Intelligent energy management system and method based on multi-dimensional data analysis

Abstract

The invention discloses an intelligent energy management system and method based on multi-dimensional data analysis, and relates to the technical field of energy management, and the method comprises the steps: collecting equipment energy data, external data and behavior data; performing real-time preprocessing on the data through a distributed computing architecture of edge nodes; capturing a long-term dependency relationship in the time sequence data, and predicting a future energy demand; a multi-objective optimization algorithm is introduced to balance the relationship between energy consumption and customer experience, a strategy library is constructed, and a preliminary strategy is automatically generated; when it is monitored that the actual energy consumption deviation prediction value exceeds a deviation threshold value, the environment comfort index exceeds a comfort threshold value range or a specific user behavior mode is recognized, a strategy dynamic adjustment mechanism is triggered; and real-time distribution and execution of strategies are realized through a distributed collaborative algorithm among edge nodes. According to the invention, the energy consumption is effectively reduced while the requirement of customers on the environmental comfort is met.

Description

Technical Field

[0001] The present invention relates to the field of energy management technology, and in particular to an intelligent energy management system and method based on multi-dimensional data analysis. Background Art

[0002] With the continuous growth of global energy demand and the increasingly tight energy supply, energy management has become a focus of attention across various industries. Traditional energy management methods rely primarily on manual experience and simple rules for energy regulation. They lack accurate predictions of energy consumption and comprehensive consideration of multiple factors, making it difficult to find the optimal balance between energy consumption and customer experience.

[0003] Traditional methods often fail to accurately capture dynamic changes and long-term trends in energy consumption, leading to a mismatch between energy supply and actual demand, resulting in energy waste or insufficient supply. For example, in some commercial venues, the inability to properly adjust the operating parameters of energy equipment based on factors such as real-time customer traffic and weather changes leads to unnecessary increases in energy consumption.

[0004] Traditional energy management pays little attention to the customer experience. In the pursuit of energy conservation, it may reduce environmental comfort, affecting customer satisfaction and loyalty. For example, excessively lowering the air conditioning temperature setting or lighting brightness to save energy may make customers feel uncomfortable.

[0005] Traditional energy management systems have slow data processing and decision-making speeds, making them unable to respond promptly to sudden changes and anomalies in energy consumption. Furthermore, the lack of effective coordination between various energy devices makes it difficult to achieve overall optimized operation. Summary of the Invention

[0006] The purpose of the present invention is to provide an intelligent energy management system and method based on multi-dimensional data analysis to solve the problems raised in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent energy management method based on multi-dimensional data analysis, the method comprising:

[0008] Collect equipment energy data, external data and behavioral data;

[0009] Pre-process data in real time through the distributed computing architecture of edge nodes;

[0010] Capturing long-term dependencies in time series data to predict future energy demand;

[0011] Introducing a multi-objective optimization algorithm to balance the relationship between energy consumption and customer experience, building a strategy library, and automatically generating preliminary strategies;

[0012] When the actual energy consumption deviation from the predicted value exceeds the deviation threshold, the environmental comfort index exceeds the comfort threshold range, or a specific user behavior pattern is identified, the policy dynamic adjustment mechanism is triggered;

[0013] Through the distributed collaborative algorithm among edge nodes, real-time distribution and execution of strategies are achieved.

[0014] According to the above scheme, the device energy data includes the device's power, operating status, health status, operating time and energy consumption; the external data includes weather data, holiday information, special times and real-time passenger flow data; the behavioral data includes stay area, stay time and environmental interaction data.

[0015] According to the above solution, the real-time preprocessing includes:

[0016] After receiving the collected data, each edge node performs data validity verification to eliminate outliers and missing data; a unified timestamp alignment mechanism is used to establish data timing association across devices.

[0017] According to the above scheme, the method of capturing long-term dependencies in time series data and predicting future energy demand includes:

[0018] Integrate the device energy data, external data, and behavioral data, unify the data format and scope, and construct a multi-dimensional data set;

[0019] Constructing an energy forecasting model, inputting a multi-dimensional data set, capturing long-term dependencies in time series data, extracting key features, and outputting future energy demand forecast results; the future energy demand forecast includes short-term forecasts and long-term forecasts;

[0020] The energy prediction model parameters are optimized using the back propagation algorithm and the gradient descent method through the multi-dimensional data set.

[0021] According to the above scheme, the initial strategy includes:

[0022] Construct a multi-objective optimization model and define an optimization objective function; the optimization objective function includes minimizing energy consumption and maximizing customer experience; set constraints, including equipment operation constraints, environmental comfort constraints, and total energy constraints; the equipment operation constraints include maximum power limits and minimum start-stop intervals; the environmental comfort constraints include comfort range thresholds for temperature, humidity, and illumination;

[0023] Randomly generating a number of device operating parameter combinations to construct an initial population; the device operating parameter combinations include adjustable variables;

[0024] The system uses energy demand forecast results, equipment operating status, environmental parameters, and behavioral data as inputs and iterates. Each iteration includes a complete optimization loop, which includes calculating the objective function value of each individual in the current population, performing non-dominated sorting to stratify the solution set, maintaining the diversity distribution of solutions through congestion calculation, and generating a new generation of populations through selection, crossover, and mutation operations. When the Pareto front improvement rate falls below a set threshold for three consecutive generations or reaches the maximum number of iterations, the calculation is terminated and the Pareto optimal solution set containing non-dominated solutions is output.

[0025] Analyze the non-dominated solutions in the Pareto optimal solution set and extract the device operating parameter combinations corresponding to the non-dominated solutions; establish a solution set mapping relationship table, wherein the solution set mapping relationship table associates each non-dominated solution with an executable device control instruction set to form a strategy library;

[0026] The policy library includes device control parameters, execution conditions and expected effects.

[0027] According to the above solution, the automatic generation of a preliminary strategy includes:

[0028] Constructing a decision model, wherein the decision factors of the decision model include current time period characteristics, real-time energy fluctuation rate, real-time energy equipment health status, passenger flow density change rate and user behavior pattern;

[0029] By using the decision model, the comprehensive fitness of each non-dominated solution in the strategy library is evaluated, the non-dominated solution with the highest comprehensive fitness is selected, and the associated executable device control instruction set is obtained through the solution set mapping relationship table to generate a preliminary strategy;

[0030] The preliminary strategy generates a hierarchical control instruction sequence including immediate execution instructions, delayed execution instructions and conditional trigger instructions;

[0031] During the implementation of the strategy, the energy data, external data and behavioral data of the equipment are monitored in real time, and a dynamic adjustment trigger mechanism is set up. The trigger mechanism includes parameter deviation trigger based on threshold, immediate trigger in response to emergencies and periodic timed trigger;

[0032] A multi-level fault tolerance mechanism is constructed, which includes primary fault tolerance, intermediate fault tolerance and advanced fault tolerance; the primary fault tolerance includes automatic retry of failure instructions, the intermediate fault tolerance includes switching to a backup strategy, and the advanced fault tolerance includes enabling a safe mode and issuing an alarm.

[0033] According to the above solution, the dynamic policy adjustment mechanism includes:

[0034] Based on the execution effect of the current policy, when it is detected that the actual energy consumption deviates from the predicted value by more than the deviation threshold, the environmental comfort index exceeds the comfort threshold range, or a specific user behavior pattern is recognized, a policy dynamic adjustment mechanism is triggered; the policy dynamic adjustment mechanism includes locking the target device control parameters that generate deviations in the current policy and retaining the device control parameters of the unaffected devices; based on the policy library, extracting associated non-dominated solutions and performing local re-optimization; combining the current time period characteristics, real-time energy volatility, real-time energy device health status, passenger flow density change rate, and user behavior pattern, and screening out the standby policy with the highest comprehensive fitness from the Pareto optimal solution set through a decision model, and converting the standby policy into a hierarchical control instruction sequence;

[0035] Coordinate the control instructions of each region through the distributed cooperation mechanism of the edge nodes. When multiple policies act on the same device simultaneously, the instructions are fused according to the preset priority.

[0036] According to the above solution, the real-time distribution and execution of the policy are achieved through the distributed cooperation algorithm between edge nodes, including:

[0037] After each edge node receives the hierarchical control instruction sequence, based on the device energy data and the external data, perform instruction adaptability verification to determine the priority and weight distribution of instruction execution; establish a message bus based on the publish-subscribe mode, exchange device status and instruction execution feedback through a lightweight communication protocol, and complete the status synchronization between nodes;

[0038] When detecting multi-node instruction conflicts, adopt the last-write-wins algorithm based on timestamps to coordinate the execution order; the timestamp of instruction I1 is T1, and the timestamp of instruction I2 is T2. If T1 < T2, then execute I1 first; where I represents the index of the instruction number and T represents the timestamp;

[0039] For composite instructions involving multi-region linkage, the two-phase commit protocol is initiated by the leading node to ensure atomic execution. The leading node sends a prepare instruction to all participating nodes, and each node checks its own status and returns a ready or failed message. If all nodes are ready, the leading node sends a commit instruction, and each node executes the instruction; if a node fails, the leading node sends a rollback instruction; at the same time, monitor the execution status of each node through a heartbeat mechanism; the real-time data updates generated during the execution process are fed back to the central processing unit through the backpropagation mechanism to form a closed-loop control flow.

[0040] An intelligent energy management system based on multi-dimensional data analysis, the system includes: a data collection and processing module, an intelligent analysis module, a dynamic execution module, a distributed cooperation module, and an interaction interface module;

[0041] The data acquisition and processing module includes a multi-source data acquisition module and an edge pre-processing module; the multi-source data acquisition module is used to collect device energy data, external data, and behavioral data in real time; the edge pre-processing module is used to complete data cleaning and time sequence alignment at the edge node to complete data standardization;

[0042] The intelligent analysis module includes an energy forecasting module, a strategy generation module, a strategy decision module, and a strategy library module; the energy forecasting module constructs an energy forecasting model to predict future energy demand, including short-term and long-term forecasts; the strategy generation module constructs a multi-objective optimization model to generate a Pareto optimal solution set that balances energy consumption and customer experience; the strategy decision module selects the optimal control strategy based on the data collected by the data acquisition and processing module; and the strategy library module is used to store and manage strategies;

[0043] The dynamic execution module includes an instruction distribution module, a conflict coordination module and a fault tolerance control module; the instruction distribution module is used to dispatch hierarchical instructions to edge nodes; the conflict coordination module is used to resolve multi-node instruction conflicts and ensure execution consistency; the fault tolerance control module has a real-time three-level fault tolerance mechanism;

[0044] The distributed collaboration module includes an edge collaboration module and a resource scheduling module; the edge collaboration module realizes state synchronization between nodes through a publish-subscribe mechanism; the resource scheduling module dynamically allocates edge computing resources;

[0045] The interactive interface module includes a visualization module and a strategy configuration module; the visualization module displays the real-time energy consumption heat map and passenger flow distribution data of the equipment in each area; the strategy configuration module predicts the energy saving rate and comfort change based on the keywords selected by the user.

[0046] According to the above scheme, when data anomalies are monitored, a hierarchical alarm mechanism is triggered; the alarm levels include emergency, warning and prompt; the emergency is a red signal, the warning is a yellow signal, and the prompt is a blue signal; the maintenance record is automatically associated with the equipment, the health status is updated, and an alarm report is generated; the alarm report includes the alarm level, alarm time, alarm area or equipment, alarm maintenance record and equipment health status.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. The present invention uses multi-dimensional data to plan energy supply and improve energy utilization efficiency;

[0049] 2. The present invention not only meets customers’ demands for environmental comfort, but also effectively reduces energy consumption;

[0050] 3. The present invention sets up a dynamic strategy adjustment mechanism to ensure that the energy management strategy always adapts to changes in actual conditions, thereby improving the flexibility and adaptability of energy management;

[0051] 4. The present invention realizes the real-time distribution and execution of strategies through the distributed collaborative algorithm among edge nodes, ensuring that various energy devices can work together and improving the operating efficiency of the overall energy management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flowchart of the steps of an intelligent energy management method based on multi-dimensional data analysis of the present invention;

[0053] Figure 2 This is a structural diagram of an intelligent energy management system based on multi-dimensional data analysis in the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0055] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, an intelligent energy management method based on multi-dimensional data analysis, the method comprising:

[0056] S1, collects equipment energy data, external data and behavioral data;

[0057] Specifically, for example, smart energy management of coffee shops is carried out, and equipment energy data, external data and behavioral data are collected; for example: the collected equipment energy data include: coffee machine: current power is 1.8kW (maximum 2.2kW), and the cumulative usage time today is 4.2 hours; air conditioner: set temperature 24℃, actual temperature 25.5℃, and the remaining filter life is 83%; the collected external data include: weather: outdoor temperature 30℃, humidity 70%; time period characteristics: Saturday morning at 10:30 (peak holiday period); the collected behavioral data include: customer distribution: window seat occupancy rate is 90%, and there are 5 standing customers in the bar area; special behavior: the customer at table 3 stayed for 45 minutes.

[0058] S2, pre-processes the data in real time through the distributed computing architecture of edge nodes;

[0059] Specifically, after receiving the collected data, each edge node performs data validity verification to eliminate outliers and missing data; a unified timestamp alignment mechanism is used to establish data timing association across devices.

[0060] S3, capture long-term dependencies in time series data and predict future energy demand;

[0061] Specifically, the device energy data, external data and behavioral data are integrated, the data format and range are unified, and a multi-dimensional data set is constructed; an energy prediction model is constructed, the multi-dimensional data set is input, the long-term dependencies in the time series data are captured, key features are extracted, and future energy demand prediction results are output; the future energy demand prediction includes short-term prediction and long-term prediction; the energy prediction model parameters are optimized using the back propagation algorithm and gradient descent method through the multi-dimensional data set; for example: input historical energy consumption: the average energy consumption in the same period of the past 7 days is 8.2kWh; environmental parameters: indoor and outdoor temperature difference is 6.5℃, and the sunlight intensity is 850W / m 2 ; Passenger flow pattern: Peak passenger flow characteristics on weekend mornings (+40% compared to weekdays); Equipment status: Two coffee machines are online at the same time, and the freezer defrost cycle is started; Output prediction results: Short-term prediction (next hour): Total energy consumption demand: 9.5±0.8kWh; Due to peak passenger flow and outdoor temperature, the air conditioning load will increase by 25%; Long-term prediction (on the same day): A second wave of passenger flow peak will occur between 3 and 4 pm.

[0062] S4. Introduce a multi-objective optimization algorithm to balance the relationship between energy consumption and customer experience, build a strategy library, and automatically generate preliminary strategies;

[0063] Specifically, a multi-objective optimization model is constructed and an optimization objective function is defined; the optimization objective function includes minimizing energy consumption and maximizing customer experience; for example: controlling energy consumption within 90% of the budget (<8.5kWh / h), and maintaining the customer area comfort score ≥4.2 points (total score 5 points); setting constraints, the constraints include equipment operation constraints, environmental comfort constraints and total energy constraints; the equipment operation constraints include maximum power limits and minimum start-stop intervals; the environmental comfort constraints include comfort range thresholds for temperature, humidity and illumination; randomly generating several groups of equipment operation parameter combinations to construct an initial population; the equipment operation parameter combinations include adjustable variables; using energy demand forecast results, equipment operation status, environmental parameters and behavior data as inputs, and iterating; each iteration contains a complete optimization cycle, the optimization cycle includes calculating the objective function value of each individual in the current population, performing non-dominated sorting to stratify the solution set, maintaining the diversity distribution of the solution through congestion calculation, and selecting, crossover and mutation. The operation generates a new generation of population; when the Pareto front improvement rate is lower than the set threshold for three consecutive generations or reaches the maximum number of iterations, the calculation is terminated, and the Pareto optimal solution set containing non-dominated solutions is output; the non-dominated solutions in the Pareto optimal solution set are analyzed, and the equipment operating parameter combinations corresponding to the non-dominated solutions are extracted; a solution set mapping relationship table is established, and the solution set mapping relationship table associates each non-dominated solution with an executable equipment control instruction set to form a strategy library; the strategy library includes equipment control parameters, execution conditions and expected effects; for example: Strategy 1: coffee machine mode: coffee machine A is turned on, coffee machine B is turned on; air conditioning setting: 26℃; lighting adjustment: area A increases brightness by 20%; expected energy consumption is 7.8kWh; customer area comfort score: 4.3; Strategy 2: coffee machine mode: coffee machine A is turned on, coffee machine B is turned off; air conditioning setting: 25℃; lighting adjustment: area B increases brightness by 20%; expected energy consumption is 7.2kWh; customer area comfort score: 4.0; this is only an example for illustration and is not limiting;

[0064] Furthermore, the decision factors of the decision model include current time period characteristics, real-time energy fluctuation rate, real-time energy equipment health status, passenger flow density change rate and user behavior pattern; through the decision model, the comprehensive fitness of each non-dominated solution in the strategy library is evaluated, the non-dominated solution with the highest comprehensive fitness is selected, and the associated executable device control instruction set is obtained through the solution set mapping relationship table to generate a preliminary strategy; the preliminary strategy generates a hierarchical control instruction sequence including immediate execution instructions, delayed execution instructions and conditional trigger instructions; for example: select strategy 1, because the passenger flow is large on weekend mornings, the coffee production speed needs to be guaranteed, and the brightness improvement in area A can increase the customer's stay time.

[0065] Furthermore, during the implementation of the strategy, the energy data, external data and behavioral data of the equipment are monitored in real time, and a dynamic adjustment trigger mechanism is set, which includes parameter deviation triggering based on thresholds, immediate triggering in response to emergencies, and periodic timing triggering; a multi-level fault tolerance mechanism is constructed, which includes primary fault tolerance, intermediate fault tolerance, and advanced fault tolerance; the primary fault tolerance includes automatic retry failure instructions, the intermediate fault tolerance includes switching to backup strategies, and the advanced fault tolerance includes enabling safety mode and alarming; for example: coffee machine overheating alarm: primary fault tolerance: automatic retry startup (2 times with an interval of 5 minutes); intermediate fault tolerance: switch to the backup machine and notify maintenance; advanced fault tolerance: shut down the faulty equipment and trigger electrical detection.

[0066] S5. When the actual energy consumption deviation from the predicted value exceeds the deviation threshold, the environmental comfort index exceeds the comfort threshold range, or a specific user behavior pattern is identified, the policy dynamic adjustment mechanism is triggered;

[0067] Specifically, based on the current strategy execution effect, when it is monitored that the actual energy consumption deviates from the predicted value by more than the deviation threshold, the environmental comfort index exceeds the comfort threshold range, or a specific user behavior pattern is identified, the strategy dynamic adjustment mechanism is triggered; the strategy dynamic adjustment mechanism includes locking the target device control parameters that cause deviations in the current strategy and retaining the device control parameters of unaffected devices; based on the strategy library, extracting associated non-dominated solutions and performing local re-optimization; combining the current time period characteristics, real-time energy fluctuation rate, real-time energy device health status, passenger flow density change rate and user behavior pattern, through the decision model, screen out the backup strategy with the highest comprehensive fitness from the Pareto optimal solution set, and convert the backup strategy into a hierarchical control instruction sequence; for example: identifying a specific user behavior pattern: when a customer sits down in area A, the brightness of area A is increased by 20%; when it is identified that the seat in area B is vacant for more than 5 minutes, turn off the spotlight of the corresponding seat in area B; this is only an example and is not limited.

[0068] Furthermore, the distributed collaboration mechanism of edge nodes is used to coordinate control instructions for each region. When multiple policies act on the same device at the same time, instructions are integrated according to preset priorities.

[0069] S6, through the distributed collaborative algorithm between edge nodes, to achieve real-time distribution and execution of strategies;

[0070] Specifically, after receiving the hierarchical control instruction sequence, each edge node performs instruction adaptability verification based on the device energy data and the external data, and determines the priority and weight distribution of instruction execution. A message bus based on the publish-subscribe model is established to exchange device status and instruction execution feedback through a lightweight communication protocol to achieve state synchronization between nodes.

[0071] Furthermore, when a multi-node instruction conflict is detected, a timestamp-based eventual consistency algorithm is used to coordinate the execution order; for complex instructions involving multi-region linkage, the dominant node initiates a two-phase commit protocol to ensure atomic execution, and at the same time monitors the execution status of each node through a heartbeat mechanism; the real-time data updates generated during the execution process are fed back to the central processing unit through the backpropagation mechanism, forming a closed-loop control flow.

[0072] The present invention provides another technical solution, an intelligent energy management method based on multi-dimensional data analysis and multi-command conflict coordination;

[0073] For example: Area A and Area B of a coffee shop share the same central air-conditioning unit;

[0074] Based on the current data, two strategies are triggered simultaneously: Strategy X: The temperature in Area A needs to be lowered to 24°C (due to high customer traffic and the high temperature in Area A of the coffee shop); Strategy Y: The temperature in Area B needs to be raised to 27°C (due to decreased customer traffic and equipment overheating alarms).

[0075] After receiving the hierarchical control instruction sequence, each edge node performs instruction adaptability verification based on the device energy data and the external data to determine the priority and weight distribution of instruction execution. According to the edge node verification, strategy X has a priority of 3 and strategy Y has a priority of 2. Strategy X is a comfort strategy with a weight coefficient of 0.6, and strategy Y is an energy efficiency strategy with a weight coefficient of 0.3.

[0076] Use a timestamp-based eventual consistency algorithm to coordinate execution order;

[0077] The final implementation strategy is to adopt a weighted compromise value of 25.5℃, with additional compensation measures: increasing directional air supply in area A and turning on local circulation fans in area B.

[0078] The present invention provides another technical solution, an intelligent energy management system based on multi-dimensional data analysis, which includes: a data acquisition and processing module, an intelligent analysis module, a dynamic execution module, a distributed collaboration module and an interactive interface module;

[0079] The data acquisition and processing module includes a multi-source data acquisition module and an edge pre-processing module; the multi-source data acquisition module is used to collect device energy data, external data, and behavioral data in real time; the edge pre-processing module is used to complete data cleaning and time sequence alignment at the edge node to complete data standardization;

[0080] The intelligent analysis module includes an energy forecasting module, a strategy generation module, a strategy decision module, and a strategy library module; the energy forecasting module constructs an energy forecasting model to predict future energy demand, including short-term and long-term forecasts; the strategy generation module constructs a multi-objective optimization model to generate a Pareto optimal solution set that balances energy consumption and customer experience; the strategy decision module selects the optimal control strategy based on the data collected by the data acquisition and processing module; and the strategy library module is used to store and manage strategies;

[0081] The dynamic execution module includes an instruction distribution module, a conflict coordination module and a fault tolerance control module; the instruction distribution module is used to dispatch hierarchical instructions to edge nodes; the conflict coordination module is used to resolve multi-node instruction conflicts and ensure execution consistency; the fault tolerance control module has a real-time three-level fault tolerance mechanism;

[0082] The distributed collaboration module includes an edge collaboration module and a resource scheduling module; the edge collaboration module realizes state synchronization between nodes through a publish-subscribe mechanism; the resource scheduling module dynamically allocates edge computing resources;

[0083] The interactive interface module includes a visualization module and a strategy configuration module; the visualization module displays the real-time energy consumption heat map and passenger flow distribution data of the equipment in each area; the strategy configuration module predicts the energy saving rate and comfort change based on the keywords selected by the user.

[0084] Specifically, when data anomalies are monitored, a graded alarm mechanism is triggered; the alarm levels include emergency, warning, and prompt; the emergency is a red signal, the warning is a yellow signal, and the prompt is a blue signal; the maintenance record is automatically associated with the equipment, the health status is updated, and an alarm report is generated; the alarm report includes the alarm level, alarm time, alarm area or equipment, alarm maintenance record, and equipment health status.

[0085] The present invention provides another technical solution, an intelligent energy management system based on multi-dimensional data analysis, and an interactive interface module;

[0086] Users can complete basic policy configuration through this module, for example: the system provides a variety of preset mode options: in peak business mode, the air conditioner will be automatically set to a constant temperature of 24°C, and the lighting system will remain at full brightness, giving priority to ensuring customer comfort; the energy-saving maintenance mode is applicable to non-business hours, the air conditioner temperature is adjusted to 28°C, and all display cabinet lighting and display equipment power are automatically turned off; the event special mode can extend the operating time of specific equipment to meet the needs of special events; this is only an example, not a limitation. Users can adjust automatic policies in real time through the interface, for example: temporarily turning off the background music system or adjusting the lighting brightness in a specific area; and the interface provides a policy execution effect visualization function, which intuitively displays the energy consumption distribution and execution effect of each area in the form of a heat map, helping users quickly understand the implementation of the policy;

[0087] When a device experiences an anomaly, the interactive interface pushes real-time alerts, including device location, anomaly type, and recommended actions. Users can confirm the action or manually adjust relevant device parameters directly within the interface. All policy adjustments and execution results are automatically saved, forming a comprehensive operation log and energy efficiency report, which can be queried and analyzed by date, region, or device type. Based on historical operation data and user habits, the system intelligently recommends optimization strategies, such as "Current passenger flow is low, recommend switching to energy-saving mode, which is expected to save 15% energy." For critical operations, the system incorporates a dual-verification mechanism to ensure the security of policy changes. The interactive interface is deeply integrated with other system modules, and any adjustments are synchronized to edge nodes for execution in real time, with the results fed back to the system, forming a complete control loop.

[0088] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. An intelligent energy management method based on multi-dimensional data analysis, characterized by: The method includes: Collect equipment energy data, external data and behavioral data; Pre-process data in real time through the distributed computing architecture of edge nodes; Capturing long-term dependencies in time series data to predict future energy demand; Introducing a multi-objective optimization algorithm to balance the relationship between energy consumption and customer experience, building a strategy library, and automatically generating preliminary strategies; When the actual energy consumption deviation from the predicted value exceeds the deviation threshold, the environmental comfort index exceeds the comfort threshold range, or a specific user behavior pattern is identified, the policy dynamic adjustment mechanism is triggered; Through the distributed collaborative algorithm among edge nodes, real-time distribution and execution of strategies are achieved.

2. The intelligent energy management method based on multi-dimensional data analysis according to claim 1, characterized in that: The device energy data includes the device's power, operating status, health status, operating time and energy consumption; the external data includes weather data, holiday information, special times and real-time passenger flow data; the behavioral data includes stay area, stay time and environmental interaction data.

3. The intelligent energy management method based on multi-dimensional data analysis according to claim 1, characterized in that: The real-time preprocessing includes: After receiving the collected data, each edge node performs data validity verification to eliminate outliers and missing data; a unified timestamp alignment mechanism is used to establish data timing association across devices.

4. The intelligent energy management method based on multi-dimensional data analysis according to claim 1, characterized in that: The method captures long-term dependencies in time series data and predicts future energy demand, including: Integrate the device energy data, external data, and behavioral data, unify the data format and scope, and construct a multi-dimensional data set; Constructing an energy forecasting model, inputting a multi-dimensional data set, capturing long-term dependencies in time series data, extracting key features, and outputting future energy demand forecast results; the future energy demand forecast includes short-term forecasts and long-term forecasts; The energy prediction model parameters are optimized using the back propagation algorithm and the gradient descent method through the multi-dimensional data set.

5. The intelligent energy management method based on multi-dimensional data analysis according to claim 1, characterized in that: The initial strategy includes: Construct a multi-objective optimization model and define an optimization objective function; the optimization objective function includes minimizing energy consumption and maximizing customer experience; set constraints, including equipment operation constraints, environmental comfort constraints, and total energy constraints; the equipment operation constraints include maximum power limits and minimum start-stop intervals; the environmental comfort constraints include comfort range thresholds for temperature, humidity, and illumination; Randomly generating a number of combinations of equipment operating parameters to construct an initial population; the equipment operating parameter combinations include adjustable variables; The system uses energy demand forecast results, equipment operating status, environmental parameters, and behavioral data as inputs and iterates. Each iteration includes a complete optimization loop, which includes calculating the objective function value of each individual in the current population, performing non-dominated sorting to stratify the solution set, maintaining the diversity distribution of solutions through congestion calculation, and generating a new generation of populations through selection, crossover, and mutation operations. When the Pareto front improvement rate falls below a set threshold for three consecutive generations or reaches the maximum number of iterations, the calculation is terminated and the Pareto optimal solution set containing non-dominated solutions is output. Analyze the non-dominated solutions in the Pareto optimal solution set and extract the device operating parameter combinations corresponding to the non-dominated solutions; establish a solution set mapping relationship table, wherein the solution set mapping relationship table associates each non-dominated solution with an executable device control instruction set to form a strategy library; The policy library includes device control parameters, execution conditions and expected effects.

6. The intelligent energy management method based on multi-dimensional data analysis according to claim 4, characterized in that: The automatic generation of a preliminary strategy includes: Constructing a decision model, wherein the decision factors of the decision model include current time period characteristics, real-time energy fluctuation rate, real-time energy equipment health status, passenger flow density change rate and user behavior pattern; By using the decision model, the comprehensive fitness of each non-dominated solution in the strategy library is evaluated, the non-dominated solution with the highest comprehensive fitness is selected, and the associated executable device control instruction set is obtained through the solution set mapping relationship table to generate a preliminary strategy; The preliminary strategy generates a hierarchical control instruction sequence including immediate execution instructions, delayed execution instructions and conditional trigger instructions; During the implementation of the strategy, the energy data, external data and behavioral data of the equipment are monitored in real time, and a dynamic adjustment trigger mechanism is set up. The trigger mechanism includes parameter deviation trigger based on threshold, immediate trigger in response to emergencies and periodic timed trigger; A multi-level fault tolerance mechanism is constructed, which includes primary fault tolerance, intermediate fault tolerance and advanced fault tolerance; the primary fault tolerance includes automatic retry of failure instructions, the intermediate fault tolerance includes switching to a backup strategy, and the advanced fault tolerance includes enabling a safe mode and issuing an alarm.

7. The intelligent energy management method based on multi-dimensional data analysis according to claim 1, characterized in that: The dynamic policy adjustment mechanism includes: Based on the current policy execution effect, when it is monitored that the actual energy consumption deviates from the predicted value by more than the deviation threshold, the environmental comfort index exceeds the comfort threshold range, or a specific user behavior pattern is identified, a dynamic policy adjustment mechanism is triggered. The dynamic policy adjustment mechanism includes locking the target device control parameters that cause deviations in the current policy and retaining the device control parameters of unaffected devices; extracting associated non-dominated solutions based on the policy library and performing local re-optimization; combining the current time period characteristics, real-time energy fluctuation rate, real-time energy device health status, passenger flow density change rate and user behavior pattern, through a decision model, screening the backup strategy with the highest comprehensive fitness from the Pareto optimal solution set, and converting the backup strategy into a hierarchical control instruction sequence; The distributed collaboration mechanism of edge nodes is used to coordinate control instructions for each region. When multiple policies act on the same device at the same time, instructions are integrated according to preset priorities.

8. The intelligent energy management method based on multi-dimensional data analysis according to claim 1, characterized in that: The distributed collaborative algorithm among edge nodes is used to achieve real-time distribution and execution of policies, including: After receiving the hierarchical control instruction sequence, each edge node performs instruction adaptability verification based on the device energy data and the external data, and determines the priority and weight distribution of instruction execution. A message bus based on the publish-subscribe model is established to exchange device status and instruction execution feedback through a lightweight communication protocol to achieve state synchronization between nodes. When a multi-node instruction conflict is detected, a timestamp-based eventual consistency algorithm is used to coordinate the execution order; for complex instructions involving multi-region linkage, the dominant node initiates a two-phase commit protocol to ensure atomic execution, and at the same time monitors the execution status of each node through a heartbeat mechanism; the real-time data updates generated during the execution process are fed back to the central processing unit through a backpropagation mechanism, forming a closed-loop control flow.

9. An intelligent energy management system based on multi-dimensional data analysis, characterized by: The system includes: data acquisition and processing module, intelligent analysis module, dynamic execution module, distributed collaboration module and interactive interface module; The data acquisition and processing module includes a multi-source data acquisition module and an edge pre-processing module; the multi-source data acquisition module is used to collect device energy data, external data, and behavioral data in real time; the edge pre-processing module is used to complete data cleaning and time sequence alignment at the edge node to complete data standardization; The intelligent analysis module includes an energy forecasting module, a strategy generation module, a strategy decision module, and a strategy library module; the energy forecasting module constructs an energy forecasting model to predict future energy demand, including short-term and long-term forecasts; the strategy generation module constructs a multi-objective optimization model to generate a Pareto optimal solution set that balances energy consumption and customer experience; the strategy decision module selects the optimal control strategy based on the data collected by the data acquisition and processing module; and the strategy library module is used to store and manage strategies; The dynamic execution module includes an instruction distribution module, a conflict coordination module and a fault tolerance control module; the instruction distribution module is used to dispatch hierarchical instructions to edge nodes; the conflict coordination module is used to resolve multi-node instruction conflicts and ensure execution consistency; the fault tolerance control module has a real-time three-level fault tolerance mechanism; The distributed collaboration module includes an edge collaboration module and a resource scheduling module; the edge collaboration module realizes state synchronization between nodes through a publish-subscribe mechanism; the resource scheduling module dynamically allocates edge computing resources; The interactive interface module includes a visualization module and a strategy configuration module; the visualization module displays the real-time energy consumption heat map and passenger flow distribution data of the equipment in each area; the strategy configuration module predicts the energy saving rate and comfort change based on the keywords selected by the user.

10. The intelligent energy management system based on multi-dimensional data analysis according to claim 9, characterized in that: When data anomalies are detected, a graded alarm mechanism is triggered; the alarm levels include emergency, warning, and prompt; the emergency is a red signal, the warning is a yellow signal, and the prompt is a blue signal; the maintenance record is automatically associated with the equipment, the health status is updated, and an alarm report is generated; the alarm report includes the alarm level, alarm time, alarm area or equipment, alarm maintenance record, and equipment health status.

Images