A Blockchain-Based Distributed Building Self-Organizing Decision-Making Method and System

Abstract

This invention discloses a blockchain-based distributed building self-organizing decision-making method and system. By collecting multi-dimensional operational data and distributed node status data of the building system, it generates decision-making tasks using smart contracts, achieves collaborative decision-making among nodes using consensus algorithms, and combines artificial intelligence to extract target optimization parameters. It then generates and broadcasts encrypted self-organizing decision-making instructions and performs simulation verification to optimize the building system. This invention improves the intelligence, collaboration, and security of building system decision-making, offering advantages such as fast real-time response, accurate processing, and reliable data.

Description

Technical Field

[0001] This invention relates to the fields of building intelligent control and blockchain technology, specifically to a blockchain-based distributed building self-organizing decision-making method and system, applicable to decision-making and optimization control in smart building systems. Background Technology

[0002] With the development of intelligent and green buildings, the automation and intelligence levels of building systems are constantly improving. Traditional building control systems typically rely on centralized data processing and decision-making mechanisms, which suffer from problems such as uneven distribution of computing resources, low data reliability, and untimely response, making it difficult to cope with complex operating environments with multiple scenarios and variables.

[0003] In recent years, blockchain technology, due to its decentralized, tamper-proof, and traceable characteristics, has been gradually introduced into building information management to improve data security and collaborative efficiency. Meanwhile, the development of artificial intelligence and edge computing technologies has also provided building systems with more powerful data processing and decision support capabilities. However, existing technologies have not yet formed an effective self-organizing decision-making mechanism that combines blockchain, artificial intelligence, and distributed node resources, and lack scalable and highly reliable building system optimization solutions.

[0004] Therefore, there is an urgent need to provide a technical solution that integrates multi-source data processing, intelligent task allocation, consensus collaboration mechanism and self-organizing optimization control to improve the intelligence level and system autonomy of building system operation. Summary of the Invention

[0005] The purpose of this invention is to provide a blockchain-based distributed building self-organizing decision-making method and system to solve the problems of centralized decision-making, low response efficiency, poor data reliability and insufficient optimization capabilities in existing building systems. It realizes intelligent task generation, node collaborative consensus, self-organizing optimization control and result feedback driven by multi-source data, thereby improving the intelligence, collaboration and security of building system operation.

[0006] Specifically, according to the present invention, a blockchain-based distributed building self-organizing decision-making method is provided, the method comprising:

[0007] When a decision request is initiated by the building system, multi-dimensional operational data of the building system and status data of distributed nodes are acquired. Based on the multi-dimensional operational data and the status data, a target decision task is generated through a smart contract in the blockchain network. When it is determined that the distributed nodes have reached a decision consensus based on the target decision task and a preset consensus algorithm, target optimization parameters are extracted based on the multi-dimensional operational data. A self-organizing decision instruction is generated based on the target optimization parameters. The self-organizing decision instruction is simulated and verified based on the multi-dimensional operational data to predict the operating state of the building system after executing the self-organizing decision instruction. If the simulation result meets the preset optimization target, the self-organizing decision instruction is broadcast through the blockchain network to execute the building system optimization; otherwise, the self-organizing decision instruction is adjusted and the simulation verification is repeated.

[0008] Furthermore, the step of generating the target decision task through a smart contract in the blockchain network based on the multi-dimensional operational data and the state data includes:

[0009] The multidimensional operational data is standardized to generate a target operational dataset in a unified format; the target operational dataset is cleaned using a target denoising algorithm, and the cleaned data is enhanced with features to obtain optimized operational data; based on the optimized operational data and the state data, a target decision task is generated through the task allocation logic in the smart contract, wherein the target decision task includes prediction parameters required to support the simulation and verification of the self-organizing decision instruction.

[0010] Furthermore, the step of generating the target decision task based on the optimized operating data and the state data through the task allocation logic in the smart contract includes:

[0011] Based on the target clustering algorithm, the optimized operation data is classified into multiple building operation scenario features; the computing power and trust level of each distributed node are determined according to the status data; and a target decision task is generated through the task allocation logic of the smart contract based on the building operation scenario features, the computing power, and the trust level.

[0012] Furthermore, before the step of cleaning the target running dataset according to the target denoising algorithm, the method further includes:

[0013] Acquire historical multidimensional operational data and corresponding historical decision result data; generate a data cleaning training sample set based on the historical multidimensional operational data and the historical decision result data; train a target denoising model based on the target machine learning algorithm and the data cleaning training sample set; calculate the loss value between the predicted value and the true value of the target denoising model based on the target loss function, and adjust the parameters of the target denoising model through an optimization algorithm so that the loss value is less than a preset threshold.

[0014] Furthermore, the step of extracting target optimization parameters based on the multidimensional operational data when determining that distributed nodes have reached a decision consensus based on the target decision task and the preset consensus algorithm includes: verifying the consistency of the task execution results of distributed nodes through the preset consensus algorithm based on the target decision task; when the verification is successful, performing semantic analysis on the multidimensional operational data based on the target parameter extraction model in the artificial intelligence module; and extracting target optimization parameters related to the target decision task based on the semantic analysis results.

[0015] Furthermore, the step of generating self-organizing decision instructions based on the target optimization parameters and broadcasting the self-organizing decision instructions through the blockchain network to execute building system optimization includes: generating identification information of the self-organizing decision instructions based on the target optimization parameters; encrypting the self-organizing decision instructions through a smart contract to generate encrypted decision instructions; broadcasting the encrypted decision instructions through the blockchain network and executing building system optimization based on the decryption verification results of distributed nodes.

[0016] Furthermore, after the steps of generating self-organizing decision instructions based on the target optimization parameters and broadcasting the self-organizing decision instructions through the blockchain network to execute building system optimization, the method further includes: obtaining execution result data of the self-organizing decision instructions; generating a building system optimization effect evaluation report based on the execution result data, the multi-dimensional operation data, and the simulation verification results; storing the optimization effect evaluation report in the blockchain network and granting access permissions to distributed nodes according to preset sharing rules.

[0017] Furthermore, after the step of generating self-organizing decision instructions based on the target optimization parameters and broadcasting the self-organizing decision instructions through the blockchain network to execute building system optimization, the method further includes: when an abnormality in the operation of the building system is detected, acquiring multi-dimensional operational data under the abnormal scenario; generating anomaly handling suggestions based on the multi-dimensional operational data under the abnormal scenario and historical decision data in the blockchain network; broadcasting the anomaly handling suggestions to distributed nodes through a smart contract, and executing the anomaly handling operation based on the consensus result of the distributed nodes.

[0018] This invention also provides a blockchain-based distributed building self-organizing decision-making system, the system comprising:

[0019] The data acquisition module is configured to acquire multi-dimensional operational data of the building system and status data of the distributed nodes when the building system initiates a decision request through the data acquisition unit of the distributed nodes;

[0020] The task generation module is configured to generate the target decision task through a smart contract in the blockchain network based on the multidimensional operational data and the status data.

[0021] The parameter extraction module is configured to extract target optimization parameters based on the multi-dimensional operational data by the artificial intelligence module when it is determined that distributed nodes have reached a decision consensus based on the target decision task and the preset consensus algorithm.

[0022] The decision execution module is configured to generate self-organizing decision instructions based on the target optimization parameters, and to simulate and verify the self-organizing decision instructions based on the multi-dimensional operational data to predict the operating state of the building system after executing the self-organizing decision instructions; if the simulation result meets the preset optimization target, the self-organizing decision instructions are broadcast through the blockchain network to execute the building system optimization; otherwise, the self-organizing decision instructions are adjusted and the simulation verification is repeated.

[0023] Compared with the prior art, the present invention has the following advantages:

[0024] 1. A self-organizing decision-making mechanism that integrates multi-dimensional operational data and distributed node status data is proposed. Combined with smart contracts to generate target decision-making tasks, it breaks through the limitations of traditional building systems that rely on centralized node scheduling, realizes the autonomous generation of data-driven distributed tasks, and effectively improves the system's decentralized processing capability and intelligent perception capability.

[0025] 2. By constructing task allocation logic for distributed nodes, taking into account the computing power and trust level of the nodes, dynamic classification and accurate assignment of tasks are achieved, which solves the problems of low resource utilization efficiency and poor node collaboration in the existing system, and improves the overall computing efficiency and trust collaboration level of the system.

[0026] 3. Based on the blockchain network and the preset consensus algorithm, the consistency of the execution results of each node is verified, and the encrypted self-organizing decision instructions are automatically broadcast after consensus is reached, which ensures the transparency and immutability of instruction execution and improves the security and reliability of decision release and execution.

[0027] 4. Supports anomaly detection and intelligent processing mechanisms. When the building system malfunctions, it can generate anomaly handling suggestions and achieve consensus response by combining historical decision data, thereby enhancing the system's self-recovery and resilience in emergency scenarios and breaking through the limitations of traditional systems' delayed post-event response.

[0028] 5. A closed-loop decision feedback mechanism and simulation verification were constructed. Simulation verification was carried out before the decision was executed, and the result data was extracted and an optimization effect evaluation report was generated after the decision was executed. The data was stored and access was opened based on blockchain, which enhanced the system's continuous optimization capability and process traceability, and promoted the evolution of building systems towards high autonomy, high transparency and high security. Attached Figure Description

[0029] Figure 1 A flowchart illustrating a blockchain-based distributed building self-organizing decision-making method is provided in this invention.

[0030] Figure 2 This invention provides a flowchart of a method for evaluating and sharing the optimization effects of a building system;

[0031] Figure 3 This invention provides a flowchart of a distributed response method for handling anomalies in building systems;

[0032] Figure 4 This invention provides a framework diagram for a blockchain-based distributed building self-organizing decision-making system.

[0033] Figure label:

[0034] A blockchain-based distributed building self-organizing decision-making system 101, comprising a data acquisition module 1011, a task generation module 1012, a parameter extraction module 1013, and a decision execution module 1014. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature; in the description of this application, unless otherwise stated, "multiple" means two or more.

[0037] To more clearly illustrate the technical solution of the present invention, the present invention will be described in detail below with reference to specific embodiments, but it should not be construed as a limitation on the scope of protection of the present invention.

[0038] In this embodiment, as Figure 1 As shown, a blockchain-based distributed building self-organizing decision-making method is provided. This method can be implemented through the following steps to ensure that the building system achieves efficient, transparent and reliable optimization decisions in a distributed environment.

[0039] Specifically, step S01: Upon detecting a decision request initiated by the building system, acquire multidimensional operational data of the building system and status data of distributed nodes. The building system can be an intelligent building management system, such as a system for managing energy consumption, environmental control, or equipment operation in an office building. When the system detects a triggering event requiring decision-making (e.g., excessively high indoor temperature or abnormal energy consumption), it collects relevant data through the data acquisition units of the distributed nodes. Multidimensional operational data includes, but is not limited to, real-time data such as temperature, humidity, energy consumption, equipment operating status, and occupancy density within the building. This data is acquired through sensors, IoT devices, or the building management system in the form of time-series or structured data. For example, a temperature sensor records the indoor temperature every minute, generating multidimensional data points containing timestamps and temperature values. The status data of the distributed nodes includes information such as the computing power, storage capacity, online status, and network connection quality of each node. For example, a distributed node may be an edge computing device deployed in the building, whose status data may indicate that its current CPU utilization is 30%, available memory space is 2GB, and network latency is 10ms. This data acquisition is achieved through standardized data acquisition protocols (such as MQTT or HTTP) to ensure the real-time nature and accuracy of the data. Those skilled in the art will understand that the frequency and range of data acquisition can be adjusted according to the specific needs of the building system; for example, high-frequency acquisition is used for dynamic environmental control, while low-frequency acquisition is used for long-term trend analysis.

[0040] Further, step S02: Based on the multi-dimensional operational data and the state data, a target decision task is generated through a smart contract in the blockchain network. The blockchain network consists of multiple distributed nodes, each running a blockchain client and collectively maintaining a decentralized ledger. The smart contract is an automated execution program pre-deployed in the blockchain network, containing logical rules for generating the decision task. Specifically, the smart contract first performs preliminary processing on the multi-dimensional operational data and state data, such as calculating the average building energy consumption or detecting node availability using a data aggregation function. Then, the smart contract transforms the data into a specific decision task according to predefined task generation logic. For example, if the multi-dimensional operational data indicates that the air conditioning energy consumption on a certain floor exceeds a threshold, and the state data shows that multiple nodes have sufficient computing power, the smart contract will generate a target decision task, such as "optimize the air conditioning operating parameters on this floor to reduce energy consumption." The task generation logic can be based on a rule engine or simple conditional judgments, such as "if energy consumption > 100kWh and node computing power > 50%, then generate an optimization task." The execution of smart contracts ensures that the task generation process is transparent and tamper-proof, and can be understood by those skilled in the art. The logic of smart contracts can be customized according to actual application scenarios (such as energy-saving optimization or equipment maintenance), but does not involve complex data cleaning or feature enhancement operations.

[0041] Further, in step S03: when the distributed nodes reach a consensus based on the target decision task and the preset consensus algorithm, the target optimization parameters are extracted based on the multi-dimensional operational data. After the target decision task is generated, the distributed nodes perform corresponding calculations or analyses based on the task content, such as calculating the optimal air conditioning temperature setpoint or equipment operating schedule. The preset consensus algorithm is used to ensure that all nodes agree on the task execution result. Common consensus algorithms include Proof-of-Work (PoW), Proof-of-Stake (PoS), or Practical Byzantine Fault Tolerance (PBFT). Taking PBFT as an example, each node votes on the task execution result, and a consensus is considered to have been reached only when more than 2 / 3 of the nodes agree on the result. For example, if the target decision task is to optimize air conditioning operating parameters, the nodes may calculate the set temperature as 24℃, 25℃, and 24℃ respectively. The consensus algorithm confirms 24℃ as the consensus result by comparing these results. After reaching a consensus, the system extracts the target optimization parameters based on the multi-dimensional operational data, such as extracting the optimization parameters "set temperature = 24℃, operating time = 8 hours" from temperature, energy consumption, and personnel density data. The extraction process can be achieved through simple statistical analysis (such as mean or weighted average) or predefined rules, such as "selecting the temperature value with the lowest energy consumption and meeting comfort requirements." Those skilled in the art will understand that the rules for parameter extraction can be adjusted according to the optimization objectives of the building system (such as energy saving or comfort), but do not involve complex semantic analysis or feature decomposition.

[0042] Further, in step S04: a self-organizing decision instruction is generated based on the target optimization parameters, and the self-organizing decision instruction is simulated and verified based on the multi-dimensional operating data to predict the operating state of the building system after executing the self-organizing decision instruction; if the simulation result meets the preset optimization target, the self-organizing decision instruction is broadcast through the blockchain network to execute the building system optimization; otherwise, the self-organizing decision instruction is adjusted and the simulation verification is repeated.

[0043] Understandably, before issuing instructions, the system will simulate and verify the decision-making instructions on a platform such as digital twin simulation. By simulating the building system's response after executing the instructions, the system can assess whether the preset optimization goals have been achieved, such as improved energy efficiency, reduced elevator waiting time, or increased user satisfaction.

[0044] If the simulation results meet the optimization objectives, the smart contract will call the blockchain network to broadcast the results across the entire network, triggering the actual building system to respond with the command. If the results do not meet the expectations, the system will adjust the optimization parameters based on the deviation analysis results and repeat the simulation verification process until the requirements are met before execution.

[0045] Furthermore, the simulation verification is based on a digital twin model of the building system. This model constructs a virtual building environment in real-time synchronization, integrating multiple physical environment and control system simulation modules, such as temperature and humidity sensors, electrical load models, equipment response characteristics, and personnel behavior patterns. The simulation platform acquires the latest multi-dimensional operational data to perform high-fidelity reconstruction of the current state of the building system, which serves as the basis for verification.

[0046] In its implementation, the system inputs the self-organizing decision-making instructions to be executed into the control module of the simulation platform, and drives the virtual building's operating logic to simulate the actual response process. For example, when the instruction involves adjusting the air conditioning system's operating strategy, the system will simulate the start-stop frequency of the air conditioning compressor, changes in the outlet temperature, and the heat transfer process interacting with the external environment, predicting the trend of temperature and humidity changes inside the building after the instruction is executed.

[0047] Meanwhile, the system sets multiple evaluation indicators to quantify the effectiveness of command execution, including but not limited to energy consumption change rate (kWh), average comfort index per person (PMV / PPD), system stability (such as current fluctuation amplitude), response latency, and task completion rate. During the simulation, these indicators will be compared and analyzed with preset optimization targets to determine whether the decision command has reached the optimization threshold.

[0048] Furthermore, the self-organizing decision-making instructions are control commands generated based on target optimization parameters, such as "adjust the air conditioner setting temperature on a certain floor to 24℃ and set the running time to 8 hours." The instruction generation process can be completed using instruction templates in smart contracts, for example, by filling optimization parameters into a predefined instruction format. The generated self-organizing decision-making instructions are sent to all relevant nodes via a blockchain network broadcast protocol (such as a P2P network), ensuring the transparency and consistency of the instructions. Nodes receiving the instructions (such as building management systems or equipment controllers) execute optimization operations according to the instruction content, such as adjusting the operating parameters of the air conditioning equipment via API calls. During execution, the blockchain network records the broadcast and execution status of the instructions, ensuring the traceability of the operations. For example, the execution result of the instruction can be recorded as a transaction on the blockchain, including information such as timestamps, execution node IDs, and execution status. Those skilled in the art will understand that the format and broadcasting method of the instructions can be adjusted according to the specific implementation of the blockchain network (such as Ethereum or Hyperledger Fabric), but this does not involve encryption processing or effect evaluation.

[0049] In summary, this method acquires multi-dimensional operational and state data, along with simulation verification results. It then utilizes smart contracts on a blockchain network to generate target decision-making tasks. After reaching consensus, optimization parameters are extracted, and self-organizing decision-making instructions are generated and broadcast, ultimately optimizing the building system. This process ensures decentralized, transparent, and trustworthy decision-making, making it suitable for scenarios such as smart building management and energy optimization. Those skilled in the art can adjust the data acquisition method, consensus algorithm, or instruction format according to specific application scenarios without introducing complex model training or semantic analysis.

[0050] In some embodiments, the step of generating a target decision task through a smart contract in a blockchain network based on the multidimensional operational data and the state data can be implemented in the following ways to ensure that the generated target decision task accurately reflects the operational needs of the building system and the processing capabilities of the distributed nodes.

[0051] Specifically, the multidimensional operational data is standardized to generate a target operational dataset in a unified format. Multidimensional operational data is typically collected by various sensors and devices within a building system, and the data formats may be heterogeneous. For example, temperature data may be expressed in degrees Celsius in a floating-point format (e.g., 25.5℃), while energy consumption data may be expressed in kilowatt-hours in an integer format (e.g., 100kWh). Standardization aims to convert this heterogeneous data into a unified format for easier subsequent processing. For instance, all numerical data can be converted to floating-point format, and a unified timestamp and data source identifier can be added to each data entry. In implementation, standardization can be accomplished using data transformation scripts, such as using a Python script to map temperature and energy consumption data to a unified JSON structure, such as {"timestamp":"2025-05-20T10:00:00","type":"temperature","value":25.5,"unit":"C"}. Furthermore, standardization also includes handling missing values, such as using interpolation to fill in missing time-series data (e.g., linear interpolation to calculate missing temperature values). Those skilled in the art will understand that the details of the standardization process can be adjusted according to the source and type of data. For example, for non-numerical data (such as the on / off status of equipment), it can be converted into Boolean values ​​or enumerated values, but without involving complex feature extraction or cluster analysis.

[0052] Further, the target running dataset is cleaned using a target denoising algorithm, and feature enhancement is performed on the cleaned data to obtain optimized running data. The target denoising algorithm aims to remove noise and outliers from the target running dataset to improve data quality. For example, the target denoising algorithm can be a statistical anomaly detection method, such as the Z-score method, to detect and remove data points that deviate from the mean by three times the standard deviation. Assuming the target running dataset contains a set of temperature data [25.0, 25.2, 30.0, 25.1], where 30.0 is an outlier, it can be removed after the Z-score method calculates its deviation. Data cleaning also includes handling duplicate data, such as removing duplicate records by checking for timestamps. The cleaned data is then subjected to feature enhancement to extract more meaningful features to support decision-making tasks. Feature enhancement can include calculating derived features, such as aggregating hourly energy consumption data into daily average energy consumption, or calculating the rate of change of temperature data (e.g., temperature change of 0.1°C per minute). In specific implementations, feature enhancement can be achieved through simple mathematical operations, such as using a sliding window to calculate the moving average of energy consumption. Optimized operational data is the result of cleaning and enhancement, such as a dataset containing average daily energy consumption and temperature change rates. Those skilled in the art will understand that the choice of denoising algorithms and feature enhancement methods can be adjusted based on data characteristics, such as using median filtering instead of the Z-score method, without involving the training of machine learning models or scene classification.

[0053] Further, based on the optimized operating data and the status data, a target decision task is generated through task allocation logic in the smart contract. This target decision task includes the prediction parameters required to support the simulation and verification of the self-organizing decision instructions. The optimized operating data provides a reliable representation of the building system's operating status, such as average daily energy consumption and temperature change rate, while the status data reflects the available resources of distributed nodes, such as CPU utilization and memory capacity. The task allocation logic in the smart contract is a set of predefined rules used to generate the target decision task based on this data. For example, the task allocation logic could be a conditional statement: if the optimized operating data shows that the average daily energy consumption of a certain floor exceeds 100 kWh, and the status data shows that the CPU utilization of at least 3 nodes is below 50%, then an energy consumption optimization task is generated, such as "reduce the air conditioning power of this floor". In specific implementations, the smart contract can be written as Solidity code and run on an Ethereum-compatible blockchain network. An example code is provided below.

[0054]

[0055] The smart contract receives optimized operational data (avgEnergy) and status data (nodeCpuUsage) and generates a task description based on the conditions. After task generation, it is stored on the blockchain to ensure transparency and immutability. Those skilled in the art will understand that the specific rules for task allocation logic can be customized according to the optimization goals of the building system (such as energy saving or equipment maintenance), but do not involve complex operations such as node computing power grading or semantic analysis. Furthermore, the target decision task includes not only scheduling objectives and execution requirements, but also key prediction parameters needed to support subsequent simulation verification. For example, in a building air conditioning energy-saving optimization task, the prediction parameters included in the target decision task may include: the expected changes in occupant density in different areas of the building, short-term forecast curves of external temperature, air conditioning system response delay models, estimated regional heat capacity, and equipment start-up and shutdown power consumption models, etc. These parameters can be directly used as inputs to the simulation model to accurately reproduce the dynamic response process of the building environment under command execution.

[0056] Furthermore, the system supports uncertainty modeling for prediction parameters. The confidence interval or range of variation for each prediction parameter will be marked in the target decision task, ensuring that the simulation system can evaluate the robustness of the command under certain disturbance conditions. For example, if the external temperature prediction has a fluctuation range of ±2℃, the simulation system can simulate the impact of this change on the building's internal temperature control system to improve the actual credibility and engineering usability of the simulation verification.

[0057] In summary, this step transforms multidimensional operational data into optimized operational data through standardization, data cleaning, and feature enhancement. Combined with state data, it generates target decision tasks through the task allocation logic of smart contracts. These target decision tasks include the prediction parameters required to support the simulation and verification of the self-organizing decision instructions. This process ensures high efficiency in data processing and reliable task generation, making it suitable for distributed building management scenarios. Those skilled in the art can adjust the standardization format, noise reduction algorithm, or task allocation logic according to specific application scenarios, but the above steps can be implemented without introducing complex model training or scenario classification.

[0058] In some embodiments, the step of generating a target decision task through task allocation logic in a smart contract based on the optimized operating data and the state data can be implemented in the following way to ensure that task allocation fully considers the building operation scenario and the performance differences of distributed nodes, thereby generating an efficient and adaptable target decision task.

[0059] Specifically, based on the target clustering algorithm, the optimized operation data is classified into multiple building operation scenario features. The optimized operation data is a standardized and feature-enhanced dataset, containing features such as daily energy consumption, temperature change rate, and personnel density. The target clustering algorithm is used to group this data into different building operation scenarios, such as "peak energy consumption scenario," "low-load comfort scenario," or "equipment-intensive operation scenario." A feasible target clustering algorithm is K-means clustering, which divides the optimized operation data into K clusters by minimizing the distance from data points to cluster centers. For example, suppose the optimized operation data contains the following feature vectors: [daily energy consumption = 120 kWh, temperature change rate = 0.1℃ / min, personnel density = 0.5 people / m²]. 2 The K-means algorithm can assign data points to "peak energy consumption scenarios" based on feature similarity. In implementation, the K value can be preset based on the data distribution (e.g., K=3). By iteratively calculating cluster centers, data points are assigned to the nearest cluster to obtain scenario features, such as "peak energy consumption scenario: energy consumption > 100kWh, temperature change rate < 0.2℃ / min". The clustering results are stored in structured data format, such as JSON: {"scene":"high_energy","features":{"avg_energy":120,"temp_rate":0.1,"density":0.5}}. Those skilled in the art will understand that the choice of clustering algorithm (such as K-means or DBSCAN) and the determination of the K value can be adjusted according to the data scale and scenario requirements, but do not involve complex semantic analysis or model training.

[0060] Furthermore, the computing power and trust level of each distributed node are determined based on the aforementioned status data. The status data includes the operating status of the distributed nodes, such as CPU utilization, available memory, network latency, etc., as well as historical performance data of the nodes, used to assess the trust level. Computing power can be determined by quantifying the node's hardware performance and current load; for example, nodes with CPU utilization below 50% and available memory greater than 1GB can be labeled as having "high computing power." In a specific implementation, a computing power scoring formula can be defined, for example: Computing power score = (1 - CPU utilization) × 0.5 + (Available memory / Total memory) × 0.5. The trust level is based on the node's historical behavior, such as task completion rate and data consistency records. Assuming a node has successfully completed 95 out of the past 100 tasks without data tampering, its trust level can be set to 0.95 (range 0 to 1). In implementation, the trust level can be calculated using historical transaction logs stored on the blockchain, such as the consensus voting success rate of the nodes. After processing the state data, a node state table is generated, for example: {"node_id":"Node1","compute_score":0.8,"trust_level":0.95}. Those skilled in the art will understand that the evaluation methods for computing power and trust level can be adjusted according to node type and application scenario, for example, by adding network bandwidth as a computing power indicator, but without involving complex machine learning models.

[0061] Understandably, based on the building's operational scenario characteristics, computing power, and trust level, a target decision task is generated through the task allocation logic of a smart contract. The building's operational scenario characteristics provide the context for the task; for example, a "peak energy consumption scenario" requires energy consumption optimization, while computing power and trust level determine which nodes are suitable to execute the task. The task allocation logic in the smart contract is a set of rules used to match scenario characteristics and node capabilities. For example, the logic rule could be: if the scenario is a "peak energy consumption scenario" and the node's computing power score is >0.7 and trust level is >0.9, then assign the task "Optimize air conditioning operating parameters." In specific implementations, the smart contract can be written in Solidity and run on a blockchain network. A code example is as follows:

[0062]

[0063]

[0064] The smart contract receives scene characteristics, a compute score, and a trust level, and generates a target decision task based on these conditions. Once generated, the task is recorded on the blockchain to ensure the allocation process is transparent and immutable. Those skilled in the art will understand that the specific rules for task allocation can be adjusted based on the scene characteristics and the priority of node capabilities, such as prioritizing allocation to nodes with higher trust levels, but this does not involve encryption or semantic analysis.

[0065] In summary, this step uses a target clustering algorithm to classify the optimized operational data into scenarios, evaluates the computing power and trust level of nodes based on state data, and generates target decision tasks through the task allocation logic of smart contracts. This process ensures that task allocation is highly matched with the building operation scenario and node capabilities, making it suitable for distributed building management scenarios. Those skilled in the art can adjust the clustering algorithm, scoring formula, or allocation logic according to specific scenarios, without introducing complex data cleaning models or effect evaluation methods.

[0066] In some embodiments, before cleaning the target running dataset according to the target denoising algorithm, the target denoising model is trained through the following steps to ensure that the data cleaning process can effectively remove noise and improve the quality of the target running dataset.

[0067] Specifically, historical multidimensional operational data and corresponding historical decision-making results data are acquired. Historical multidimensional operational data includes long-term operational records of the building system, such as time-series data like temperature, energy consumption, and humidity, stored in a building management database or blockchain ledger. For example, the data might include hourly energy consumption (e.g., 100 kWh) and temperature (e.g., 25.0°C). Historical decision-making results data are records of optimization decisions associated with this data, such as "adjust the air conditioner to 24°C". Data is obtained through database queries or blockchain APIs, in a structured record format, such as {"time":"2024-05-20T10:00:00","energy":100,"temperature":25.0,"decision":"SetAC to 24C"}. Those skilled in the art will understand that the data acquisition method can be adjusted according to the system architecture, such as using distributed storage, but does not involve scenario classification.

[0068] Further, a data cleaning training sample set is generated based on the historical multidimensional operational data and the historical decision result data. The training sample set provides labeled data for the target noise reduction model, where the historical multidimensional operational data serves as input, and the labels are derived based on the decision results. For example, if a certain energy consumption data (e.g., 1000 kWh) far exceeds the normal range (50-200 kWh) and is not used for effective decision-making, it is labeled as "noise". Label generation can employ statistical methods, such as the Z-score method, labeling data that deviates from the mean by three times the standard deviation. The sample set format is: {"input":{"energy":1000,"temperature":25.0},"label":"noise"}. The sample size can be adjusted based on historical data, for example, extracting 100,000 records. Those skilled in the art will understand that the labeling rules can be optimized based on data characteristics, but do not involve semantic analysis.

[0069] Understandably, a target denoising model is trained based on the target machine learning algorithm and the cleaned training sample set. The target machine learning algorithm can be a random forest, suitable for classification tasks handling multi-dimensional running data. The training samples are divided into a training set and a validation set (80% training, 20% validation), with features including numerical values ​​such as energy consumption and temperature. The random forest learns the relationship between features and labels by constructing decision trees, for example, determining whether high-energy-consumption data is noise. Those skilled in the art will understand that the algorithm and parameters (such as the number of trees) can be adjusted according to the data scale, but complex deep learning models should be avoided.

[0070] Further, the loss value between the predicted value and the true value of the target denoising model is calculated according to the target loss function, and the parameters of the target denoising model are adjusted by an optimization algorithm so that the loss value is less than a preset threshold. The target loss function uses cross-entropy loss to measure the difference between the predicted label and the true label. For example, if the predicted noise probability is 0.9 and the true label is "noise", the loss value is -log(0.9). The optimization algorithm uses gradient descent to adjust the model parameters (such as feature weights). Training is iterated until the loss on the validation set is less than a preset threshold (e.g., 0.1). The optimized model can accurately identify noisy data, for example, labeling 1000kWh as noise. Those skilled in the art will understand that the loss function and threshold can be adjusted according to the scenario, but this does not involve consensus verification.

[0071] In summary, this step, by acquiring historical data, generating training samples, training the target denoising model, and optimizing parameters, supports data cleaning, ensures the accuracy of noise identification, and is suitable for distributed building management scenarios. Those skilled in the art can adjust the data format, algorithm, or thresholds, but this can be achieved without complex analysis.

[0072] In some embodiments, when determining that distributed nodes have reached a decision consensus based on the target decision task and the preset consensus algorithm, the step of extracting target optimization parameters based on the multidimensional operational data can be implemented in the following ways to ensure that the parameter extraction process is efficient and accurate, and is applicable to distributed building self-organizing decision-making scenarios.

[0073] Based on the target decision task, a pre-defined consensus algorithm is used to verify the consistency of the task execution results of distributed nodes. The target decision task is a task generated by a smart contract in the blockchain network, such as "optimizing the air conditioning operating parameters on a certain floor to reduce energy consumption." Distributed nodes independently perform calculations or analyses based on the task content, such as calculating the optimal air conditioning temperature setpoint. The pre-defined consensus algorithm is used to verify the consistency of the execution results of each node; common algorithms include Practical Byzantine Fault Tolerance (PBFT). In PBFT, each node submits its task execution result (e.g., setting the temperature to 24℃) to the blockchain network. The master node in the network collects the results and votes; verification is successful only when more than 2 / 3 of the nodes' results are consistent. For example, assuming 5 nodes submit results of [24℃, 24℃, 25℃, 24℃, 24℃], the PBFT algorithm confirms 24℃ as the consensus result. In specific implementation, the consensus process is completed through the protocol stack of the blockchain client, and the verification result is recorded as a blockchain transaction, ensuring transparency and immutability. Those skilled in the art will understand that the choice of consensus algorithm (such as PBFT or PoS) and voting threshold can be adjusted according to network size, but do not involve feature decomposition or semantic scoring.

[0074] Furthermore, upon successful verification, semantic analysis is performed on the multidimensional operational data based on the target parameter extraction model in the artificial intelligence module. The target parameter extraction model is a pre-trained artificial intelligence model used to extract optimization parameters relevant to the target decision-making task from the multidimensional operational data. The multidimensional operational data includes real-time data such as temperature, energy consumption, and humidity, for example [temperature = 25.0℃, energy consumption = 120kWh, humidity = 60%]. Semantic analysis aims to identify patterns or features in the data related to the task objective (such as energy saving or comfort). For example, for the task "optimize air conditioning operating parameters," the model analyzes the data to determine which parameters (such as the temperature setpoint) have the greatest impact on energy consumption. In implementation, the target parameter extraction model can be a simple regression model (such as linear regression), predicting optimal parameters by analyzing the correlation between historical data and decision results. For example, the model might learn the relationship that "a 1℃ decrease in temperature leads to a 5kWh reduction in energy consumption." The semantic analysis process is implemented through model inference, taking multidimensional operational data as input and outputting task-related parameter suggestions. Those skilled in the art will understand that the choice of model can be adjusted according to the complexity of the data, such as using a decision tree instead of a regression model, without involving complex multidimensional feature decomposition.

[0075] Understandably, target optimization parameters related to the target decision-making task are extracted based on the semantic analysis results. The semantic analysis results are the output of the target parameter extraction model, such as a set of parameter suggestions: {"Temperature setpoint": 24℃, "Running time": 8 hours}. The process of extracting target optimization parameters involves aligning these suggestions with the objectives of the target decision-making task. For example, if the task objective is "to reduce energy consumption while maintaining comfort," parameters that meet the conditions are selected from the analysis results, such as a temperature of 24℃ (both energy-efficient and within the comfort range of 20-26℃). In specific implementations, the extraction process can be achieved through rule-based filtering, for example, setting a threshold rule: "Select parameters with energy consumption reduction > 10% and temperature within 20-26℃". The extracted parameters are stored in a structured format, such as JSON: {"parameter":"temperature","value":24,"unit":"C"}, and recorded on a blockchain to ensure traceability. Those skilled in the art will understand that the rules for parameter extraction can be adjusted according to the task objective, such as prioritizing comfort or energy efficiency, but do not involve semantic relevance scoring or broadcast encryption.

[0076] In summary, this step verifies the consistency of task execution results of distributed nodes through a pre-defined consensus algorithm, performs semantic analysis on multi-dimensional operational data using a target parameter extraction model, and extracts target optimization parameters based on the analysis results. This process ensures that the extracted parameters are highly relevant to the decision-making task and is suitable for distributed building management scenarios. Those skilled in the art can adjust the consensus algorithm or model selection according to the task type and data characteristics, without introducing complex feature decomposition or encryption processing.

[0077] In some embodiments, the step of performing semantic analysis on the multidimensional operational data based on the target parameter extraction model in the artificial intelligence module can be implemented in the following way to ensure that the optimization parameters related to the target decision-making task are accurately extracted from the multidimensional operational data, which is applicable to distributed building self-organizing decision-making scenarios.

[0078] Specifically, the multidimensional operational data undergoes multidimensional feature decomposition to obtain multiple feature subsets. The multidimensional operational data is data collected in real-time from the building system, covering dimensions such as temperature, energy consumption, humidity, and personnel density. For example, a set of data might include a temperature of 25.0℃, energy consumption of 120kWh, humidity of 60%, and personnel density of 0.5 people / m². 2These data are stored in time-series or structured formats and may originate from sensors or building management systems. The purpose of multidimensional feature decomposition is to segment these data into independent subsets based on feature type for subsequent analysis. For example, the decomposition process can divide the data into subsets such as temperature (e.g., 25.0℃, 25.2℃, 24.8℃), energy consumption (e.g., 120kWh, 115kWh, 122kWh), humidity (e.g., 60%, 58%, 61%), and population density (e.g., 0.5, 0.4, 0.6 people / m²). 2 In implementation, decomposition can be accomplished using data processing tools, such as grouping data by field and storing it as separate lists or tables. The decomposed feature subsets retain the characteristics of the original data, ensuring that subsequent analysis can be performed on specific features. Those skilled in the art will understand that the decomposition method can be adjusted according to the data structure, such as grouping by time period or device type, but does not involve encryption or consensus verification.

[0079] Furthermore, based on the target semantic analysis algorithm, semantic relevance scoring is performed on the multiple feature subsets. The target semantic analysis algorithm aims to evaluate the relevance of each feature subset to the target decision task. For example, when the task is "optimize air conditioning operating parameters to reduce energy consumption," it is necessary to determine which features have the greatest impact on energy consumption. Semantic relevance scoring can be achieved through statistical methods, such as calculating the Pearson correlation coefficient between the feature subset and the task objective. Assuming the task objective is to reduce total energy consumption, the algorithm analyzes the correlation between the energy consumption subset and historical energy consumption data, obtaining a high score (e.g., 0.9, indicating a strong correlation), while the humidity subset may have a lower score (e.g., 0.2). In specific implementation, the scoring process includes the following steps: First, determine the quantitative indicator of the task objective, such as total energy consumption; then, calculate the correlation between each feature subset and this indicator; finally, normalize the score to a range of 0 to 1. For example, the temperature subset score might be 0.8, the energy consumption subset 0.9, the humidity subset 0.2, and the personnel density subset 0.5. The scoring results are stored in a structured format, such as tables or key-value pairs, recording the correlation scores for each feature subset. Those skilled in the art will understand that the scoring method can be adapted to task requirements, for example, using the Spearman correlation coefficient instead of the Pearson coefficient, without involving encryption or broadcasting operations.

[0080] Understandably, based on the semantic relevance score, the target optimization parameters with the highest relevance to the target decision task are extracted from the multiple feature subsets. The semantic relevance score provides the degree of association between each feature subset and the task objective. For example, the energy consumption subset has the highest score of 0.9, indicating that it has the greatest impact on the "reduce energy consumption" task. The process of extracting target optimization parameters involves selecting parameters consistent with the task objective from the high-scoring subsets. For example, if the task requires reducing energy consumption, target optimization parameters are extracted from the energy consumption subsets (120kWh, 115kWh, 122kWh), such as "target energy consumption = 115kWh" (taking the average as the recommended value). The extraction process is implemented through predefined rules, such as "selecting feature subsets with scores greater than 0.7 and extracting their average or recommended values ​​as optimization parameters." If the temperature subset also has a high score (e.g., 0.8), "temperature setpoint = 24℃" may be additionally extracted as a parameter to take comfort into account. The extracted parameters are stored in a structured format, such as key-value pairs: {"parameter":"energy consumption", "value":115, "unit":"kWh"}, and recorded on the blockchain to ensure traceability. Those skilled in the art will understand that the extraction rules can be optimized based on task objectives, such as prioritizing energy-saving parameters, but do not involve instruction generation or effect evaluation.

[0081] In summary, this step extracts highly relevant optimization parameters from multi-dimensional operational data through multi-dimensional feature decomposition, semantic relevance scoring, and parameter extraction. This process ensures accurate and efficient analysis results and is suitable for distributed building management scenarios. Those skilled in the art can adjust the decomposition method, scoring method, or extraction rules according to data type and task objectives, but the above steps can be implemented without introducing encryption or consensus algorithms.

[0082] In some embodiments, the step of generating self-organizing decision instructions based on the target optimization parameters and broadcasting the self-organizing decision instructions through a blockchain network to execute building system optimization can be implemented in the following ways to ensure that the decision instructions are generated and distributed securely and transparently, which is suitable for distributed building self-organizing decision-making scenarios.

[0083] Specifically, identification information for self-organizing decision instructions is generated based on the target optimization parameters. Target optimization parameters are parameters related to the decision task extracted from multi-dimensional operational data, such as "temperature setpoint = 24℃" or "air conditioning operating power = 80%", used to optimize the operation of the building system, such as reducing energy consumption or improving comfort. Self-organizing decision instructions are control commands generated based on these parameters to guide the building system in performing specific operations. The generation of identification information aims to assign a unique identifier to each decision instruction for easy tracking and verification. For example, the identification information may include the instruction's unique ID, generation timestamp, and task objective, in the format: {"Instruction ID": "CMD202505201023", "Timestamp": "2025-05-20T22:23:00", "Objective": "Reduce energy consumption"}. The generation process is completed through templates in smart contracts. For example, the target optimization parameter "temperature = 24℃" is filled into the instruction template "Adjust air conditioning to {temperature}", generating the instruction "Adjust air conditioning to 24℃", and the aforementioned identification information is attached. The identification information is stored as structured data to ensure the traceability of instructions within the blockchain network. Those skilled in the art will understand that the format and content of the identification information can be adjusted according to system requirements, such as adding device IDs or priorities, but without involving data cleaning or semantic analysis.

[0084] Furthermore, the self-organizing decision instructions are encrypted using a smart contract to generate encrypted decision instructions. This encryption aims to protect the security and integrity of the instructions during transmission over the blockchain network, preventing unauthorized tampering or access. The smart contract embeds encryption logic, using symmetric encryption algorithms (such as AES) or asymmetric encryption algorithms (such as RSA) to encrypt the instruction content. For example, for the instruction "adjust the air conditioner to 24℃", the encryption process may generate a ciphertext string, such as a hexadecimal string, ensuring that only nodes with the correct key can decrypt it. The specific implementation of the encryption includes the following steps: First, the smart contract obtains an encryption key from a pre-configured keystore; then, the instruction content (including identification information) is combined with the key for encryption; finally, an encrypted decision instruction is generated, in the format: {"Instruction ID": "CMD202505201023", "Ciphertext": "0x7a8b9c..."}. The encrypted instructions are recorded through blockchain transactions to ensure their immutability. Those skilled in the art will understand that the choice of encryption algorithm can be adjusted according to security requirements, such as using more efficient symmetric encryption, without involving effect evaluation or anomaly handling.

[0085] Understandably, the encrypted decision instructions are broadcast through the blockchain network, and building system optimization is performed based on the decryption verification results of distributed nodes. The blockchain network uses a peer-to-peer (P2P) protocol to broadcast the encrypted decision instructions to all distributed nodes, such as edge devices or controllers deployed in the building management system. The broadcast process ensures that each node receives the same copy of the instructions, maintaining system consistency. After receiving the instructions, each node decrypts them using a pre-shared decryption key, for example, restoring the ciphertext to the original instruction "Adjust the air conditioner to 24°C" using the AES decryption algorithm. Decryption verification includes checking the integrity and legality of the instructions, such as verifying whether the timestamp in the identification information is within a valid range, or confirming whether the source of the instructions is a trusted smart contract. If the verification passes, the node transmits the instructions to the building management system to perform optimization operations, such as adjusting the temperature setpoint of the air conditioning equipment via an API call. The execution result is recorded in the form of a blockchain transaction, for example: {"Instruction ID":"CMD202505201023","Execution Status":"Success","Time":"2025-05-20T22:24:00"}. Those skilled in the art will understand that the details of the broadcast protocol and decryption verification can be adjusted according to the blockchain platform (such as Ethereum or Hyperledger Fabric), but do not involve optimization effect evaluation or anomaly handling.

[0086] In summary, this step ensures the security and transparency of instruction generation and distribution by generating identification information for self-organizing decision instructions, encrypting the generated encrypted decision instructions, and broadcasting and executing optimization operations through a blockchain network. It is suitable for distributed building management scenarios. Those skilled in the art can adjust the identification format, encryption algorithm, or broadcasting mechanism according to system requirements, but the above steps can be implemented without introducing data analysis or effect evaluation.

[0087] In some embodiments, such as Figure 2 As shown, after generating self-organizing decision instructions based on the target optimization parameters and broadcasting the self-organizing decision instructions through a blockchain network to execute building system optimization, the following steps are used to evaluate and share the optimization effect, so as to ensure that the results of building system optimization can be verified and accessed, which is suitable for distributed building system self-organizing decision-making scenarios.

[0088] Specifically, the execution result data of the self-organizing decision-making instructions is acquired. These instructions are control commands generated based on target optimization parameters, such as "adjust the air conditioner set temperature to 24℃" or "reduce the fan operating power to 80%", which are broadcast via a blockchain network and executed by the building management system or equipment. The execution result data is feedback data collected from execution nodes (such as edge devices or controllers), reflecting the actual effect after the instruction is executed. For example, if the instruction is "adjust the air conditioner to 24℃", the execution result data might include the actual temperature after execution (e.g., 24.1℃), energy consumption change (e.g., from 120kWh to 110kWh), and execution status (e.g., "success" or "failure"). This data is collected through sensors or the monitoring module of the building management system, in a structured record format, such as: {"Instruction ID": "CMD202505201023", "Actual Temperature": 24.1", "Energy Consumption": 110", "Status": "Success"}. The data acquisition process is implemented through standard protocols (such as MQTT) to ensure real-time performance and accuracy. Those skilled in the art will understand that the type and frequency of the execution result data can be adjusted according to the instruction objective, such as increasing humidity or personnel density data, but without involving anomaly handling or consensus verification.

[0089] Further, step S11: Obtain the execution result data of the self-organizing decision instruction, and generate a building system optimization effect evaluation report based on the execution result data, the multi-dimensional operational data, and the simulation verification results. The execution result data provides direct feedback on the instruction execution, while the multi-dimensional operational data includes the building's operational status before and after instruction execution, such as temperature, energy consumption, and humidity. The optimization effect evaluation report evaluates the effectiveness of the instruction by comparing the execution results with the expected goals. For example, if the goal is to reduce energy consumption by 10%, and the execution results show that energy consumption has decreased from 120kWh to 110kWh, the report will record "Energy consumption reduced by 8.3%, close to the target." The process of generating the report includes the following steps: First, extract key indicators (such as energy consumption and temperature) from the execution result data; then, compare them with historical or expected values ​​in the multi-dimensional operational data to calculate the optimization effect (such as the percentage reduction in energy consumption); next, call the simulation verification results generated in the previous simulation phase to simulate the consistency and deviation between the prediction and the actual execution; finally, organize them into a structured report, including the instruction ID, execution time, optimization indicators, and evaluation conclusions. The report format can be a table or text, for example: {“Instruction ID”:“CMD202505201023”,“Execution Time”:“2025-05-20T22:24:00”,“Energy Consumption Reduction”:“8.3%”,“Conclusion”:“Optimization effect is good”}. Those skilled in the art will understand that the report content and evaluation indicators can be adjusted according to optimization goals, such as adding comfort assessments, but will not include recommendations for handling anomalies.

[0090] Further, step S12: The optimization effect evaluation report is stored in the blockchain network, and access permissions are granted to distributed nodes according to preset sharing rules. The blockchain network provides decentralized storage, ensuring the transparency and immutability of the report. The storage process uploads the evaluation report to the blockchain as a transaction record, for example, by calling the storage function through a smart contract to write the report content into the distributed ledger. After the report is stored, a unique transaction hash (such as "0x9a8b7c...") is assigned for easy subsequent querying. The preset sharing rules define which distributed nodes can access the report, such as "only nodes with a trust level higher than 0.9 are allowed to access" or "open to all nodes participating in consensus". In implementation, the sharing rules are implemented through the access control logic of the smart contract, such as verifying access permissions based on node address or role. The access process is completed through the blockchain query interface, and the node obtains the decrypted report content after submitting a request. Those skilled in the art will understand that the storage method and sharing rules can be adjusted according to the blockchain platform, for example, using IPFS (InterPlanetary File System, a decentralized distributed file storage and sharing protocol) to store large files or setting more granular access permissions, but without involving abnormal scenario analysis.

[0091] In summary, this step, by acquiring execution result data, generating an optimization effect evaluation report, and storing it on the blockchain network, while simultaneously granting access permissions according to sharing rules, ensures the transparency and verifiability of the optimization effect, making it suitable for distributed building management scenarios. Those skilled in the art can adjust the data collection methods, evaluation indicators, or sharing rules according to system requirements, but the functionality of the above steps can be achieved without introducing anomaly handling or semantic analysis.

[0092] In some embodiments, such as Figure 3 As shown, after broadcasting the self-organizing decision instructions through the blockchain network to execute building system optimization, the following steps are used to handle abnormal scenarios, so as to ensure that the building system can quickly resume normal operation under abnormal conditions. This is applicable to self-organizing decision-making scenarios of distributed building systems.

[0093] Specifically, in step S21: when an abnormality in the building system is detected, multi-dimensional operational data under the abnormal scenario is acquired. An abnormality in the building system may include equipment failure, abnormal energy consumption, or exceeding environmental parameter limits, such as air conditioning system temperature adjustment failure or a sudden surge in energy consumption. Anomaly detection is achieved through the monitoring module of the building management system. For example, when the temperature exceeds the normal range (e.g., 20-26℃) for 10 consecutive minutes or energy consumption exceeds three times the standard deviation of the historical average, an anomaly alarm is triggered. The multi-dimensional operational data under the abnormal scenario is real-time data collected from sensors or devices, covering dimensions such as temperature, energy consumption, humidity, and equipment status. For example, if the temperature on a certain floor reaches 30℃ (abnormally high), the multi-dimensional operational data may include: {"Time": "2025-05-20T22:30:00", "Temperature": 30.0, "Energy Consumption": 150kWh, "Humidity": 65%, "Equipment Status": "Running"}. Data acquisition is completed through IoT protocols (such as CoAP) to ensure real-time performance and accuracy. Those skilled in the art will understand that the threshold for anomaly detection and the dimensions of data collection can be adjusted according to system requirements, such as adding air quality data, without involving data cleaning or feature enhancement.

[0094] Further, in step S21: anomaly handling suggestions are generated based on the multi-dimensional operational data under the abnormal scenario and historical decision data in the blockchain network. The multi-dimensional operational data under the abnormal scenario provides a snapshot of the current system state, while the historical decision data is a record of past decisions extracted from the blockchain network, such as "adjust the air conditioner to 24°C" or "shut down the faulty equipment". The process of generating anomaly handling suggestions includes the following steps: First, analyze the abnormal scenario data to identify key features of the anomaly, such as a temperature of 30°C indicating overheating; then, query the historical decision data on the blockchain to find similar handling records for abnormal scenarios, such as "increase ventilation power" in the past when the temperature exceeded the limit; finally, generate suggestions based on rules or pattern matching, such as "suggest increasing ventilation power to 90% and checking the air conditioning equipment". Suggestions are generated in a structured format, such as: {"Abnormal ID": "ERR202505201030", "Abnormal Type": "Temperature Too High", "Suggestion": "Increase ventilation power to 90%"}. Rules can be based on a predefined anomaly handling knowledge base, such as "If the temperature > 28°C, prioritize increasing ventilation". Those skilled in the art will understand that the rule generation suggestions can be adjusted according to the type of anomaly, such as suggesting a reduction in device power for anomalies in energy consumption, but this does not involve semantic analysis or optimization of parameter extraction.

[0095] Understandably, step S22 involves broadcasting the anomaly handling suggestion to distributed nodes via a smart contract and executing the anomaly handling operation based on the consensus result of the distributed nodes. The anomaly handling suggestion is broadcast via a smart contract on the blockchain network, ensuring that all distributed nodes (such as edge devices or management nodes) receive a consistent suggestion. The broadcast process uses a peer-to-peer (P2P) protocol, distributing the suggestion as a transaction record, such as a transaction containing the anomaly ID and the suggestion content. Distributed nodes perform consensus verification on the suggestion using a preset consensus algorithm (such as Practical Byzantine Fault Tolerance, PBFT). In PBFT, nodes vote to confirm the rationality of the suggestion, such as checking whether the suggestion matches the current anomaly data; consensus is reached only when more than 2 / 3 of the nodes agree. After consensus is reached, the anomaly handling operation is executed, such as adjusting the ventilation power to 90% via a building management system API call. The execution result is recorded on the blockchain, for example: {“Exception ID”:“ERR202505201030”,“Execution Status”:“Success”,“Time”:“2025-05-20T22:32:00”}. Those skilled in the art will understand that the consensus algorithm and execution method can be adjusted according to the blockchain platform, for example, using Proof-of-Stake (PoS) instead of PBFT, without involving cryptographic processing or performance evaluation.

[0096] In summary, this step, by acquiring multi-dimensional operational data under abnormal scenarios, generating anomaly handling suggestions, and broadcasting and executing them via smart contracts, ensures the rapid recovery of the building system in abnormal situations, making it suitable for distributed building management scenarios. Those skilled in the art can adjust data collection methods, suggestion generation rules, or consensus mechanisms based on the anomaly type and system architecture, but the functionality of the above steps can be achieved without introducing data cleaning or optimizing parameter extraction.

[0097] Another embodiment of the present invention also provides a blockchain-based distributed building self-organizing decision-making system 101, such as... Figure 4 As shown, the system 101 achieves decentralized decision-making and optimization functions through the following modules, which are suitable for smart building management scenarios, such as energy optimization or environmental control in commercial buildings.

[0098] Specifically, the data acquisition module 1011 is configured to acquire multi-dimensional operational data of the building system and status data of the distributed nodes when a decision request is initiated by the building system, through the data acquisition unit of the distributed nodes. The data acquisition module 1011 is deployed on the distributed nodes, such as edge computing devices or IoT gateways, and is responsible for collecting the building system's operational data and node status in real time. Multi-dimensional operational data includes temperature, energy consumption, humidity, and personnel density, originating from sensors or management systems within the building. For example, when the system detects a decision request (such as an optimization request triggered by excessive energy consumption), the module collects data via IoT protocols (such as MQTT) and generates structured records, such as: {"Time": "2025-05-20T22:32:00", "Temperature": 25.0", "Energy Consumption": 120kWh, "Humidity": 60%}. The status data of the distributed nodes includes CPU utilization, available memory, and network connection status, for example: {"Node ID": "Node1", "CPU Utilization": 30%, "Memory": 2GB}. The data acquisition unit communicates with sensors or devices through standardized interfaces to ensure the real-time nature and accuracy of the data. Those skilled in the art will understand that the frequency and dimensions of data acquisition can be adjusted according to the building type; for example, high-frequency acquisition may be used for dynamic control but does not involve data cleaning or feature enhancement.

[0099] Furthermore, the task generation module 1012 is configured to generate the target decision task through a smart contract in the blockchain network based on the multi-dimensional operational data and the state data. The task generation module 1012 runs on the smart contract layer of the blockchain network, using the multi-dimensional operational data and state data to generate the decision task. For example, if multi-dimensional operational data shows that the energy consumption of a certain floor is 120kWh (higher than the threshold of 100kWh), and the state data indicates that multiple nodes have sufficient computing power, the module generates the task "Optimize the air conditioning operating parameters of this floor" through a smart contract. The smart contract contains predefined task generation rules, such as "If energy consumption > 100kWh and node CPU utilization < 50%, then generate an energy-saving task." The generated task is stored in a structured format, for example: {"Task ID": "TASK202505201032", "Target": "Reduce energy consumption", "Area": ​​"Floor A"}, and recorded on the blockchain to ensure transparency. The task generation process is completed automatically through the smart contract, without manual intervention. Those skilled in the art will understand that task generation rules can be adjusted according to optimization goals, such as comfort optimization, but do not involve scenario classification or trust level assessment.

[0100] Understandably, the parameter extraction module 1013 is configured to extract target optimization parameters from the multi-dimensional operational data through the artificial intelligence module when a consensus is reached among distributed nodes based on the target decision task and a preset consensus algorithm. The parameter extraction module 1013 works in conjunction with the consensus mechanism of the blockchain network and the artificial intelligence module. After a task is generated, distributed nodes execute the task and submit results, such as calculating the optimal air conditioning temperature. A preset consensus algorithm (such as Practical Byzantine Fault Tolerance, PBFT) verifies the consistency of the results, for example, confirming that a majority of nodes recommend a temperature of 24°C. After consensus is reached, the artificial intelligence module (such as a pre-trained regression model) analyzes the multi-dimensional operational data and extracts optimization parameters, such as "temperature setpoint = 24°C". The analysis process is based on rules or simple models, such as comparing the historical relationship between energy consumption and temperature to determine energy-saving parameters. The extracted parameters are stored in a structured format, such as {"parameter":"temperature","value":24","unit":"°C"}, and recorded on the blockchain. Those skilled in the art will understand that consensus algorithms and artificial intelligence models can be adjusted according to system requirements, such as using decision tree models, but without involving feature decomposition or encryption.

[0101] Furthermore, the decision execution module 1014 is configured to generate self-organizing decision instructions based on the target optimization parameters, and to simulate and verify the self-organizing decision instructions based on the multi-dimensional operational data to predict the building system's operational status after executing the self-organizing decision instructions. If the simulation result meets the preset optimization target, the self-organizing decision instructions are broadcast through the blockchain network to execute building system optimization; otherwise, the self-organizing decision instructions are adjusted and the simulation verification is repeated. After passing the simulation verification, the decision execution module 1014 generates instructions based on the extracted parameters, for example, converting "temperature = 24℃" into the instruction "adjust the air conditioner to 24℃". After the instructions are generated, they are distributed to all relevant nodes, such as the building management system or equipment controller, through the peer-to-peer (P2P) broadcast protocol of the blockchain network. The broadcast instructions are recorded in the form of transactions, for example: {"Instruction ID": "CMD202505201033", "Content": "Adjust the air conditioner to 24℃"}. Receiving nodes execute the instructions, for example, by adjusting equipment parameters through API calls to complete the optimization operation. The execution results are recorded on the blockchain to ensure traceability. Those skilled in the art will understand that the instruction format and broadcast protocol can be adapted to a blockchain platform, such as using Ethereum's transaction mechanism, but without involving encryption or effect evaluation.

[0102] In summary, this system 101, through its data acquisition module 1011, task generation module 1012, parameter extraction module 1013, and decision execution module 1014, realizes a complete process from data collection to decision execution, ensuring the decentralization, transparency, and efficiency of building system optimization, and is suitable for distributed building management scenarios. Those skilled in the art can adjust the module functions according to building requirements, such as optimizing the data collection frequency or consensus algorithm, without needing to introduce data cleaning or anomaly handling to achieve the system's functionality.

[0103] The above description is merely an exemplary embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention specification and drawings under the technical concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A blockchain-based distributed building self-organizing decision method, characterized in that, The method comprises: When detecting that the building system initiates a decision request, obtaining multi-dimensional operation data of the building system and state data of the distributed nodes; According to the multi-dimensional operation data and the state data, generating a target decision task through a smart contract in a blockchain network; When it is determined that the distributed nodes reach a decision consensus according to the target decision task and a preset consensus algorithm, extracting a target optimization parameter according to the multi-dimensional operation data, wherein the step of extracting the target optimization parameter according to the multi-dimensional operation data when it is determined that the distributed nodes reach the decision consensus according to the target decision task and the preset consensus algorithm comprises: performing consistency verification on the task execution results of the distributed nodes through the preset consensus algorithm according to the target decision task; when the verification is passed, performing semantic analysis on the multi-dimensional operation data according to a target parameter extraction model in an artificial intelligence module; extracting a target optimization parameter related to the target decision task according to the semantic analysis result; and the step of performing semantic analysis on the multi-dimensional operation data according to the target parameter extraction model in the artificial intelligence module comprises: performing multi-dimensional feature decomposition on the multi-dimensional operation data to obtain a plurality of feature subsets; performing semantic correlation scoring on the plurality of feature subsets based on a target semantic analysis algorithm; and extracting a target optimization parameter with the highest correlation degree with the target decision task from the plurality of feature subsets according to the semantic correlation scoring; Generating a self-organizing decision instruction according to the target optimization parameter, and performing simulation verification on the self-organizing decision instruction according to the multi-dimensional operation data to predict the operation state of the building system after the self-organizing decision instruction is executed; if the simulation result meets a preset optimization target, broadcasting the self-organizing decision instruction through the blockchain network to execute building system optimization; otherwise, adjusting the self-organizing decision instruction and repeating the simulation verification. The step of generating a self-organizing decision instruction according to the target optimization parameter and broadcasting the self-organizing decision instruction through a blockchain network to perform building system optimization comprises: generating identification information of the self-organizing decision instruction according to the target optimization parameter; performing encryption processing on the self-organizing decision instruction through a smart contract to generate an encrypted decision instruction; broadcasting the encrypted decision instruction through the blockchain network and performing building system optimization according to a decryption verification result of a distributed node; and after the step of generating a self-organizing decision instruction according to the target optimization parameter and broadcasting the self-organizing decision instruction through a blockchain network to perform building system optimization, the method further comprises: obtaining execution result data of the self-organizing decision instruction; generating a building system optimization effect evaluation report according to the execution result data, the multi-dimensional running data and the simulation verification result; storing the optimization effect evaluation report to the blockchain network and opening access authority to the distributed node according to a preset sharing rule; and after the step of generating a self-organizing decision instruction according to the target optimization parameter and broadcasting the self-organizing decision instruction through a blockchain network to perform building system optimization, the method further comprises: when detecting that the building system is running abnormally, obtaining multi-dimensional running data under an abnormal scene; generating an abnormal processing suggestion according to the multi-dimensional running data under the abnormal scene and historical decision data in the blockchain network; broadcasting the abnormal processing suggestion to the distributed node through the smart contract and performing an abnormal processing operation according to a consensus result of the distributed node.

2. The method of claim 1, wherein, The step of generating a target decision task through a smart contract in the blockchain network according to the multi-dimensional running data and the state data comprises: performing standardization processing on the multi-dimensional running data to generate a target running data set in a unified format; performing data cleaning on the target running data set according to a target denoising algorithm and performing feature enhancement on the cleaned data to obtain optimization running data; generating a target decision task through a task allocation logic in the smart contract according to the optimization running data and the state data, wherein the target decision task comprises a prediction parameter required for supporting simulation verification of the self-organizing decision instruction.

3. The method of claim 2, wherein, The step of generating a target decision task through a task allocation logic in the smart contract according to the optimization running data and the state data comprises: performing scene classification on the optimization running data based on a target clustering algorithm to obtain a plurality of building running scene features; determining a computing capability and a trust level of each distributed node according to the state data; generating a target decision task through a task allocation logic of the smart contract according to the building running scene features, the computing capability and the trust level.

4. The method of claim 2, wherein, Before the step of performing data cleaning on the target running data set according to a target denoising algorithm, the method further comprises: obtaining historical multi-dimensional running data and corresponding historical decision result data; generating a data cleaning training sample set according to the historical multi-dimensional running data and the historical decision result data; training a target denoising model according to a target machine learning algorithm and the data cleaning training sample set; According to a target loss function, a loss value between a predicted value and an actual value of the target denoising model is calculated, and a parameter of the target denoising model is adjusted by an optimization algorithm, so that the loss value is less than a preset threshold.

5. A blockchain-based distributed building self-organizing decision system, characterized in that, The system is configured to perform the method of any one of claims 1-4, and the system comprises: a data acquisition module configured to acquire multi-dimensional operation data of the building system and state data of the distributed nodes through a data acquisition unit of the distributed nodes when a decision request of the building system is detected; a task generation module configured to generate the target decision task through a smart contract in the blockchain network according to the multi-dimensional operation data and the state data; a parameter extraction module configured to extract a target optimization parameter according to the multi-dimensional operation data through the artificial intelligence module when a decision consensus of the distributed nodes is determined according to the target decision task and a preset consensus algorithm, wherein the parameter extraction includes consistency verification, semantic analysis, multi-dimensional feature decomposition, semantic correlation scoring, and correlation degree highest parameter extraction; a decision execution module configured to generate a self-organizing decision instruction according to the target optimization parameter, and to simulate and verify the self-organizing decision instruction according to the multi-dimensional operation data to predict a running state of the building system after the self-organizing decision instruction is executed; if a simulation result meets a preset optimization target, the self-organizing decision instruction is broadcast through the blockchain network to execute building system optimization; otherwise, the self-organizing decision instruction is adjusted and the simulation verification is repeated, an execution result data is acquired after execution, an optimization effect evaluation report is generated according to the execution result data, the multi-dimensional operation data, and the simulation verification result, the optimization effect evaluation report is stored in the blockchain network, and access permission is opened to the distributed nodes according to a preset sharing rule; an exception processing module configured to acquire multi-dimensional operation data under an abnormal scenario when an abnormal running of the building system is detected, to generate an exception handling suggestion according to the multi-dimensional operation data under the abnormal scenario and historical decision data in the blockchain network, to broadcast the exception handling suggestion to the distributed nodes through the smart contract, and to execute an exception handling operation according to a consensus result of the distributed nodes.

Images