Big data-based intelligent energy optimization management method and system

Abstract

This invention relates to the field of smart energy management technology, specifically to a smart energy optimization management method and system based on big data. The method includes: acquiring a set of energy operation records containing energy consumption sequences, equipment operating status sequences, and environmental impact factor sequences; further decomposing the energy consumption sequences into multi-dimensional components to obtain basic energy consumption, flexible adjustable load, and energy consumption noise components; evaluating equipment operating conditions to generate an operating efficiency spectrum; and extracting environmental factors to obtain an environmental sensitivity vector. Multiple features are then input into a feature fusion center to synthesize a panoramic feature model of the energy system. Based on this model, a load forecasting and optimization decision engine generates a future load demand forecast curve and an optimization scheduling strategy, including equipment start-up and shutdown sequences, power adjustment commands, and demand response plans. This method achieves precise optimization scheduling of the energy system through refined feature decomposition and multi-feature fusion modeling.

Description

Technical Field

[0001] This invention relates to the field of smart energy management technology, and in particular to a smart energy optimization management method and system based on big data. Background Technology

[0002] Existing energy optimization management technologies typically collect operational data such as energy consumption, equipment operating status, and environmental impact factors. They use conventional statistical methods to monitor data and predict basic loads, and formulate simplified dispatching schemes based on overall energy consumption data. These technologies only perform basic data collection and shallow analysis of various operational data. They do not perform multi-dimensional decomposition of energy consumption sequences, quantitative assessment of equipment health and operational efficiency, or feature extraction of environmental impact factor sequences to form quantitative vectors. Various heterogeneous data are processed independently without correlation analysis or feature synthesis.

[0003] Existing technologies cannot precisely differentiate the composition of energy consumption, equipment operating efficiency lacks standardized quantitative representation, the impact of environmental factors on the energy system is difficult to quantify, multi-source operating data cannot be deeply integrated, and it is impossible to construct a comprehensive energy system characteristic model. Load forecasting results do not match actual operating conditions sufficiently, dispatching strategies can only achieve single power regulation, and cannot form a comprehensive dispatching scheme that includes equipment start-up and shutdown sequences, power regulation commands, and demand response plans, resulting in insufficient precision in energy system management.

[0004] It is necessary to complete the multi-dimensional decomposition of energy consumption sequence, generate equipment operating efficiency spectrum and environmental sensitivity vector, synthesize a panoramic feature model of energy system through feature fusion center, and generate accurate load forecast curve and multi-dimensional optimization scheduling strategy based on the model to improve the technical limitations of existing energy management. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a smart energy optimization management method and system based on big data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a smart energy optimization management method based on big data, comprising: Obtain a set of energy operation records, which includes energy consumption sequences, equipment operating status sequences, and environmental impact factor sequences; The energy consumption sequence is decomposed in multiple dimensions to obtain the basic energy consumption component, the flexible and adjustable load component, and the energy consumption noise component. The health and efficiency of the equipment operating condition sequence are evaluated to generate the equipment operating efficiency spectrum. The environmental impact factor sequence is feature extracted to obtain the environmental sensitivity vector. The basic energy consumption component, flexible adjustable load component, energy consumption noise component, equipment operating efficiency spectrum, and environmental sensitivity vector are input into the feature fusion center for correlation analysis and feature synthesis to generate a panoramic feature model of the energy system. Based on the aforementioned panoramic feature model of the energy system, a load forecasting and optimization decision engine generates load demand forecast curves and corresponding optimization scheduling strategies for future time periods. The optimization scheduling strategies include equipment start-up and shutdown sequences, power adjustment commands, and demand response plans.

[0007] As a further aspect of the present invention, the energy consumption sequence is decomposed into a multi-dimensional component to obtain a basic energy consumption component, a flexible and adjustable load component, and an energy consumption noise component, including: A trend decomposition algorithm is applied to the energy consumption sequence to extract a long-term trend line that changes gradually over a long period of time. The long-term trend line constitutes the basic energy consumption component. Clustering and pattern recognition are applied to the remaining sequence after removing the basic energy consumption component to identify load segments with regular start-stop or adjustable characteristics, and the load segments are aggregated to form the flexible adjustable load component. After further removing the influence of the basic energy consumption component and the flexible adjustable load component from the energy consumption sequence, the remaining residual part with drastic fluctuations and no obvious pattern is defined as the energy consumption noise component.

[0008] As a further aspect of the present invention, the health and efficiency of the equipment operating condition sequence are evaluated to generate an equipment operating efficiency spectrum, including: Key operating parameters are extracted from the equipment operating condition sequence, including operating temperature, working pressure, vibration amplitude, and output power. The key operating parameters are compared with the preset rated parameter range of the equipment under standard operating conditions, and the deviation index of each parameter is calculated. By combining the deviation indices of all key operating parameters, a health score representing the overall operational health status of the equipment is calculated through weighted fusion. The ratio of output power to corresponding input energy is analyzed over time to form an efficiency curve that reflects the dynamic changes in equipment energy efficiency. The health score and the efficiency curve are integrated in the time dimension to form the equipment operating efficiency spectrum.

[0009] As a further aspect of the present invention, the basic energy consumption component, flexible adjustable load component, energy consumption noise component, equipment operating efficiency spectrum, and environmental sensitivity vector are input into a feature fusion center for correlation analysis and feature synthesis to generate a panoramic feature model of the energy system, including: Within the feature fusion center, a correlation model is established between the basic energy consumption component and the environmental sensitivity vector to quantify the contribution of environmental impact to basic energy consumption. Establish a mapping relationship between the flexible adjustable load component and the equipment operating efficiency spectrum to characterize the adjustable potential of the flexible load under different equipment efficiency states; The energy consumption noise component is correlated with external random event logs to identify random factors that cause abnormal energy consumption. The correlation model, mapping relationship, and correlation analysis results are spatiotemporally aligned and feature-stitched with the original energy consumption sequence, equipment operating status sequence, and environmental impact factor sequence to construct a multi-dimensional, highly correlated data structure, namely the panoramic feature model of the energy system.

[0010] As a further aspect of the present invention, based on the aforementioned energy system panoramic feature model, a load forecasting curve and corresponding optimization scheduling strategy for future time periods are generated through a load forecasting and optimization decision engine, including: Using the panoramic feature model of the energy system as input, the prediction module in the load forecasting and optimization decision engine is driven. The prediction module combines historical load patterns, future environmental forecast data and planned events to deduce the load demand forecast curve for a specified future period. The load demand forecast curve, the current energy price signal, the power grid dispatch instructions, and the equipment operating efficiency spectrum and flexible adjustable load component information extracted from the energy system panoramic feature model are all input into the decision module of the load forecasting and optimization decision engine. Under the premise of satisfying all equipment operation constraints and grid requirements, the decision module performs global optimization with the objective function of minimizing energy cost or maximizing energy efficiency, and outputs the equipment start-up and shutdown sequence, power adjustment instructions and demand response plan containing specific operation times and quantities, which together constitute the optimized scheduling strategy.

[0011] As a further aspect of the present invention, the load demand forecast curve, the current energy price signal, the grid dispatch command, and the equipment operating efficiency spectrum and flexible adjustable load component information extracted from the energy system panoramic feature model are input together into the decision module of the load forecasting and optimization decision engine, including: The power grid dispatch instructions are analyzed to extract mandatory load reduction requirements, time-limited electricity price incentives, or renewable energy consumption indicators. The expected operating efficiency range of each device in the future time period is obtained from the device operating efficiency spectrum, and the adjustable capacity range and adjustment rate limit of each flexible load are obtained from the flexible adjustable load component information. An optimization time slot is established with time step as the unit. Within each optimization time slot, the value of the load demand forecast curve in the optimization time slot is taken as the basic load that must be met, the flexible and adjustable load component is taken as the optimizable variable, the energy price signal is taken as the cost coefficient, and the information in the power grid dispatch command and the equipment operating efficiency spectrum is transformed into the constraints of the optimization problem.

[0012] As a further aspect of the present invention, the establishment of optimized time slots in units of time steps, wherein within each optimized time slot, the value of the load demand forecast curve in the optimized time slot is used as the basic load that must be met, includes: Determine the time resolution for optimization decisions, and divide the future time period evenly into continuous optimization time slots according to the time resolution; Read the predicted load value corresponding to each optimized time slot from the load demand prediction curve, and directly set the predicted load value as the rigid load that must be met within the corresponding optimized time slot, which is denoted as the basic load that must be met in the optimized time slot. In the optimization model of each optimization time slot, it is ensured that the sum of the total power provided by all operating equipment in the optimization time slot and the energy storage charging and discharging power is equal to the sum of the basic load that the optimization time slot must satisfy and the actual value of the flexible adjustable load of the optimization time slot decision.

[0013] As a further aspect of the present invention, the decision-making module, under the premise of satisfying all equipment operating constraints and grid requirements, performs global optimization with the objective function of minimizing energy cost or maximizing energy efficiency, including: Construct a mathematical optimization model that includes decision variables, such as the start / stop status and operating power of each device in each optimization time slot, and the adjustment amount of each flexible load in each optimization time slot; The minimum start-up and shutdown time, maximum and minimum output, ramp rate limit, and efficiency curve obtained from the equipment's operating efficiency spectrum are converted into inequality constraints on the decision variables. The mandatory requirements in the power grid dispatching instructions are converted into equal or inequality constraints on the total load or the output of specific equipment. The objective function is set as the total energy cost, which consists of the cost of purchasing electricity from the grid in each time slot, the cost of equipment operation and maintenance, and the penalty cost; or the objective function is set as the reciprocal of the total energy efficiency of the system. The mathematical optimization model is solved using mixed integer programming or dynamic programming algorithms to obtain the globally optimal sequence of decision variable values.

[0014] As a further aspect of the present invention, it also includes: The optimized scheduling strategy is compiled into a sequence of control instructions that can be issued, and the sequence of control instructions is distributed to the corresponding energy system terminals for execution, specifically including: The equipment start-up and shutdown sequence, power regulation command and demand response plan in the optimized scheduling strategy are analyzed and decomposed into operation actions, action execution time points and action parameters for a single specific energy equipment or load unit. Each operation action, action execution time point, and action parameter is translated into a low-level control command that the target device or load unit can directly recognize and execute, based on the communication protocol and control interface specification used by the target device or load unit. According to the chronological order of the action execution time, all the underlying control instructions are sorted, and the execution conditions of the instructions with dependencies are marked. Finally, they are assembled into the control instruction sequence that can be issued with time sequence marks. The process of sorting all low-level control instructions according to the chronological order of their execution times and marking execution conditions for instructions with dependencies includes: The underlying control instructions are sorted in ascending order based on the action execution time point as the primary keyword. Examine all low-level control commands and identify logical dependencies, such as a device start command that can only be executed after another device shut down command has been executed, or a power adjustment command that can only be executed after a device start command has been confirmed. Add a pre-instruction identifier to the underlying control instructions that have dependencies. The pre-instruction identifier indicates that the underlying control instruction can only be issued after the pre-instruction it depends on has been confirmed to be executed successfully. This forms an instruction queue with timing and condition markers, and completes the final assembly of the issueable control instruction sequence.

[0015] As a further aspect of the present invention, the present invention also includes a smart energy optimization management system based on big data, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the smart energy optimization management method based on big data as described above.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: By decomposing the energy consumption sequence into multiple dimensions, we can identify the basic energy consumption component, the flexible and adjustable load component, and the energy consumption noise component. By conducting health and efficiency assessments on the equipment operating condition sequence, we can generate an equipment operating efficiency spectrum. By extracting features from the environmental impact factor sequence, we can obtain an environmental sensitivity vector. The different components of the energy consumption data are clearly distinguished. The operating health status and working efficiency of the equipment are presented in the form of a quantitative spectrum. The degree of impact of environmental factors on the energy system is transformed into a standardized vector. The interference components of the energy consumption data are effectively removed. The quantitative representation of the equipment operating status is more in line with the actual operating status. The quantitative results of environmental impact can be directly adapted to the analysis logic of the energy system.

[0017] By inputting the basic energy consumption component, flexible and adjustable load component, energy consumption noise component, equipment operating efficiency spectrum, and environmental sensitivity vector into the feature fusion center for correlation analysis and feature synthesis, a panoramic feature model of the energy system can be generated. Based on this model, the load forecasting and optimization decision engine can output load demand forecast curves and optimized scheduling strategies including equipment start-up and shutdown sequences, power adjustment commands, and demand response plans. Multiple heterogeneous features are deeply coupled and correlated, and the operating characteristics of the energy system are comprehensively integrated and presented. The scheduling commands cover multiple dimensions of equipment control and demand response. The scheduling content is highly matched with the actual energy consumption composition, equipment status, and environmental impact of the energy system. The logic of the scheduling strategy formulation is consistent with the panoramic operating characteristics of the energy system. Attached Figure Description

[0018] Figure 1 The flowchart shows the smart energy optimization management method based on big data as described in this invention. Figure 2 A flowchart generated for the equipment operating efficiency spectrum; Figure 3 This is a 24-hour basic load forecast curve. Figure 4 A 24-hour power dispatch curve for multi-energy devices; Figure 5 This is a graph showing the success rate of executing control commands for smart energy system equipment. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1 The system acquires a set of energy operation records, including energy consumption sequences recording historical energy usage, equipment operating condition sequences reflecting equipment status, and environmental impact factor sequences recording external factors such as temperature and humidity. These three sequences undergo deep processing. The energy consumption sequence is decomposed into basic energy consumption, flexible adjustable load, and energy noise components. Simultaneously, the equipment operating condition sequence is assessed for health and efficiency, generating an equipment operating efficiency spectrum that comprehensively reflects equipment health and energy efficiency changes. Features are extracted from the environmental impact factor sequence to obtain an environmental sensitivity vector. All components obtained from the above decomposition and assessment, including basic energy consumption, flexible adjustable load, energy noise, equipment operating efficiency spectrum, and environmental sensitivity vector, are input into a feature fusion center for cross-dimensional correlation analysis and deep feature synthesis. This constructs a comprehensive panoramic feature model of the energy system, capable of depicting the overall state and internal relationships of the energy system. Based on the generated panoramic feature model of the energy system, it is input into the load forecasting and optimization decision engine. This engine can deduce the load demand forecast curve that the system may face in a specific future period, and simultaneously generate an optimized scheduling strategy designed to achieve economical and efficient operation. This strategy specifically includes equipment start-up and shutdown sequence, power regulation command and demand response plan.

[0022] In one embodiment of the present invention, see [reference] Figure 2The process of multi-dimensionally decomposing the energy consumption sequence to obtain the basic energy consumption component, the flexible adjustable load component, and the energy consumption noise component specifically involves applying a trend decomposition algorithm to the energy consumption sequence. This algorithm can extract a long-term, gently changing long-term trend line, which constitutes the basic energy consumption component. After removing the basic energy consumption component, clustering and pattern recognition techniques are applied to the remaining sequence to identify load segments with regular start-stop patterns or exhibiting adjustable characteristics. These identified load segments are then aggregated to form the flexible adjustable load component. Furthermore, the influence of the separated basic energy consumption component and flexible adjustable load component is removed from the original energy consumption sequence. The remaining residual portion, characterized by drastic fluctuations and no obvious pattern, is defined as the energy consumption noise component.

[0023] The process of assessing the health and efficiency of equipment operating condition sequences and generating an equipment operating efficiency spectrum includes extracting key operating parameters from the equipment operating condition sequences. These parameters typically cover operating temperature, working pressure, vibration amplitude, and output power. Each extracted key operating parameter is compared one by one with a preset range of rated parameters for the equipment under standard operating conditions, and the deviation index for each parameter is calculated. Then, by combining the deviation indices of all key operating parameters, a single health score characterizing the overall operating health of the equipment is calculated using a preset weighted fusion method. Simultaneously, the change in the ratio of output power to corresponding input energy over time is analyzed, forming an efficiency curve reflecting the dynamic changes in equipment energy efficiency. Finally, the calculated health score and efficiency curve are integrated over time to form the equipment operating efficiency spectrum.

[0024] In practical implementation, taking the energy management system of a commercial building as an example, the energy operation record set includes hourly energy consumption sequences, equipment operating status sequences, and environmental impact factor sequences from the past year. The energy consumption sequence records the building's total electricity consumption data, the equipment operating status sequence includes operating parameters of chillers, air conditioning fans, and lighting systems, and the environmental impact factor sequence includes outdoor temperature and humidity data. The energy consumption sequence is decomposed in multiple dimensions to obtain a basic energy consumption component, a flexible adjustable load component, and an energy consumption noise component. In practice, a trend decomposition algorithm is applied to the energy consumption sequence, using a moving average method to extract a long-term, gently changing long-term trend line. This long-term trend line constitutes the basic energy consumption component. In the commercial building example, the basic energy consumption component reflects the building's basic electricity needs, such as the energy consumption of lighting and basic equipment. Data shows that the basic energy consumption component fluctuates little between days, presenting a stable baseline. Clustering and pattern recognition were applied to the remaining sequence after removing the baseline energy consumption component. K-means clustering was used to identify load segments with regular start-stop or adjustable characteristics. These load segments were aggregated to form a flexible adjustable load component. In the example, the flexible adjustable load component corresponds to the start-stop of the air conditioning system and the operation of the elevator. Data shows that these loads change frequently during working hours, exhibiting obvious adjustable characteristics. Further removal of the baseline energy consumption component and the flexible adjustable load component from the energy consumption sequence resulted in a residual portion with drastic fluctuations and no clear pattern, defined as the energy consumption noise component. This noise component may be caused by random events such as temporary equipment activation or personnel activity. Data comparison shows that the decomposed baseline energy consumption component is smooth, the flexible adjustable load component is regular, while the energy consumption noise component has no specific pattern.

[0025] In some embodiments, a health and efficiency assessment is performed on the equipment operating condition sequence to generate an equipment operating efficiency spectrum. Key operating parameters are extracted from the equipment operating condition sequence, including operating temperature, operating pressure, vibration amplitude, and output power. In the commercial building example, for a chiller unit, operating temperature, operating pressure, vibration amplitude, and output power are extracted. The key operating parameters are compared with the preset rated parameter range of the equipment under standard operating conditions, and the deviation index of each parameter is calculated. The deviation index is calculated using the following formula:

[0026] in: This represents the deviation index of the i-th parameter. This represents the actual measured value. This represents the rated value under standard operating conditions. A health score, characterizing the overall operational health status of the equipment, is calculated by weighted fusion of deviations from all key operating parameters. The formula for calculating the health score is:

[0027] in: Indicates health score, This represents the weight of the i-th parameter. This represents the deviation index, with the sum of all weights equal to 1. Analyzing the change in the ratio of output power to corresponding input energy over time forms an efficiency curve reflecting the dynamic changes in equipment energy efficiency. In the example, the energy efficiency ratio of the chiller unit changes over time, and the efficiency curve shows a decrease in the energy efficiency ratio during high-temperature periods. Integrating the health score and efficiency curve over time forms the equipment operating efficiency spectrum. This spectrum is a time-series data structure, with each time point containing a health score and energy efficiency ratio value, used for a comprehensive assessment of equipment status.

[0028] Optionally, the trend decomposition algorithm can also use seasonal decomposition methods to better handle periodic energy consumption sequences. It is understandable that clustering and pattern recognition algorithms can select different numbers of clusters based on load characteristics to adapt to different application scenarios. In terms of data comparison, the original energy consumption sequence shows complex fluctuations. After multi-dimensional decomposition, the basic energy consumption component reveals long-term trends, the flexible and adjustable load component highlights the adjustable portion, the energy consumption noise component captures random fluctuations, and for the equipment operating condition sequence, the health score provides a quantitative health indicator, while the efficiency curve shows energy efficiency changes. The integrated equipment operating efficiency spectrum provides detailed basis for optimized scheduling.

[0029] In one embodiment of the present invention, the basic energy consumption component, the flexible adjustable load component, the energy consumption noise component, the equipment operating efficiency spectrum, and the environmental sensitivity vector are input into a feature fusion center for correlation analysis and feature synthesis to generate a panoramic feature model of the energy system. The process involves establishing a quantitative correlation model between the basic energy consumption component and the environmental sensitivity vector within the feature fusion center to quantify the contribution of different environmental factors to basic energy consumption. Simultaneously, a mapping relationship is established between the flexible adjustable load component and the equipment operating efficiency spectrum to depict the adjustable potential of the flexible load under different equipment efficiency states. Furthermore, the energy consumption noise component is correlated with recorded external random event logs to identify random factors causing abnormal energy consumption fluctuations. Finally, the correlation model, mapping relationship, and correlation analysis results obtained in the above steps are spatiotemporally aligned and feature-stitched with the original energy consumption sequence, equipment operating state sequence, and environmental impact factor sequence to construct a multi-dimensional, highly correlated data structure, which is the panoramic feature model of the energy system.

[0030] Based on a panoramic feature model of the energy system, a load forecasting and optimization decision engine generates load demand forecast curves and corresponding optimization scheduling strategies for future periods. Specifically, the panoramic feature model of the energy system serves as the core input, driving the forecasting module within the load forecasting and optimization decision engine. This module combines historical load patterns, future environmental forecast data, and known planned events to deduce the load demand forecast curve for a specified future period. Subsequently, the obtained load demand forecast curve, current energy price signals, received grid dispatch instructions, and equipment operating efficiency spectra and flexible adjustable load component information extracted from the panoramic feature model of the energy system are input into the decision module of the load forecasting and optimization decision engine. Under the premise of satisfying all equipment operating constraints and grid requirements, the decision module performs a global optimization solution with the objective function of minimizing energy costs or maximizing system energy efficiency. The output includes equipment start-up and shutdown sequences, power regulation instructions, and demand response plans containing specific operation times and quantities; these together constitute the optimized scheduling strategy.

[0031] In practical implementation, the components obtained from decomposition and evaluation are input into a feature fusion center for synthesis. Based on the results, predictions and strategies are generated. Taking an integrated energy system in an industrial park that includes photovoltaic, energy storage, refrigeration units, air compressors, and production line loads as an example, the basic energy consumption component, flexible adjustable load component, energy consumption noise component, equipment operating efficiency spectrum, and environmental sensitivity vector are all used as inputs to the feature fusion center. Within the feature fusion center, a correlation model is established between the basic energy consumption component and the environmental sensitivity vector. The environmental sensitivity vector includes features such as outdoor temperature and solar irradiance. The correlation model quantifies the contribution of environmental impacts to basic energy consumption. In the industrial park example, the correlation model reveals that the basic energy consumption component increases significantly during the high-temperature period in summer and is strongly positively correlated with the temperature sensitivity vector. A mapping relationship was established between flexible adjustable load components and equipment operating efficiency spectra. Flexible adjustable load components include adjustable air conditioning load and production buffer load. The equipment operating efficiency spectrum reflects the real-time energy efficiency of refrigeration units and air compressors. The mapping relationship characterizes the adjustability potential of flexible loads under different equipment efficiency states. For example, when the operating efficiency spectrum of a refrigeration unit shows high efficiency, the adjustability potential of the associated air conditioning flexible load increases. Correlation analysis was performed between energy consumption noise components and external random event logs. These logs record information such as temporary equipment maintenance and rush production tasks. The correlation analysis identified random factors leading to abnormal energy consumption. Data comparison showed that the peak values ​​of energy consumption noise components often coincided temporally with the rush production events recorded in the logs. The correlation model, mapping relationship, and correlation analysis results are spatiotemporally aligned and feature-stitched with the original energy consumption sequence, equipment operating status sequence, and environmental impact factor sequence to construct a multi-dimensional, highly correlated data structure. This structure is the panoramic feature model of the energy system. In the example, the panoramic feature model of the energy system is a data matrix containing multi-dimensional features such as timestamps, base load, adjustable potential, equipment health, and environmental coefficients.

[0032] In some embodiments, based on the energy system panoramic feature model, a load forecasting and optimization decision engine generates load demand forecast curves and corresponding optimization scheduling strategies for future periods. Using the energy system panoramic feature model as input, the engine drives a forecasting module. This module combines historical load patterns, future environmental forecast data, and planned events to deduce the load demand forecast curve for a specified future period. For industrial parks, planned events include known production plans, and future environmental forecast data comes from weather forecasts. The forecasting module outputs load demand forecast curves for the next 24 hours at 15-minute intervals. The load demand forecast curves, current energy price signals, grid dispatch instructions, and equipment operating efficiency spectra and flexible adjustable load component information extracted from the energy system panoramic feature model are input into the decision module of the load forecasting and optimization decision engine. The energy price signal is a time-of-use price, and the grid dispatch instructions include peak-hour load reduction requirements for the following day. Under the premise of satisfying all equipment operation constraints and grid requirements, the decision-making module performs global optimization with the objective function of minimizing energy cost or maximizing energy efficiency. The output includes equipment start-up and shutdown sequence, power regulation command and demand response plan containing specific operation time and operation quantity. The equipment start-up and shutdown sequence specifies the charging and discharging time of the energy storage system and the start-up and shutdown plan of the refrigeration unit. The power regulation command sets the operating power point of the air compressor. The demand response plan describes the load shifting scheme of the production line. These together constitute the optimized scheduling strategy.

[0033] Optionally, the contribution of the correlation model can be quantified using a multiple linear regression method, taking multiple factors in the environmental sensitivity vector as independent variables and the basic energy consumption component as the dependent variable, to fit the contribution coefficients of each environmental factor. It can be understood that the output of the feature fusion center is the panoramic feature model of the energy system, which can serve as a unified data interface for the load forecasting and optimization decision engine and other advanced application modules. In terms of data comparison, the original energy consumption sequence, equipment operating status sequence, and environmental impact factor sequence are independent time-series data. However, the panoramic feature model of the energy system generated after processing by the feature fusion center is a structured data object that is strictly aligned in the time dimension and deeply integrated in the feature dimension. Its dimensionality is significantly higher than any original input sequence, and it includes the correlation between input sequences and derived features. For the generation of optimization scheduling strategies, the data comparison is reflected in the fact that the input is macroscopic or external instructions such as prediction curves and price signals, while the output is a specific, time-sequential, and executable set of control actions, such as "start charging of energy storage unit 1 at 14:00, with a power of 500kW." The decision module completes the transformation from objectives and constraints to specific action plans.

[0034] In one embodiment of the present invention, the process of inputting load demand forecast curves, energy price signals, grid dispatch instructions, equipment operating efficiency spectra, and flexible adjustable load component information into a decision module includes parsing the grid dispatch instructions and extracting mandatory load reduction requirements, time-limited electricity price incentive information, or renewable energy consumption indicators. The expected operating efficiency range of each piece of equipment in the future time period is obtained from the equipment operating efficiency spectrum, and the adjustable capacity range and adjustment rate limit of each flexible load are obtained from the flexible adjustable load component information. Optimization time slots are established in units of time steps. Within each optimization time slot, the value of the load demand forecast curve in that optimization time slot is taken as the basic load that must be satisfied, the flexible adjustable load component is taken as an optimizable variable, and the energy price signal is taken as a cost coefficient. The information in the grid dispatch instructions and equipment operating efficiency spectra is transformed into constraints for the optimization problem.

[0035] The process involves establishing optimization time slots based on time steps. Within each optimization time slot, the value of the load demand forecast curve within that time slot is used as the mandatory basic load. Specifically, the time resolution for optimization decisions is determined, and the future optimization period is evenly divided into consecutive optimization time slots based on this time resolution. The predicted load value corresponding to each optimization time slot is read from the load demand forecast curve and directly set as the rigid load that must be met within the corresponding optimization time slot, denoted as the mandatory basic load for that optimization time slot. In the optimization model for each optimization time slot, it is ensured that the sum of the total power provided by all operating equipment and the charging / discharging power of the energy storage system within that optimization time slot is equal to the sum of the mandatory basic load for that optimization time slot and the actual value of the flexible adjustable load determined by the optimization time slot decision.

[0036] In practical implementation, taking an industrial park microgrid system that includes photovoltaic power generation, battery energy storage, adjustable industrial load, and essential production load as an example, the load demand forecast curve is the predicted value for the next 24 hours at 15-minute intervals, the energy price signal is the known time-of-use electricity price, and the grid dispatch instructions are issued in document form. The grid dispatch instructions are parsed to extract mandatory load reduction requirements, time-based electricity price incentive information, or renewable energy consumption indicators. In the example, parsing the instructions yields a mandatory load reduction requirement of "at least 200 kilowatts of total load must be reduced between 13:00 and 15:00" and time-based electricity price incentive information of "photovoltaic power generation enjoys additional subsidies between 11:00 and 13:00". The expected operating efficiency range of each device in the future period is obtained from the equipment operating efficiency spectrum. The equipment operating efficiency spectrum shows that the battery energy storage in the current healthy state has a charge and discharge efficiency range of 92% to 95%. The adjustable capacity range and adjustment rate limit of each flexible load are obtained from the flexible adjustable load component information. The flexible adjustable load component information shows that the adjustable capacity range of the electric furnace load is 0 to 500 kW, and the adjustment rate limit is no more than 100 kW per minute. An optimization time slot is established with time step as the unit. Within each optimization time slot, the value of the load demand forecast curve in the optimization time slot is taken as the basic load that must be met, the flexible and adjustable load component is taken as the optimizable variable, and the energy price signal is taken as the cost coefficient. The information in the grid dispatch instructions and equipment operating efficiency spectrum is transformed into the constraints of the optimization problem. In the example, the time step is set to 15 minutes, and the next 24 hours are divided into 96 consecutive optimization time slots. Within each optimization time slot, the forecast value read from the load demand forecast curve is set as the basic load that must be met, the power value of the electric furnace load is set as the optimizable variable, the time-of-use electricity price is taken as the electricity purchase cost coefficient, the mandatory load reduction requirement is transformed into an inequality constraint on the total load, and the equipment efficiency range is transformed into an equality constraint on the battery charging and discharging efficiency.

[0037] In some embodiments, the optimization time slots are established in units of time steps. Within each optimization time slot, the value of the load demand forecast curve in the optimization time slot is used as the basic load that must be met. The time resolution for optimization decisions is determined, and the time resolution is set to 15 minutes. Based on the 15-minute time resolution, the next 24 hours are evenly divided into 96 consecutive optimization time slots. The predicted load value corresponding to each optimization time slot is read from the load demand forecast curve, and the predicted load value is directly set as the rigid load that must be met within the corresponding optimization time slot, denoted as the basic load that must be met for the optimization time slot. In the example, the predicted load value corresponding to the optimization time slot at 2 PM (i.e., the 56th time slot) is 1500 kW, and this 1500 kW is set as the basic load that must be met for that optimization time slot. In the optimization model of each optimization time slot, it is ensured that the sum of the total power provided by all operating equipment in the optimization time slot and the energy storage charging and discharging power is equal to the sum of the basic load that must be met for the optimization time slot and the actual value of the flexible adjustable load of the optimization time slot decision. This is achieved through the power balance equation, the formula is:

[0038] in: It represents the sum of the output of all power generation equipment and the discharge power of energy storage during the optimized time slot t (it is negative when energy storage is charging). This represents the basic load that must be satisfied for the optimized time slot t. This represents the actual value of the flexible adjustable load after optimizing the time slot t decision.

[0039] Optionally, the time resolution can be set to different values ​​such as 5 minutes, 10 minutes, or 30 minutes, depending on the required level of precision in system adjustment. It can be understood that the baseline load that must be met is a fixed numerical parameter in each optimization time slot, derived from load forecast results, and remains constant during a single optimization solution. Regarding data comparison, the grid dispatch instructions input to the decision module are in text or code form, which are parsed and transformed into specific numerical constraints, such as "reduce at least 200 kW" being transformed into "total load ≤ predicted total load - 200 kW". The flexible adjustable load component information, when input, is a set of parameters describing adjustable capacity and rate; in the optimization time slot model, it is transformed into decision variables. And its upper and lower bound constraints. The load demand forecast curve is a continuous forecast curve, which is discretized into a specific numerical point for each optimal time slot when establishing the optimal time slot. This discretization process transforms the continuous prediction problem into a mathematical optimization problem that can be solved independently or in conjunction with each time slot.

[0040] See Figure 3This is a 24-hour basic load forecast curve, fully presenting the rigid basic load variation pattern of the industrial park's microgrid over 24 hours. From 0:00 to 6:00, the load in time slots 0-24 remains between 700 and 900 kW, representing the period of lowest energy consumption and corresponding to non-production / low load conditions. From 6:00 to 12:00, the load in time slots 24-48 gradually increases from 900 kW to 1400 kW, reflecting the gradual recovery of production activities. From 12:00 to 18:00, the load in time slots 48-72 reaches its peak, representing the period of most intense energy consumption. From 18:00 to 24:00, the load in time slots 72-96 gradually decreases from 1400 kW to 900 kW, indicating the gradual end of production activities. The light red shaded area in the graph clearly marks the mandatory load reduction period (13:00-15:00, time slots 52-60), which is the core constraint after the grid dispatch command is converted, requiring a minimum reduction of 200 kW in total load during this period. This period falls precisely within the peak load range of the day, reflecting the dispatching instructions' intent to regulate peak load.

[0041] In one embodiment of the present invention, the decision-making module, under the premise of satisfying all equipment operation constraints and grid requirements, performs global optimization with the objective function of minimizing energy cost or maximizing energy efficiency. The process involves constructing a mathematical optimization model containing decision variables, including the start-up / shutdown state and operating power of each device in each optimization time slot, and the adjustment amount of each flexible load in each optimization time slot. The minimum start-up / shutdown time, maximum / minimum output limits, ramp rate limits, and efficiency curves obtained from the equipment operating efficiency spectrum are transformed into inequality constraints on the decision variables. Mandatory requirements in grid dispatch instructions are converted into equality or inequality constraints on the total system load or the output of specific equipment. The objective function is set as the total system energy cost, which consists of the cost of purchasing electricity from the grid in each time slot, equipment operation and maintenance costs, and potential penalty costs; alternatively, the objective function can be set as the reciprocal of the total system energy efficiency. Mixed-integer programming or dynamic programming algorithms are used to solve the constructed mathematical optimization model to obtain the globally optimal sequence of decision variable values.

[0042] In practical implementation, taking a factory energy system including a gas turbine, an electric chiller, a battery energy storage system, and necessary process loads as an example, the inputs received by the decision module include the load demand forecast curve for the next 24 hours with a time resolution of 15 minutes, time-of-use energy price signals, grid dispatch instructions including peak-hour demand limits, and equipment operating efficiency spectrum and flexible adjustable load component information extracted from the energy system panoramic feature model. A mathematical optimization model containing decision variables is constructed. The decision variables include the start-stop state and operating power of each device in each optimization time slot, and the adjustment amount of each flexible load in each optimization time slot. For the factory energy system in the example, the decision variables specifically include the start-stop state (0 / 1 variable) and power generation (continuous variable) of the gas turbine in each of the 96 time slots, the start-stop state and cooling power of the electric chiller, the charging and discharging power of the battery energy storage system (signed variable), and the power adjustment amount of the flexible load of the process production line. Refer to Table 1 for the range of its decision variables.

[0043] Table 1: Equipment Decision Variables and Basic Parameters

[0044] The minimum start-up and shutdown time, maximum and minimum output, ramp rate limits, and efficiency curves obtained from the equipment's operating efficiency spectrum are converted into inequality constraints on decision variables. The minimum start-up and shutdown time constraint requires that the gas turbine must run continuously for at least 4 time slots (1 hour) once started and must remain shut down for at least 2 time slots (0.5 hours) once shut down. This is achieved by introducing auxiliary variables and constraint inequalities. The maximum and minimum output constraints are directly reflected in the lower and upper limits of the decision variables in the table above, such as the gas turbine's power generation capacity. satisfy The ramp rate limit requires that the power variation of the gas turbine in adjacent time slots not exceed 200 kilowatts per minute, i.e. The equipment operating efficiency spectrum provides the energy efficiency ratio curves of the electric chiller under different load rates. These curves are transformed into a nonlinear relationship between its output cooling power and input electrical power, and are added as constraints to the optimization model. The mandatory requirements in the power grid dispatch instructions are converted into equality or inequality constraints on the total load or the output of specific equipment. The power grid dispatch instructions require the factory's total power consumption between 13:00 and 15:00 (corresponding to time slots t=52 to t=60) not to exceed a certain limit. This translates into a constraint on time slot t: the power purchased from the grid ,in It is one of the decision variables.

[0045] In some embodiments, the objective function is set as total energy cost, which consists of the cost of purchasing electricity from the grid for each time slot, equipment operation and maintenance costs, and penalty costs. The cost of purchasing electricity from the grid is calculated based on time-of-use pricing and the amount of electricity purchased. The equipment operation and maintenance costs are related to the power generation and start-up / shutdown frequency of the gas turbine. The penalty costs are set for violations of soft constraints such as indoor temperature comfort. The specific mathematical expression of the objective function is to minimize:

[0046] in: Represents the total cost. This represents the electricity price in time slot t. This represents the amount of electricity purchased from the grid in time slot t. This represents the operation and maintenance cost coefficient per unit of electricity generated by a gas turbine. This represents the power generation of the gas turbine in time slot t. This indicates the cost of starting a gas turbine once. A 0 / 1 variable characterizing whether time slot t has started. It is a very large positive number used as the penalty coefficient. These are slack variables that violate soft constraints. Alternatively, the objective function can be set as the reciprocal of the system's total energy efficiency, defined as the ratio of total output energy to total input energy; minimizing its reciprocal is equivalent to maximizing total energy efficiency. Mixed-integer programming or dynamic programming algorithms are used to solve the mathematical optimization model, obtaining the globally optimal sequence of decision variable values. In the example, since the model includes 0 / 1 integer variables such as the start-stop states of the gas turbine and electric chiller, a mixed-integer linear programming algorithm is used. The solver outputs the optimal decision variable values ​​for each of the next 96 time slots, including the start-stop state sequences of the gas turbine and electric chiller, their respective power sequences, the charging and discharging power sequences of the battery energy storage system, and the adjustment sequence of the flexible load.

[0047] See Figure 4This is a 24-hour power dispatch curve for multi-energy equipment, fully demonstrating the power dispatch strategies and constraints of three core equipment types: gas turbines, electric chillers, and battery energy storage. Gas turbines and electric chillers start synchronously during the day (8-20 hours) to match peak daytime production loads; they shut down at night to reduce maintenance costs. Battery energy storage charges during off-peak electricity price periods at night (0-10 hours) and discharges during peak daytime electricity price periods (10-24 hours), achieving peak shaving and valley filling combined with electricity price arbitrage. The power of all equipment remains within preset upper and lower limits, meeting equipment operation safety and grid dispatch requirements. Battery discharge power is strictly controlled within the range of -500kW to 300kW to avoid overcharging and over-discharging, which could damage battery life. The gas turbine, as the base load power source, provides stable base power; the electric chiller, as an adjustable load, matches cooling demand; and battery energy storage, as a buffer unit, smooths power fluctuations. The three work together to meet the total power demand of both base and flexible loads, verifying the feasibility of the power balance equation.

[0048] In one embodiment of the present invention, the optimized scheduling strategy is compiled into a sequence of control instructions that can be issued, and the sequence of control instructions is distributed to the corresponding energy system terminals for execution. Specifically, this includes parsing the equipment start-up and shutdown sequence, power regulation instructions, and demand response plan in the optimized scheduling strategy, and decomposing them into operational actions, action execution times, and action parameters for individual specific energy devices or load units. Each operational action, action execution time, and action parameter is translated into low-level control instructions that the target device or load unit can directly recognize and execute, based on the communication protocol and control interface specifications adopted by the target device or load unit. All generated low-level control instructions are sorted according to the chronological order of action execution times, and execution conditions are marked for instructions with logical dependencies. Finally, they are assembled into a sequence of control instructions with timing markers that can be issued.

[0049] The process involves sorting all low-level control commands according to their execution time sequence and marking execution conditions for commands with dependencies. Specifically, all low-level control commands are sorted in ascending order based on their execution time. All low-level control commands are examined to identify logical dependencies. These dependencies include conditions such as a device start command requiring another device stop command to be executed before execution, or a power adjustment command requiring confirmation of a successful device start command before execution. A pre-command identifier is added to low-level control commands with dependencies. This identifier indicates that the low-level control command can only be issued after its dependent pre-command has been confirmed to have been successfully executed. This forms a command queue with timing and condition markers, completing the final assembly of the sequence of control commands that can be issued.

[0050] In practical implementation, the optimized scheduling strategy is generated by the decision module, which includes equipment start-up and shutdown sequences, power adjustment commands, and demand response plans. The optimized scheduling strategy's equipment start-up and shutdown sequences, power adjustment commands, and demand response plans are analyzed and decomposed into operational actions, execution times, and parameters for individual specific energy devices or load units. In the building energy system example, one optimized scheduling strategy is "Start the central air conditioning unit at 10:30, setting the cooling power to 300kW." After analysis, the decomposed operational action is "Start," the execution time is "10:30," and the parameter is "cooling power setpoint 300kW." Another demand response plan, "Reduce the lighting circuit power to 70% from 14:00 to 15:00," is decomposed into the operational action "Adjust power" for the lighting circuit controller, the execution time is "14:00," and the parameter is "target power percentage 70%." Each operation action, execution time point, and action parameter is translated into a low-level control instruction that the target device or load unit can directly recognize and execute, based on the communication protocol and control interface specification used by the target device or load unit. For a central air conditioning host using the Modbus TCP protocol, the operation action "start" is translated into an instruction frame that writes a specific value to a specific register address, and the action parameter "300kW" is quantized into the corresponding register value and filled into the data field of the instruction frame. For a lighting circuit controller using the KNX protocol, the operation action "adjust power" is translated into a message written to a specific group address, and the action parameter "70%" is converted into the corresponding control byte value. The action execution time point information is retained for scheduling the timing of instruction issuance.

[0051] In some embodiments, all underlying control instructions are sorted according to the chronological order of their execution times, and execution conditions are marked for instructions with dependencies. Finally, these are assembled into a sequence of control instructions with time-series markers that can be issued. In the example, the multiple underlying control instructions parsed and translated from the optimized scheduling strategy have execution times of 10:30, 14:00, 10:28, etc., and the sorted instruction sequence is in the order of the 10:28 instruction, the 10:30 instruction, and the 14:00 instruction. Marking execution conditions for instructions with dependencies involves sorting the underlying control instructions in ascending order using the execution time as the primary keyword, checking all underlying control instructions, and identifying logical dependencies. In the building example, dependencies include the central air conditioning unit's chilled pump start instruction only being executed after the unit's start instruction is completed, or the battery pack's discharge power adjustment instruction only being executed after confirmation feedback is received regarding the successful closure of the battery pack's grid connection switch. A pre-command identifier is added to the underlying control commands with dependencies. The pre-command identifier indicates that the underlying control command can only be issued after the pre-command it depends on has been confirmed to be executed successfully. This forms a command queue with timing and condition markers, completing the final assembly of the control command sequence that can be issued. For the start command of the chilled pump, its pre-command identifier is set as the unique number of the start command of the central air conditioning unit. The start command of the central air conditioning unit is set as the unique number of the battery pack ready command.

[0052] Optionally, the translation of low-level control instructions can support multiple industrial protocols, adapted through a protocol plugin library. It can be understood that the instruction queue structure with timing and condition markers allows the instruction issuing and execution unit to execute instructions sequentially according to strict timelines and logical conditions, avoiding conflicts that may arise from concurrent control. Regarding data comparison, the optimized scheduling strategy is a relatively high-level instruction description oriented towards system scheduling logic, such as "reduce non-critical loads during peak electricity price periods," while the compiled and issued control instruction sequence is code or messages that are immediately executable, oriented towards specific device controller hardware, such as "write the value 0 to register 40001 of the PLC with IP address 192.168.1.10 at 14:00:00." In the instruction sequence assembly stage, the instruction set before sorting is only loosely organized according to the source device or policy entry. The instruction queue formed after sorting and labeling dependencies is a task list with a clear global execution order and logical constraints, reliably executable by the scheduling engine.

[0053] See Figure 5This is a chart showing the success rate of control command execution in a smart energy system. It visually presents the success rate of five types of core equipment in executing optimized scheduling commands, with 90% as the success threshold for compliance verification. All equipment has a success rate higher than the 90% threshold, indicating that the overall execution effect of optimized scheduling commands at the terminal equipment level meets the standards, and the system control reliability is strong. The success rates of all five types of equipment are consistently above 92%, with central air conditioning (98%) and water pump systems (97%) performing best, reflecting the control stability of core loads and power equipment. The fresh air system (92%) has the lowest value, but still meets the threshold requirement, possibly affected by environmental variables or equipment aging. This is a closed-loop verification from optimized decision-making to actual energy consumption, directly reflecting the efficiency of control command sequence issuance, equipment response capability, and communication reliability.

[0054] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A smart energy optimization management method based on big data, characterized in that, The method includes: Obtain a set of energy operation records, which includes energy consumption sequences, equipment operating status sequences, and environmental impact factor sequences; The energy consumption sequence is decomposed in multiple dimensions to obtain the basic energy consumption component, the flexible and adjustable load component, and the energy consumption noise component. The health and efficiency of the equipment operating condition sequence are evaluated to generate the equipment operating efficiency spectrum. The environmental impact factor sequence is feature extracted to obtain the environmental sensitivity vector. The basic energy consumption component, flexible adjustable load component, energy consumption noise component, equipment operating efficiency spectrum, and environmental sensitivity vector are input into the feature fusion center for correlation analysis and feature synthesis to generate a panoramic feature model of the energy system. Based on the aforementioned panoramic feature model of the energy system, a load forecasting and optimization decision engine generates load demand forecast curves and corresponding optimization scheduling strategies for future time periods. The optimization scheduling strategies include equipment start-up and shutdown sequences, power adjustment commands, and demand response plans.

2. The smart energy optimization management method based on big data according to claim 1, characterized in that, The energy consumption sequence is decomposed into three components: basic energy consumption component, flexible adjustable load component, and energy consumption noise component, including: A trend decomposition algorithm is applied to the energy consumption sequence to extract a long-term trend line that changes gradually over a long period of time. The long-term trend line constitutes the basic energy consumption component. Clustering and pattern recognition are applied to the remaining sequence after removing the basic energy consumption component to identify load segments with regular start-stop or adjustable characteristics, and the load segments are aggregated to form the flexible adjustable load component. After further removing the influence of the basic energy consumption component and the flexible adjustable load component from the energy consumption sequence, the remaining residual part with drastic fluctuations and no obvious pattern is defined as the energy consumption noise component.

3. The smart energy optimization management method based on big data according to claim 2, characterized in that, The health and efficiency of the equipment operating condition sequence are evaluated to generate an equipment operating efficiency spectrum, including: Key operating parameters are extracted from the equipment operating condition sequence, including operating temperature, working pressure, vibration amplitude, and output power. The key operating parameters are compared with the preset rated parameter range of the equipment under standard operating conditions, and the deviation index of each parameter is calculated. By combining the deviation indices of all key operating parameters, a health score representing the overall operational health status of the equipment is calculated through weighted fusion. The ratio of output power to corresponding input energy is analyzed over time to form an efficiency curve that reflects the dynamic changes in equipment energy efficiency. The health score and the efficiency curve are integrated in the time dimension to form the equipment operating efficiency spectrum.

4. The smart energy optimization management method based on big data according to claim 3, characterized in that, The basic energy consumption component, flexible adjustable load component, energy consumption noise component, equipment operating efficiency spectrum, and environmental sensitivity vector are input into the feature fusion center for correlation analysis and feature synthesis to generate a panoramic feature model of the energy system, including: Within the feature fusion center, a correlation model is established between the basic energy consumption component and the environmental sensitivity vector to quantify the contribution of environmental impact to basic energy consumption. Establish a mapping relationship between the flexible adjustable load component and the equipment operating efficiency spectrum to characterize the adjustable potential of the flexible load under different equipment efficiency states; The energy consumption noise component is correlated with external random event logs to identify random factors that cause abnormal energy consumption. The correlation model, mapping relationship, and correlation analysis results are spatiotemporally aligned and feature-stitched with the original energy consumption sequence, equipment operating status sequence, and environmental impact factor sequence to construct a multi-dimensional, highly correlated data structure, namely the panoramic feature model of the energy system.

5. The smart energy optimization management method based on big data according to claim 4, characterized in that, Based on the aforementioned energy system panoramic feature model, a load forecasting and optimization decision engine generates load demand forecast curves and corresponding optimization scheduling strategies for future time periods, including: Using the panoramic feature model of the energy system as input, the prediction module in the load forecasting and optimization decision engine is driven. The prediction module combines historical load patterns, future environmental forecast data and planned events to deduce the load demand forecast curve for a specified future period. The load demand forecast curve, the current energy price signal, the power grid dispatch instructions, and the equipment operating efficiency spectrum and flexible adjustable load component information extracted from the energy system panoramic feature model are all input into the decision module of the load forecasting and optimization decision engine. Under the premise of satisfying all equipment operation constraints and grid requirements, the decision module performs global optimization with the objective function of minimizing energy cost or maximizing energy efficiency, and outputs the equipment start-up and shutdown sequence, power adjustment instructions and demand response plan containing specific operation times and quantities, which together constitute the optimized scheduling strategy.

6. The smart energy optimization management method based on big data according to claim 5, characterized in that, The load demand forecast curve, current energy price signals, grid dispatch instructions, and equipment operating efficiency spectrum and flexible adjustable load component information extracted from the energy system panoramic feature model are input together into the decision module of the load forecasting and optimization decision engine, including: The power grid dispatch instructions are analyzed to extract mandatory load reduction requirements, time-limited electricity price incentives, or renewable energy consumption indicators. The expected operating efficiency range of each device in the future time period is obtained from the device operating efficiency spectrum, and the adjustable capacity range and adjustment rate limit of each flexible load are obtained from the flexible adjustable load component information. An optimization time slot is established with time step as the unit. Within each optimization time slot, the value of the load demand forecast curve in the optimization time slot is taken as the basic load that must be met, the flexible and adjustable load component is taken as the optimizable variable, the energy price signal is taken as the cost coefficient, and the information in the power grid dispatch command and the equipment operating efficiency spectrum is transformed into the constraints of the optimization problem.

7. The smart energy optimization management method based on big data according to claim 6, characterized in that, The establishment of optimized time slots based on time steps, whereby within each optimized time slot, the value of the load demand forecast curve in the optimized time slot is used as the basic load that must be met, includes: Determine the time resolution for optimization decisions, and divide the future time period into continuous optimization time slots evenly according to the time resolution; Read the predicted load value corresponding to each optimized time slot from the load demand prediction curve, and directly set the predicted load value as the rigid load that must be met within the corresponding optimized time slot, which is denoted as the basic load that must be met in the optimized time slot. In the optimization model of each optimization time slot, it is ensured that the sum of the total power provided by all operating equipment in the optimization time slot and the energy storage charging and discharging power is equal to the sum of the basic load that the optimization time slot must satisfy and the actual value of the flexible adjustable load of the optimization time slot decision.

8. The smart energy optimization management method based on big data according to claim 7, characterized in that, The decision-making module, under the premise of satisfying all equipment operating constraints and grid requirements, performs global optimization with the objective function of minimizing energy cost or maximizing energy efficiency, including: Construct a mathematical optimization model that includes decision variables, such as the start / stop status and operating power of each device in each optimization time slot, and the adjustment amount of each flexible load in each optimization time slot; The minimum start-up and shutdown time, maximum and minimum output, ramp rate limit, and efficiency curve obtained from the equipment's operating efficiency spectrum are converted into inequality constraints on the decision variables. The mandatory requirements in the power grid dispatching instructions are converted into equal or inequality constraints on the total load or the output of specific equipment. The objective function can be set as the total energy cost, which consists of the cost of purchasing electricity from the grid in each time slot, the cost of equipment operation and maintenance, and the penalty cost; or the objective function can be set as the reciprocal of the total energy efficiency of the system. The mathematical optimization model is solved using mixed integer programming or dynamic programming algorithms to obtain the globally optimal sequence of decision variable values.

9. The smart energy optimization management method based on big data according to claim 8, characterized in that, Also includes: The optimized scheduling strategy is compiled into a sequence of control instructions that can be issued, and the sequence of control instructions is distributed to the corresponding energy system terminals for execution, specifically including: The equipment start-up and shutdown sequence, power regulation command and demand response plan in the optimized scheduling strategy are analyzed and decomposed into operation actions, action execution time points and action parameters for a single specific energy equipment or load unit. Each operation action, action execution time point, and action parameter is translated into a low-level control command that the target device or load unit can directly recognize and execute, based on the communication protocol and control interface specification used by the target device or load unit. According to the chronological order of the action execution time, all the underlying control instructions are sorted, and the execution conditions of the instructions with dependencies are marked. Finally, they are assembled into the control instruction sequence that can be issued with time sequence marks. The process of sorting all low-level control instructions according to the chronological order of their execution times and marking execution conditions for instructions with dependencies includes: The underlying control instructions are sorted in ascending order based on the action execution time point as the primary keyword. Examine all low-level control commands and identify logical dependencies, such as a device start command that can only be executed after another device shut down command has been executed, or a power adjustment command that can only be executed after a device start command has been confirmed. Add a pre-instruction identifier to the underlying control instructions that have dependencies. The pre-instruction identifier indicates that the underlying control instruction can only be issued after the pre-instruction it depends on has been confirmed to be executed successfully. This forms an instruction queue with timing and condition markers, and completes the final assembly of the issueable control instruction sequence.

10. A smart energy optimization management system based on big data, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the smart energy optimization management method based on big data as described in any one of claims 1 to 9.

Images