occur in the future period, but also more accurately capture complex time series features, and finally obtain accurate problem trend prediction results. This step significantly improves the prediction accuracy and helps the property management team prepare countermeasures in advance.

[0106] Step 303: Based on the problem trend prediction result, use the graph neural network to model the relationship network within the property, and introduce the random forest algorithm to evaluate the mutual influence intensity between different facilities, and generate a relationship network analysis report containing potential risks and optimization opportunities;

[0107] In this step, the graph neural network (GNN) refers to a method used to model and analyze the relationships between nodes in complex network structures and is applicable to the interactions between facilities; the random forest algorithm refers to an ensemble learning method that improves the stability and accuracy of prediction by constructing multiple decision trees; the relationship network analysis report refers to a document that details the degree of mutual influence between the internal facilities of the property and potential risk points;

[0108] Based on the problem trend prediction results obtained in step 302, use a graph neural network (GNN) to model the relationship network between the internal facilities of the property and evaluate how these facilities influence each other. Introduce the random forest algorithm to further quantify the mutual influence intensity between different facilities and generate a relationship network analysis report containing potential risks and optimization opportunities. This step helps the property management team deeply understand the correlation effects between facilities and prepare countermeasures in advance to reduce potential risks and seize optimization opportunities.

[0109] Step 304: Combine the problem trend prediction results, the relationship network analysis report, and the mutual influence intensity between facilities, and apply the multi-criteria decision analysis method to comprehensively calculate the development trend in the time dimension, the correlation effect in the space dimension, and the interaction between facilities, and generate a problem list including emergency problem responses and a set of opportunity points for optimizing the daily management process;

[0110] In this step, combine the problem trend prediction results of step 302, the relationship network analysis report of step 303, and the mutual influence intensity between facilities, and apply the multi-criteria decision analysis (MCDA) method to comprehensively calculate the development trend in the time dimension, the correlation effect in the space dimension, and the interaction between facilities. Finally, generate a problem list including emergency problem responses and a set of opportunity points for optimizing the daily management process. This step provides comprehensive and scientific decision support to help the property management team formulate optimal strategies, ensure the effective allocation of resources, and continuously improve the service quality.

[0111] Based on this, the present invention provides a specific embodiment. In step 303, use the facility influence matrix to identify the key facility nodes with mutual influence intensity and the association paths of the key facility nodes, and conduct a quantitative analysis of the synergy effects caused by the key facility nodes based on the multi-criteria decision analysis method to obtain a list of potential risk points and optimization opportunities, which specifically includes the following steps:

[0112] Step 401: Use the facility influence matrix to deeply analyze the mutual influence intensity between different facilities, identify the key facility nodes with high mutual influence intensity and the association paths of the key facility nodes, and obtain a key facility node graph;

[0113] In this step, the facility influence matrix refers to a matrix that quantifies the mutual influence intensity between different facilities and reflects the direct or indirect relationship between facilities; the in-depth analysis refers to a detailed and comprehensive study of the mutual influence intensity between facilities to reveal hidden relationships and patterns; the key facility nodes refer to the facility points that play an important role in the facility network, and their failures or optimizations may have a significant impact on the overall system;; the key facility node graph refers to a visual chart that shows the key facility nodes and their association paths;

[0114] Using the inter-facility influence matrix, deeply analyze the mutual influence intensity between different facilities. Through the analysis, identify the key facility nodes with high mutual influence intensity and their associated paths, and generate a key facility node graph. This step helps to clarify which facilities occupy important positions in the network and how these facilities interact with each other, providing a basis for subsequent evaluations.

[0115] Step 402: Based on the key facility node graph, apply the multi-criteria decision analysis method to comprehensively calculate the impacts of time, cost, and resource consumption, and quantitatively evaluate the synergy effects caused by each key facility node to generate a synergy effect score.

[0116] In this step, the impacts of time, cost, and resource consumption refer to evaluating the impacts of the synergy effects caused by each key facility node on time and resources; the synergy effect score refers to scoring the synergy effects of each key facility node according to the multi-criteria decision analysis results, reflecting its importance and potential impacts.

[0117] Based on the key facility node graph generated in Step 401, apply the multi-criteria decision analysis (MCDA) method to comprehensively calculate the impacts of time, cost, and resource consumption, and quantitatively evaluate the synergy effects caused by each key facility node. In this way, the system can generate a synergy effect score for each key facility node, helping the property management team better understand the importance and potential impacts of these nodes.

[0118] Moreover, in the existing property management methods, there is a lack of systematic and quantitative evaluation tools for key facility nodes and their synergy effects. Traditional evaluation methods often rely on manual experience judgment and are difficult to comprehensively consider the impacts of multiple factors such as time, cost, and resource consumption. By introducing the multi-criteria decision analysis (MCDA) method and combining specific calculation formulas, it is possible to quantitatively evaluate the synergy effects caused by each key facility node and generate a synergy effect score. This method not only improves the scientificity and accuracy of the evaluation but also provides solid data support for subsequent risk management and optimization. Among them, the specific calculation method for introducing the multi-criteria decision analysis (MCDA) method is as follows:

[0119] S i -αT i +βC i +γR i

[0120] Where S i represents the synergy effect score of the i-th key facility node; T i represents the impact score in the time dimension; C i represents the impact score in the cost dimension; R iIndicates the impact score on the resource consumption dimension; α, β, and γ are the weight coefficients of time, cost, and resource consumption respectively. These weights can be determined through expert scoring or historical data analysis to ensure the rationality and fairness of the comprehensive evaluation results;

[0121] Among them, the impact score T on the time dimension i is calculated as follows:

[0122]

[0123] T represents the number of observed time periods; w t represents the weight of each time period, which is set according to the importance and relevance of the time period, and comes from the time series data in the inter-facility influence matrix, as well as the performance evaluation of key facility nodes in different time periods.

[0124] Among them, the impact score C on the cost dimension i is calculated as follows:

[0125]

[0126] C represents the number of cost categories, such as labor cost, material cost, etc.; w c represents the weight of each cost category, which is set according to its importance; C i,c represents the impact degree of the key facility node i on the cost category c, which is obtained based on historical data and prediction models, and comes from the historical cost records and prediction models of property management, reflecting the impact of the facility node on various costs.

[0127] Among them, the impact score R on the resource consumption dimension i is calculated as follows:

[0128]

[0129] Among them, R represents the number of resource categories, such as electricity, water, etc.; R i,r represents the consumption degree of the key facility node i on the resource category r, which is obtained based on the inter-facility influence matrix and actual consumption data, and comes from the actual resource consumption records and influence matrix between facilities, reflecting the consumption situation of the facility node on various resources.

[0130] Step 403: According to the synergy effect score, analyze the roles and functions of key facility nodes in property management, identify potential risk points and optimization opportunities generated by the synergy effect between facilities, and classify and organize the potential risk points and optimization opportunities to obtain the classified potential risk points and optimization opportunities;

[0131] In this step, potential risk points refer to factors or situations that may have a negative impact on property management; optimization opportunities refer to opportunities to improve existing management processes to increase efficiency or reduce costs; categorization refers to organizing the identified risk points and optimization opportunities by category for subsequent processing.

[0132] Based on the synergy scores obtained in step 402, further analyze the roles and functions of key facility nodes in property management, and identify potential risk points and optimization opportunities arising from the synergy between facilities. Then, categorize and organize these potential risk points and optimization opportunities to ensure that each type of problem can be addressed specifically. This step not only helps the property management team detect potential problems in advance but also provides specific entry points for optimizing the management process.

[0133] Step 404: For the categorized potential risk points and optimization opportunities, combined with historical data, evaluate the occurrence probability of each risk point and the impact degree of each risk on overall property management, and at the same time evaluate the expected benefits and implementation difficulties brought by each optimization opportunity to generate a list of potential risk points and optimization opportunities.

[0134] In this step, for the potential risk points and optimization opportunities categorized in step 403, combined with historical data, evaluate the occurrence probability of each risk point and the impact degree of each risk on overall property management. At the same time, evaluate the expected benefits and implementation difficulties brought by each optimization opportunity. Finally, generate a list of potential risk points and optimization opportunities containing detailed evaluation results. This list provides clear guidance for the property management team to help them prioritize and handle the most important issues and seize the most valuable optimization opportunities.

[0135] Based on this, the present invention provides a specific embodiment. In step 103, based on the problem list, use the Bayesian optimization algorithm to select the optimal emergency response plan from a preset emergency response plan library, and apply natural language generation technology and sentiment analysis models to automatically generate a user-friendly solution suggestion to obtain the information transmission content, which specifically includes the following steps:

[0136] Step 501: According to the problem list, perform priority ranking and classification processing on each identified problem, and obtain a sequence of problems to be prioritized based on the urgency and impact scope of the problems.

[0137] In this step, the problem list refers to the list containing all the identified problems and their relevant information generated by the previous step; the priority ranking refers to sorting the problems according to their urgency and scope of influence to determine which problems need to be addressed first; the classified handling refers to classifying the problems according to their types or areas of influence for targeted handling; the urgency refers to evaluating the speed and severity of the impact of the problem on residents' lives or the normal operation of facilities; the scope of influence refers to evaluating the breadth of the problems involved, such as the number of affected residents or the area of facilities.

[0138] According to the problem list, prioritize and classify each identified problem. By comprehensively considering the urgency and scope of influence of the problems, determine which problems require immediate response, and arrange these problems in order of priority to obtain a sequence of problems to be addressed first. This step ensures the effective allocation of resources, enabling the most urgent and important problems to be addressed in a timely manner.

[0139] Step 502: According to the sequence of problems to be addressed first, use the Bayesian optimization algorithm to search in the preset emergency response plan library, calculate the historical success rate, resource consumption, and implementation difficulty of different plans, and select the best plan.

[0140] In this step, the preset emergency response plan library refers to a series of pre-prepared plans for dealing with emergencies; the historical success rate refers to the frequency of success of the plan in past similar situations; the resource consumption refers to the amount of resources required to implement the plan, including human and material resources; the implementation difficulty refers to the time and technical complexity required to execute the plan.

[0141] According to the sequence of problems to be addressed first generated in Step 501, use the Bayesian optimization algorithm to search in the preset emergency response plan library. Calculate the historical success rate, resource consumption, and implementation difficulty of different plans, and select the best plan that suits the current situation. This step ensures that the selected plan not only has a high probability of success but can also be efficiently implemented under the existing resource conditions, minimizing negative impacts to the greatest extent.

[0142] Step 503: Based on the best plan, combined with natural language generation technology, generate solution suggestions for each problem.

[0143] In this step, based on the best plan selected in Step 502, combined with natural language generation technology, generate detailed solution suggestions for each problem. These suggestions not only include specific countermeasures but also are expressed in easy-to-understand language to ensure that residents and other relevant personnel can quickly understand and take corresponding actions. This step improves the effectiveness of information transmission and the acceptance of residents.

[0144] Step 504: Optimize the proposed solution. Test the effectiveness of the information transfer content through a simulated dialogue system, and adjust the language expression to enhance the communication effect and residents' satisfaction, so as to obtain accurate and popular information transfer content;

[0145] In this step, the simulated dialogue system refers to a tool used to test and improve the human-computer interaction effect, which can simulate the dialogue process in a real scenario; the communication effect refers to the degree of understanding and willingness to respond of the recipient after the information is transferred; the residents' satisfaction refers to the degree of recognition and satisfaction of the residents with the provided information and services; the accurate and popular information transfer content refers to the optimized information, which is both accurate and easy to understand and can be effectively conveyed to the target audience;

[0146] Optimize the solution suggestions generated in step 503 by testing the effectiveness of the information transfer content through a simulated dialogue system. Adjust the language expression according to the test results to ensure that the information is both accurate and popular, enhancing the communication effect and residents' satisfaction. The final information transfer content can not only clearly convey the solution but also enhance the residents' trust and willingness to cooperate, thus improving the efficiency and quality of problem-solving.

[0147] Based on this, the present invention provides a specific embodiment. In step 104, a genetic algorithm is used to search for different combinations of management strategies for the set of opportunity points, and the influence weights of key performance indicators are quantified based on the Markov decision process to generate an optimal management strategy and a specific implementation plan, which specifically include the following steps:

[0148] Step 601: Use the set of opportunity points to perform a combined search process on different management strategies. Through genetic operations such as crossover and mutation, explore various possible combinations of management strategies to obtain a candidate management strategy set;

[0149] In this step, the combined search process refers to exploring the combinations between different management strategies through an algorithm to find the optimal solution; genetic operations such as crossover and mutation refer to the operations in the genetic algorithm used to generate new candidate management strategy combinations; the candidate management strategy set refers to a series of possible management strategy options obtained after the combined search;

[0150] Use the set of opportunity points to perform a combined search process on different management strategies. Through operations such as crossover and mutation in the genetic algorithm, explore various possible combinations of management strategies. Finally, obtain a candidate management strategy set containing multiple candidate solutions, aiming to widely explore different combinations of management strategies and provide a rich selection basis for subsequent evaluation.

[0151] Step 602: Based on the candidate management strategy set, combined with historical data and current status information, use the Markov decision process to model the execution path of each candidate strategy within a preset time, evaluate the impact weights of the candidate strategies on resource consumption and resident satisfaction, and generate a strategy evaluation report;

[0152] In this step, the Markov decision process (MDP) refers to a mathematical framework for modeling decision-making processes, which is particularly suitable for environments with uncertainty; the execution path refers to the implementation steps and expected effects of each candidate management strategy over a period of time in the future; the strategy evaluation report refers to a detailed record of the evaluation results of each candidate management strategy, including its advantages and disadvantages and expected impacts;

[0153] Based on the candidate management strategy set generated in Step 601, combined with historical data and current status information, use the Markov decision process (MDP) to model the execution path of each candidate strategy within a preset time. Evaluating the impact weights of the candidate strategies on resource consumption and resident satisfaction and generating a detailed strategy evaluation report ensures that the effects and impacts of each candidate management strategy can be quantitatively evaluated, providing a scientific basis for selecting the optimal strategy.

[0154] Step 603: According to the strategy evaluation report, comprehensively calculate the cost-benefit ratio, implementation difficulty, and expected benefits of each candidate management strategy, and select the optimal management strategy;

[0155] In this step, the cost-benefit ratio refers to the ratio of the cost of implementing the management strategy to the benefits it brings; the implementation difficulty refers to the time and technical complexity required to execute the management strategy; the expected benefits refer to the benefits expected to be brought after the implementation of the management strategy, such as efficiency improvement or cost reduction;

[0156] According to the strategy evaluation report generated in Step 602, comprehensively calculate the cost-benefit ratio, implementation difficulty, and expected benefits of each candidate management strategy. Through multi-faceted trade-offs, select the optimal management strategy that best meets the property management objectives. This step ensures that the selected strategy is not only effective in theory but also can be smoothly implemented in actual operation and bring maximum benefits.

[0157] Step 604: Refine the optimal management strategy into a specific implementation plan, clarify the responsible person, time node, and expected effect of each task, and establish corresponding monitoring and adjustment mechanisms to effectively execute the management strategy, obtaining the optimal management strategy and the specific implementation plan;

[0158] In this step, the specific implementation plan refers to refining the optimal management strategy into specific action guidelines, clarifying the responsible person, time node, and expected effect of each task; the monitoring and adjustment mechanism refers to a set of mechanisms used to track the implementation of the management strategy and make adjustments when necessary; the optimal management strategy and the specific implementation plan refer to the finally determined management strategy and its detailed implementation steps;

[0159] Refine the optimal management strategy selected in step 603 into a specific implementation plan. Clarify the responsible person, time node, and expected effect of each task, and establish a corresponding monitoring and adjustment mechanism to ensure the effective implementation of the management strategy. This step not only clarifies the specific implementation steps and responsibility allocation but also establishes a feedback mechanism to timely adjust the strategy to cope with changes in the actual situation, thus ensuring the effectiveness and flexibility of the management strategy.

[0160] Based on this, the present invention provides a specific embodiment. In step 104, a genetic algorithm is used to search for different management strategy combinations for the set of opportunity points, and the influence weights of key performance indicators are quantified based on the Markov decision process to generate the optimal management strategy and the specific implementation plan, which specifically includes the following steps:

[0161] Based on this, the present invention provides a specific embodiment. In step 105, according to the optimal management strategy and the specific implementation plan, a personalized recommendation algorithm is used to customize the design of the push mechanism, and at the same time, a feedback mechanism based on blockchain technology is established. The online learning algorithm is used to adjust the management plan in real time for the dynamic data collected during the execution process to obtain the best property management model, which specifically includes the following steps:

[0162] Step 701: Using the optimal management strategy and the specific implementation plan, combined with the historical behavior data and personal preferences of the residents, a personalized recommendation algorithm is used to customize the design of the push mechanism to obtain a personalized push plan;

[0163] In this step, the personalized recommendation algorithm refers to a technology that provides customized recommendations based on user characteristics; the personalized push plan refers to an information transmission plan customized for different resident groups;

[0164] Using the optimal management strategy and the specific implementation plan, combined with the historical behavior data and personal preferences of the residents, a personalized recommendation algorithm is used to customize the design of the push mechanism. By analyzing the behavior patterns and preferences of the residents, the system can generate personalized push plans for different managers and residents to ensure the relevance and practicality of the information, and improve the response efficiency and satisfaction of the recipients.

[0165] Step 702: Based on the personalized push plan, construct a sentiment analysis model for the time arrangement, channel selection, and format optimization of the push content to generate the best personalized push mechanism;

[0166] In this step, the timing is used to determine the best time point for pushing information to maximize residents' attention and participation; the channel selection is used to select the most suitable pushing channels (such as text messages, emails, app notifications, etc.) to ensure that the information can be effectively conveyed to the target audience; the format optimization sentiment analysis model is used to optimize the expression of information using sentiment analysis technology to make it more user-friendly and acceptable;

[0167] Based on the personalized pushing scheme generated in step 701, construct the timing of the pushing content, the channel selection, and the format optimization sentiment analysis model. Through these optimization measures, generate the best personalized pushing mechanism to ensure the timeliness, accuracy, and affinity of information transmission, thereby enhancing residents' participation and satisfaction.

[0168] Step 703: At the same time, establish a feedback mechanism based on blockchain technology to record and verify the implementation of each management decision and residents' feedback. Through the immutable feature of blockchain, the data has authenticity and transparency, enhancing residents' trust to obtain verified feedback data;

[0169] In this step, blockchain technology refers to a decentralized ledger technology that guarantees the authenticity and immutability of data; the feedback mechanism refers to a system for collecting and processing residents' feedback to ensure that the implementation of each management decision can be recorded and verified; the verified feedback data refers to the data recorded and verified through blockchain technology, which has authenticity and transparency;

[0170] At the same time, establish a feedback mechanism based on blockchain technology to record and verify the implementation of each management decision and residents' feedback. Utilize the immutable feature of blockchain to make the data have authenticity and transparency, enhancing residents' trust. This step not only improves the credibility of information but also promotes communication and cooperation between residents and the property management team to obtain verified feedback data.

[0171] Step 704: For the dynamic data collected during the execution process, combined with the verified feedback data, use online learning algorithms to perform real-time analysis and processing to obtain the latest management status update;

[0172] In this step, dynamic data refers to the new data continuously generated during the execution process, including facility status updates and new requests on the residents' interaction platform; online learning algorithms refer to a learning method that can continuously update model parameters when new data arrives; the latest management status update refers to the management status information adjusted based on the latest data;

[0173] For the dynamic data collected during the execution process, combined with the verified feedback data obtained in step 703, real-time analysis and processing are carried out using an online learning algorithm. This step ensures that the management system can quickly respond to changing requirements and environments, obtain the latest management status updates, and provide a scientific basis for subsequent adjustments.

[0174] Step 705: According to the latest management status update, adjust and optimize the existing management plan in real time to generate an optimized management plan. According to the optimized management plan, form a closed-loop management system integrating personalized push, blockchain feedback, and online learning adjustment;

[0175] In this step, according to the latest management status update obtained in step 704, the existing management plan is adjusted and optimized in real time to generate an optimized management plan. Finally, a closed-loop management system integrating personalized push, blockchain feedback, and online learning adjustment is formed. This system not only improves the intelligent level and service quality of property management, but also enhances the sense of participation and trust of residents, ensuring the efficiency and adaptability of the management plan.

[0176] Figure 2 The following is a schematic structural diagram of a property management system based on artificial intelligence provided by an embodiment of the present invention, as Figure 2 shown, the system includes:

[0177] A collection module 21, configured to collect multi-source heterogeneous real-time status information and requests by using a property facility sensor network and a resident interaction platform, preliminarily process local data through an edge computing node, and implement data routing and integration based on a distributed hash table to generate integrated data;

[0178] A construction module 22, configured to, according to the integrated data, use an artificial intelligence decision system constructed by a deep reinforcement learning algorithm, combine a long short-term memory network for time series prediction to identify potential problem trends, and analyze the relationship network between facilities by using a graph neural network to generate a problem list and a set of opportunity points for optimizing the daily management process;

[0179] A selection module 23, configured to, based on the problem list, use a Bayesian optimization algorithm to select an optimal plan from a preset emergency response plan library, and apply natural language generation technology and a sentiment analysis model to automatically generate a humanized solution recommendation to obtain information transmission content;

[0180] A quantization module 24, configured to use a genetic algorithm to search for different management strategy combinations for the set of opportunity points, quantify the influence weights of key performance indicators based on a Markov decision process, and generate an optimal management strategy and a specific implementation plan;

[0181] An adjustment module 25, configured to customize and design a push mechanism using a personalized recommendation algorithm according to the optimal management strategy and a specific implementation plan, and simultaneously establish a feedback mechanism based on blockchain technology to use an online learning algorithm to adjust the management plan in real time for the dynamic data collected during the execution process, so as to obtain an optimal property management model.

[0182] Figure 2 The described property management system based on artificial intelligence can execute Figure 1 For the described property management method based on artificial intelligence in the illustrated embodiment, its implementation principle and technical effects will not be elaborated further. For the property management system based on artificial intelligence in the above embodiment, the specific manners in which each module and unit perform operations have been described in detail in the embodiments related to the method, and will not be elaborated here in detail.

[0183] In a possible design, Figure 2 The property management system based on artificial intelligence in the illustrated embodiment can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0184] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0185] The processing component 32 is configured to collect multi-source heterogeneous real-time status information and requests using a property facility sensor network and a resident interaction platform, preliminarily process local data through an edge computing node, and implement data routing and integration based on a distributed hash table to generate integrated data;

[0186] According to the integrated data, an artificial intelligence decision system constructed using a deep reinforcement learning algorithm, combined with a long short-term memory network for time series prediction to identify potential problem trends, and a graph neural network to analyze the relationship network between facilities, to generate a problem list and a set of opportunity points for optimizing the daily management process;

[0187] Based on the problem list, use a Bayesian optimization algorithm to select an optimal pre-plan from a preset emergency response pre-plan library, and apply natural language generation technology and a sentiment analysis model to automatically generate a humanized solution suggestion to obtain information transmission content;

[0188] Use a genetic algorithm to search for different management strategy combinations for the set of opportunity points, quantify the influence weights of key performance indicators based on a Markov decision process, and generate an optimal management strategy and a specific implementation plan;

[0189] According to the optimal management strategy and specific implementation plan, a personalized recommendation algorithm is used to customize and design a push mechanism, and at the same time, a feedback mechanism based on blockchain technology is established. The online learning algorithm is used to adjust the management plan in real time for the dynamic data collected during the execution process, so as to obtain the best property management model.

[0190] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0191] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0192] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0193] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module can be an output device, an input device, etc.

[0194] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0195] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device can refer to a cloud server, and the above processing component, storage component, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0196] An embodiment of the present invention also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 shown embodiment of an artificial intelligence-based property management method.

[0197] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0198] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative work.

[0199] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0200] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A property management method based on artificial intelligence, characterized in that: include: Utilize the property facility sensor network and the resident interaction platform to collect multi-source heterogeneous real-time status information and requests, preliminarily process local data through edge computing nodes, and implement data routing and integration based on distributed hash tables to generate integrated data; Based on the integrated data, an artificial intelligence decision-making system constructed using a deep reinforcement learning algorithm is used to combine long short-term memory networks for time series prediction to identify potential problem trends, and a graph neural network is used to analyze the relationship network between facilities to generate a list of problems and a set of opportunity points for optimizing daily management processes; Based on the problem list, the Bayesian optimization algorithm is used to select the best plan from the preset emergency response plan library, and the natural language generation technology and sentiment analysis model are used to automatically generate humanized solution suggestions to obtain information transmission content; Using genetic algorithms to search for different management strategy combinations for the opportunity point set, quantifying the impact weights of key performance indicators based on the Markov decision process, and generating optimal management strategies and specific implementation plans; According to the optimal management strategy and specific implementation plan, a personalized recommendation algorithm is used to customize the design of the push mechanism. At the same time, a feedback mechanism based on blockchain technology is established. The dynamic data collected during the execution process is used to use an online learning algorithm to adjust the management plan in real time to obtain the optimal property management model.

2. The method according to claim 1, characterized in that Based on the integrated data, an artificial intelligence decision-making system built using a deep reinforcement learning algorithm combines long short-term memory networks for time series prediction to identify potential problem trends, and graph neural networks to analyze the relationship network between facilities, generating a list of problems and a collection of opportunity points for optimizing daily management processes, including: The integrated multi-source heterogeneous real-time status information is used to dynamically update the input layer of the artificial intelligence decision-making system, and the data from different sources are processed based on the adaptive data normalization technology to obtain optimized input data with unified scale. According to the optimized input data with unified scale, the long short-term memory network is combined with the self-attention mechanism to perform deep learning and long-term dependency analysis on the historical data to obtain the problem trend prediction result; Based on the problem trend prediction results, the graph neural network is used to model the internal relationship network of the property, and the random forest algorithm is introduced to evaluate the mutual influence intensity between different facilities to generate a relationship network analysis report containing potential risks and optimization opportunities; The problem trend prediction results, the relationship network analysis report and the mutual influence intensity between facilities are combined, and the multi-criteria decision analysis method is applied to comprehensively calculate the development trend in the time dimension, the correlation effect in the spatial dimension and the interaction between facilities, and generate a problem list including emergency problem responses and a set of opportunity points for optimizing daily management processes.

3. The method according to claim 2, characterized in that Based on the problem trend prediction results, the graph neural network is used to model the internal relationship network of the property, and the random forest algorithm is introduced to evaluate the mutual influence intensity between different facilities to generate a relationship network analysis report containing potential risks and optimization opportunities, including: The problem trend prediction results are used to perform graph neural network modeling on the physical connections and functional dependencies between facilities within the property, and a relational network model that reflects the interactions between facilities is obtained; Based on the relationship network model, combined with the static attributes and dynamic status information of the facilities, the random forest algorithm is used to evaluate the mutual influence intensity between different facilities to generate an influence matrix between facilities; Using the inter-facility impact matrix, identify the key facility nodes with mutual impact strength and the associated paths of the key facility nodes, conduct quantitative analysis on the synergy effects caused by the key facility nodes based on the multi-criteria decision analysis method, and obtain a list of potential risk points and optimization opportunities; Based on the list of potential risk points and optimization opportunities, a comprehensive analysis is conducted on the probability of occurrence of each risk point, the degree of impact of the risk point, and the expected benefits and implementation difficulty of each optimization opportunity to generate a relationship network analysis report.

4. The method according to claim 3, characterized in that The inter-facility impact matrix is ​​used to identify the key facility nodes and the associated paths of the key facility nodes with the intensity of mutual impact. The synergy effects caused by the key facility nodes are quantitatively analyzed based on the multi-criteria decision analysis method to obtain a list of potential risk points and optimization opportunities, including: Using the inter-facility impact matrix, we conduct an in-depth analysis of the mutual impact intensity between different facilities, identify key facility nodes with high mutual impact intensity and the associated paths of key facility nodes, and obtain a key facility node diagram; Based on the key facility node diagram, a multi-criteria decision analysis method is applied to comprehensively calculate the impact of time, cost and resource consumption, and the synergy effect caused by each key facility node is quantitatively evaluated to generate a synergy effect score; According to the synergy effect score, the roles and functions of key facility nodes in property management are analyzed, potential risk points and optimization opportunities generated by the synergy effect between facilities are identified, and the potential risk points and optimization opportunities are classified and sorted to obtain classified potential risk points and optimization opportunities; For the classified potential risk points and optimization opportunities, combined with historical data, the probability of occurrence of each risk point and the impact of each risk on the overall property management are evaluated. At the same time, the expected benefits and implementation difficulty of each optimization opportunity are evaluated to generate a list of potential risk points and optimization opportunities.

5. The method according to claim 1, characterized in that Based on the problem list, the Bayesian optimization algorithm is used to select the best plan from the preset emergency response plan library, and the natural language generation technology and sentiment analysis model are used to automatically generate humanized solution suggestions to obtain information transmission content, including: According to the problem list, each identified problem is prioritized and classified, and a priority problem sequence is obtained based on the urgency and impact scope of the problem; According to the priority problem sequence, a Bayesian optimization algorithm is used to search in a preset emergency response plan library, calculate the historical success rate, resource consumption and implementation difficulty of different plans, and select the best plan; Based on the best plan and combined with natural language generation technology, generate solution suggestions for each problem; The solution proposals are optimized, the effectiveness of information delivery content is tested through a simulated dialogue system, and language expression is adjusted to enhance communication effectiveness and resident satisfaction, so as to obtain accurate and popular information delivery content.

6. The method according to claim 1, characterized in that A genetic algorithm is used to search for different management strategy combinations for the opportunity point set, and the influence weights of key performance indicators are quantified based on the Markov decision process to generate the optimal management strategy and specific implementation plan, including: Using the opportunity point set, performing combined search processing on different management strategies, exploring multiple possible management strategy combinations through genetic operations such as crossover and mutation, and obtaining a candidate management strategy set; Based on the candidate management strategy set, combined with historical data and current status information, the Markov decision process is used to model the execution path of each candidate strategy within a preset time, evaluate the impact weight of the candidate strategy on resource consumption and resident satisfaction, and generate a strategy evaluation report; According to the strategy evaluation report, comprehensively calculate the cost-benefit ratio, implementation difficulty and expected benefits of each candidate management strategy, and select the optimal management strategy; The optimal management strategy is broken down into a specific implementation plan, clarifying the person in charge of each task, the time node and the expected effect, and formulating a corresponding monitoring and adjustment mechanism to ensure the effective implementation of the management strategy and obtain the optimal management strategy and specific implementation plan.

7. The method according to claim 1, characterized in that According to the optimal management strategy and specific implementation plan, a personalized recommendation algorithm is used to customize the design push mechanism. At the same time, a feedback mechanism based on blockchain technology is established. The dynamic data collected during the execution process is adjusted in real time using an online learning algorithm to obtain the optimal property management model, including: Using the optimal management strategy and specific implementation plan, combined with residents' historical behavior data and personal preferences, a personalized recommendation algorithm is used to customize the push mechanism to obtain a personalized push solution; Based on the personalized push solution, a time arrangement, channel selection and format optimization sentiment analysis model for push content is constructed to generate the best personalized push mechanism; At the same time, a feedback mechanism based on blockchain technology is established to record and verify the implementation of each management decision and residents’ feedback. Through the tamper-proof nature of blockchain, the data is made authentic and transparent, enhancing residents’ trust and obtaining verified feedback data. The dynamic data collected during the execution process is combined with the verified feedback data and analyzed in real time using an online learning algorithm to obtain the latest management status update; According to the latest management status update, the existing management plan is adjusted and optimized in real time to generate an optimized management plan. According to the optimized management plan, a closed-loop management system integrating personalized push, blockchain feedback and online learning adjustment is formed.

8. A property management system based on artificial intelligence, characterized in that: include: The collection module is used to collect multi-source heterogeneous real-time status information and requests using the property facility sensor network and the resident interaction platform, preliminarily process local data through edge computing nodes, and implement data routing and integration based on distributed hash tables to generate integrated data; A building module is used to construct an artificial intelligence decision-making system based on the integrated data using a deep reinforcement learning algorithm, combining a long short-term memory network to perform time series prediction to identify potential problem trends, and a graph neural network to analyze the relationship network between facilities, to generate a problem list and a set of opportunity points for optimizing daily management processes; A selection module is used to select the best plan from the preset emergency response plan library based on the problem list using a Bayesian optimization algorithm, and to automatically generate humanized solution suggestions using natural language generation technology and sentiment analysis models to obtain information delivery content; A quantification module, which is used to use a genetic algorithm to search for different management strategy combinations for the opportunity point set, quantify the impact weights of key performance indicators based on a Markov decision process, and generate an optimal management strategy and a specific implementation plan; The adjustment module is used to customize the design push mechanism using a personalized recommendation algorithm according to the optimal management strategy and specific implementation plan, and at the same time establish a feedback mechanism based on blockchain technology, and use an online learning algorithm to adjust the management plan in real time based on the dynamic data collected during the execution process to obtain the optimal property management model.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an artificial intelligence-based property management method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, an artificial intelligence-based property management method as described in any one of claims 1 to 7 is implemented.

Images