Energy management system and method based on Internet of Things and big data

Abstract

The invention discloses an energy management system and method based on the Internet of Things and big data, and particularly relates to the technical field of big data, and the method comprises the following steps: collecting the operation data of a preset type of energy device in an energy management region in real time based on the Internet of Things, and generating a single device data set; the method comprises the following steps: identifying whether equipment has an early sign of energy consumption abnormity or not through monomer analysis, performing feature extraction and analysis on abnormal equipment, generating an equipment operation quality index and a load adaptability index, inputting a pre-trained machine learning model, and dividing the equipment into high-quality and low-quality operation equipment; in combination with a feature analysis result, the number of low-quality equipment and the total amount of the equipment, an optimization strategy form is generated through fuzzy logic reasoning, and accurate classification and intelligent optimization management of the energy equipment are achieved; through classification and reasoning, limited optimization resources are preferentially allocated to low-quality operation equipment, and invalid energy consumption is reduced by improving the operation quality of energy equipment.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and more specifically, to an energy management system and method based on the Internet of Things and big data. Background Art

[0002] The Internet of Things refers to connecting various physical devices to information systems through the network, enabling automated data collection and remote control. In energy management, the Internet of Things technology is applied to sensors, smart devices, etc., which can collect real-time consumption data of energy such as electricity and transmit it to the central system through the network for monitoring and analysis. These real-time data provide the basis for subsequent energy optimization.

[0003] Big data technology stores, processes and analyzes large amounts of complex energy data, and can extract valuable information and trends from the data through methods such as data mining and pattern recognition.

[0004] With the acceleration of industrialization and modernization, various types of energy-consuming equipment are widely distributed in different energy management areas, such as industrial parks, commercial centers, and public facilities. How to analyze the energy consumption data of different time periods, different regions, and different equipment, timely discover abnormal energy use, and then take corresponding optimization measures, and combine the energy management of the Internet of Things and big data to determine the optimization strategy for adjusting energy-consuming equipment according to different usage scenarios and needs is a major problem. Therefore, an energy management system and method based on the Internet of Things and big data is proposed here to solve the above problems. Summary of the invention

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An energy management method based on the Internet of Things and big data includes the following steps:

[0007] Through the Internet of Things, real-time operation data of all energy equipment of preset types in the energy management area are obtained to obtain operation data sets of multiple single energy equipment;

[0008] Conduct single-unit analysis based on the operating data set of single energy equipment to identify early signs of abnormal energy consumption during the operation of single energy equipment;

[0009] Perform feature extraction and feature analysis on single energy devices that show early signs of abnormal energy consumption, obtain feature analysis results and transmit them to a pre-trained machine learning model to classify single energy devices into high-quality operating devices and low-quality operating devices;

[0010] The results of feature analysis in the energy management area, the number of low-quality operating devices classified as individual energy devices, and the total number of all energy devices of preset types are used together for optimization type reasoning to generate an optimization strategy form.

[0011] In a preferred embodiment, monomer analysis refers to:

[0012] In a fixed time window, a time series of operating load data is obtained from the operating data set of the single energy device. The operating load data in the time series is the load value of the single energy device at each time point. In order to obtain the central trend of the load, the average value of all load values ​​in the fixed time window is calculated. In order to measure the degree of load fluctuation, the standard deviation of all load values ​​in the fixed time window is calculated, and then the absolute value of the difference between the average value and the preset benchmark value is divided by the standard deviation. The result of the calculation is the abnormal factor AF.

[0013] In a preferred embodiment, the logic for identifying whether there are early signs of abnormal energy consumption during the operation of a single energy device is:

[0014] The abnormal factor AF is compared with the preset standard threshold. If the abnormal factor AF is greater than the preset standard threshold, an abnormal signal is generated. If the abnormal factor AF is less than or equal to the preset standard threshold, a normal signal is generated. When the normal signal is generated, it is determined that there are no early signs of abnormal energy consumption during the operation of the single energy device. At this time, the single energy device is defaulted to a high-quality operating device. When the abnormal signal is generated, it is determined that there are early signs of abnormal energy consumption during the operation of the single energy device.

[0015] In a preferred embodiment, extracting features from single energy devices that show early signs of abnormal energy consumption refers to:

[0016] In view of the operation quality and load adaptability of single energy equipment, we extract the equipment performance characteristic data group that focuses on reflecting the stability of the single energy equipment during operation, and the load adaptability characteristic data group that focuses on the response ability of the single energy equipment to load changes. The equipment performance characteristic data group generates an equipment operation quality index during feature analysis, and the load adaptability characteristic data group generates a load adaptability index during feature analysis.

[0017] In a preferred embodiment, the logic for obtaining the equipment operation quality index is:

[0018] The following assessments are made based on the equipment performance characteristic data set:

[0019] To evaluate load stability, the calculation formula is:

[0020] N represents the number of data points in the time window, L i represents the load value at the i-th moment, L i-1 Represents the load value at the i-1th moment, Mean L represents the mean value of load data, LSI represents the load stability coefficient;

[0021] To evaluate the temperature stability, the calculation formula is:

[0022] Max T is the maximum value of the temperature data in the time window, Min T is the minimum value of the temperature data in the time window, Mean T is the mean of the temperature data in the time window, TSI represents the temperature stability coefficient;

[0023] To evaluate vibration stability, the calculation formula is:

[0024] V i is the vibration value at the i-th moment in the time window, Mean V is the mean value of the vibration data in the time window, N is the number of data points in the time window, and VSI is the vibration stability coefficient;

[0025] Energy efficiency evaluation is performed, and the calculation formula is:

[0026] Output Work is the workload of a single energy device in the time window, Energy Consumption is the energy consumption of a single energy device in the time window, is the average operating power of a single energy device in the time window, and EEI represents the energy efficiency coefficient;

[0027] Integrate the evaluation results of each feature data in the equipment performance feature data group into a unified scoring mechanism, specifically:

[0028] DPQI stands for Device Performance Quality Index.

[0029] In a preferred embodiment, the logic for obtaining the load adaptability index is:

[0030] Based on the load adaptability characteristic data set, the evaluation is conducted from three aspects: the response capability of the single energy equipment to load fluctuations, the flexibility of load regulation, and the load balancing:

[0031] Load Adaptation Response Capability Assessment:

[0032]

[0033] Lt Represents the load value at time t, Mean L represents the mean value of the load data in the time window, N represents the number of data points in the time window, Max L Indicates the maximum value of the load data in the time window, Min L Indicates the minimum value of the load data in the time window, sign(x) indicates the sign function, returns 1 if x is greater than 0, returns -1 if x is less than 0, otherwise returns 0, LARI indicates the responsiveness value;

[0034] Load Regulation Sensitivity Evaluation:

[0035] L n is the load value at the nth moment, N is the number of data points in the time window, Δt is the time interval of load measurement, and LRSI is the regulation sensitivity value;

[0036] Load adaptability balance evaluation:

[0037] L k is the load value at the kth moment, LABI represents the load adaptation value;

[0038] The calculation formula of load adaptability index is:

[0039] log represents natural logarithm function, and LAI represents load adaptability index.

[0040] In a preferred embodiment, the machine learning model is a convolutional neural network model, and the equipment operation quality index and the load adaptability index are used together as input data of the convolutional neural network model, and the output result of the convolutional neural network model is 0 or 1. When the output result is 0, it indicates that the single energy equipment is classified as low-quality operation equipment, and when the output result is 1, it indicates that the single energy equipment is classified as high-quality operation equipment.

[0041] In a preferred embodiment, optimizing type inference refers to:

[0042] The equipment operation quality index, load adaptability index, and the ratio of the number of single energy equipment classified as low-quality operation equipment to the total number of all energy equipment of preset types are taken as input variables, and the optimization type of energy equipment in the energy management area is taken as the output variable. The input variables are fuzzified and the values ​​of the input variables are converted into fuzzy sets. The output variables are fuzzified and the output variables are converted into fuzzy sets. Fuzzy rules are formulated to describe the adaptation degree of the optimization type of energy equipment under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to obtain the optimization type of each energy equipment in the energy management area. Then the optimization types of all energy equipment are summarized to obtain the optimization strategy form.

[0043] In a preferred embodiment, an energy management system based on the Internet of Things and big data includes:

[0044] The data collection module obtains the operating data of all energy equipment of preset types in the energy management area in real time through the Internet of Things, and obtains the operating data sets of multiple single energy equipment

[0045] The single-unit analysis module performs single-unit analysis based on the operating data set of the single energy equipment to identify whether there are early signs of abnormal energy consumption during the operation of the single energy equipment;

[0046] The feature extraction module extracts features from individual energy devices that show early signs of abnormal energy consumption;

[0047] The feature analysis module performs feature analysis operations based on the features extracted by the feature extraction module to obtain feature analysis results;

[0048] The equipment classification module transmits the results of feature analysis to the pre-trained machine learning model to classify single energy equipment into high-quality operating equipment and low-quality operating equipment;

[0049] The optimization reasoning module performs optimization type reasoning on the results of feature analysis, the number of low-quality operating devices classified as single energy devices, and the total number of all energy devices of preset types in the energy management area to generate an optimization strategy form.

[0050] Technical effects and advantages of the present invention:

[0051] The present invention uses the equipment operation quality index and load adaptability index as the input of the convolutional neural network model and utilizes machine learning technology to divide single energy equipment into high-quality operation equipment and low-quality operation equipment. The classification results are clear and provide a clear basis for the formulation of optimization strategies.

[0052] The present invention maintains the existing status of high-quality running equipment, reduces unnecessary optimization costs, and performs targeted optimization on low-quality running equipment based on specific problems, such as equipment stability optimization, load adjustment optimization, or a combination of the two, to improve equipment operation efficiency. Using fuzzy logic reasoning methods, multiple input variables such as equipment operation quality index, load adaptability index, and low-quality equipment ratio are converted into fuzzy sets, and the equipment status and the overall operation of the region are comprehensively considered to generate a dynamically adaptive optimization strategy.

[0053] The present invention targets devices and regions in different states: Low proportion: local optimization of single devices to accurately solve the operating state of the problem equipment. High proportion: overall regional optimization to solve systemic energy distribution or load scheduling problems. By generating an optimization strategy form, the optimization types are summarized into clear operation plans, which not only solves the problems of single devices, but also improves the overall operation efficiency of regional energy management.

[0054] Through classification and reasoning, the present invention preferentially allocates limited optimization resources to low-quality operating equipment to avoid wasting resources on high-quality equipment that does not need to be optimized. It identifies equipment with abnormal energy consumption at an early stage, optimizes or repairs it in time, and prevents energy waste and high maintenance costs caused by equipment failure. Through stability analysis and load adaptability optimization, it reduces the ineffective energy consumption of equipment and achieves energy-saving goals in the region.

[0055] The present invention coordinates and optimizes the equipment in the area, improves the rationality of load distribution between equipment, and avoids energy overload or idleness. Whether it is an industrial park, commercial building or public facility, the system can flexibly adapt to energy management needs of different scales and complexities. By generating a global optimization strategy, the optimization needs of individual equipment and regions are balanced, and ultimately the economic benefits and operational reliability of the entire energy management area are improved.

[0056] The present invention reduces ineffective energy consumption by improving the operating quality of energy equipment. The system effectively reduces carbon emissions in energy use, helping to achieve low-carbon goals. It also optimizes the operating status of equipment and can reduce equipment losses caused by long-term abnormal operation, thereby extending the service life of the equipment and reducing resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0058] Figure 1 This is a schematic diagram of an energy management method based on the Internet of Things and big data in the present invention.

[0059] Figure 2 This is a schematic diagram of an energy management system based on the Internet of Things and big data in the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0061] Reference Figure 1 - Figure 2 The following examples are obtained:

[0062] Example 1: The energy management method based on the Internet of Things and big data refers to using the Internet of Things technology to collect energy usage data in real time, and processing and optimizing these data through big data analysis technology to improve energy efficiency, reduce energy costs, reduce energy waste, and ultimately achieve intelligent management and optimization of energy. This method relies on the real-time perception of the Internet of Things and the intelligent analysis of big data to achieve dynamic monitoring and optimization decisions of the energy system.

[0063] The role of the Internet of Things in energy management: The Internet of Things refers to connecting various physical devices with information systems through the network, enabling automated data collection and remote control. In energy management, the Internet of Things technology is applied to sensors, smart devices, etc., which can collect real-time consumption data of energy such as electricity and transmit it to the central system through the network for monitoring and analysis. These real-time data provide the basis for subsequent energy optimization.

[0064] Big data technology stores, processes and analyzes large amounts of complex energy data, and can extract valuable information and trends from the data through methods such as data mining and pattern recognition. For example, by analyzing energy consumption data of different time periods, different regions and different devices, abnormal energy use can be discovered, and corresponding optimization measures can be taken. Energy management combined with the Internet of Things and big data can determine the optimization strategy for adjusting energy consumption equipment according to different usage scenarios and needs.

[0065] The present invention proposes an energy management method based on the Internet of Things and big data, comprising the following steps:

[0066] Through the Internet of Things, the operating data of all energy equipment of preset types such as power equipment in the energy management area can be obtained in real time, and the operating data sets of multiple single energy equipment can be obtained; the Internet of Things technology can realize real-time monitoring of all equipment in the energy management area to ensure that the operating status of all single energy equipment is included in the analysis scope. The acquired data covers the real-time load, energy consumption, temperature, vibration and other parameters of the equipment, providing a comprehensive basic data set for subsequent analysis. The collected operating data set is the basis for subsequent single equipment analysis, feature extraction, and optimization reasoning. The real-time nature of the data ensures that problems can be discovered and handled in a timely manner, thereby improving the efficiency of energy equipment management.

[0067] Single-unit analysis is performed based on the operating data set of single energy equipment to identify whether there are early signs of abnormal energy consumption during the operation of single energy equipment; by analyzing each single device, it is possible to accurately find which devices have early signs of abnormal energy consumption, avoiding the waste of resources for comprehensive optimization of all devices. By analyzing abnormal fluctuations in equipment operating data such as abnormal load changes, it is possible to intervene before the problem becomes serious, reducing the risk of failure and energy waste. Through single-unit analysis, the scope of subsequent optimization can be narrowed to focus only on those devices that show signs of abnormality, saving resources and time. Early detection of abnormal behavior of single devices can prevent these problems from gradually expanding into regional energy management issues.

[0068] Feature extraction and feature analysis are performed on single energy devices that show early signs of abnormal energy consumption. The results of feature analysis are obtained and transmitted to the pre-trained machine learning model to classify single energy devices into high-quality operating devices and low-quality operating devices. Feature extraction and analysis of devices with early signs can clarify the specific manifestations of abnormalities in equipment operation, such as poor load adaptability and low equipment stability. Feature extraction includes key feature data such as the equipment's operating quality index and load adaptability index, which provide a basis for subsequent classification. Through pre-trained machine learning models (such as convolutional neural networks), devices can be efficiently and accurately classified based on feature data, and devices can be divided into high-quality operating devices and low-quality operating devices, which helps to clearly distinguish which devices need optimization intervention and which devices can continue to maintain their current operating status. The use of machine learning models avoids human intervention and misjudgment in traditional analysis, and improves the intelligence and efficiency of energy management through automated classification.

[0069] The results of feature analysis, the number of low-quality running devices for individual energy equipment, and the total number of all energy equipment of preset types in the energy management area are used together for optimization type reasoning to generate an optimization strategy form. Not only does it focus on the operating status of a single device, but it also combines the number and distribution of equipment in the entire energy management area to evaluate the health of energy equipment in the area. Feature analysis results: used to clarify the operating characteristics of each device; the number of low-quality equipment: reflects the distribution of problem equipment in the entire area; the total number of equipment: provides a benchmark for the overall equipment scale of the region, which is convenient for calculating the proportion of low-quality equipment. Through fuzzy logic reasoning, the most suitable optimization strategy for the current situation of the region is generated according to the input variables (feature analysis results, proportion of low-quality equipment). The determination of the optimization type is the core of energy equipment management. Through intelligent optimization reasoning, the error of human decision-making can be reduced, and the reasonable allocation of resources can be achieved. Generate an optimization strategy form: The generation of the form can clearly display the optimization type of each device or area, provide an executable optimization plan for operation and maintenance personnel, dynamically adjust the optimization strategy according to the equipment characteristics and operating conditions of different areas, and improve the flexibility of energy equipment management.

[0070] Single-unit analysis means: within a fixed time window, from the operating data set of a single energy device, a time series of operating load data is obtained. The operating load data in the time series is the load value of the single energy device at each time point. To obtain the central trend of the load, the average value of all load values ​​within the fixed time window is calculated. To measure the degree of load fluctuation, the standard deviation of all load values ​​within the fixed time window is calculated. Then, the absolute value of the difference between the average value and the preset benchmark value is divided by the standard deviation. The result of the calculation is the abnormal factor AF.

[0071] Extract the operating load data from the operating data set of the single energy equipment to form a time series, reflecting the load conditions of the equipment at different time points. Calculate the average value of all load values ​​within a fixed time window to indicate the central trend of the equipment load and reflect the typical load level of the equipment operation. Calculate the standard deviation of all load values ​​within a fixed time window to indicate the degree of load fluctuation and reflect the stability or instability of the equipment operation load. The abnormal factor quantifies the deviation of the central trend (average value) of the load from the preset reference value, and normalizes the result with the standard deviation to avoid interference from the degree of load fluctuation. By calculating the central trend of the load through the average value, it can be found whether the equipment load deviates from the normal state. The introduction of the standard deviation ensures the quantification of the degree of load fluctuation, so that the calculation of the abnormal factor can take into account the stability of the equipment operation. The abnormal factor AF normalized with the standard deviation is a relative value, which can have consistent discriminant significance between different devices or in different time windows.

[0072] The logic for identifying whether there are early signs of abnormal energy consumption during the operation of a single energy device is as follows: compare the abnormal factor AF with the preset standard threshold. If the abnormal factor AF is greater than the preset standard threshold, an abnormal signal is generated. If the abnormal factor AF is less than or equal to the preset standard threshold, a normal signal is generated. When a normal signal is generated, it is determined that there are no early signs of abnormal energy consumption during the operation of the single energy device. At this time, the single energy device is defaulted to a high-quality operation device. When an abnormal signal is generated, it is determined that there are early signs of abnormal energy consumption during the operation of the single energy device.

[0073] By quantitatively comparing abnormal factors, it is possible to accurately identify abnormal energy consumption of equipment without subjective judgment. Real-time calculation and threshold judgment of abnormal factors can capture abnormalities at the early stage of the problem and prevent small problems from evolving into major failures. Further optimization is only carried out when abnormal signals are generated, and normal signals are directly marked as high-quality operating equipment, saving computing and optimization resources. The calculation of abnormal factors comprehensively analyzes the operating status of single energy equipment by comprehensively considering the central trend and fluctuation degree of the load. By normalizing the deviation of abnormal factors from the baseline value, abnormal detection is consistent for different devices and different operating environments.

[0074] By calculating the anomaly factor in real time, it is possible to quickly identify whether there are early signs of abnormal energy consumption in individual devices. The early detection mechanism can take timely measures to reduce the risk of failure before the problem expands or affects regional energy management. When a normal signal is generated, the device is directly marked as a high-quality operating device, eliminating unnecessary feature extraction and optimization operations. Subsequent feature extraction and optimization are only performed on abnormal signal devices, focusing on problem devices, avoiding comprehensive inspection and adjustment of all devices, saving time and resources. The results of the anomaly factor and its signal generation provide a data basis for subsequent optimization decisions. High-quality equipment does not need further optimization and maintains its current state. Abnormal equipment enters the feature extraction and classification stage to provide basic data for optimization type reasoning.

[0075] Feature extraction of individual energy devices that show early signs of abnormal energy consumption refers to:

[0076] In view of the operation quality and load adaptability of single energy equipment, we extract the equipment performance characteristic data group that focuses on reflecting the stability of the single energy equipment during operation, and the load adaptability characteristic data group that focuses on the response ability of the single energy equipment to load changes. The equipment performance characteristic data group generates an equipment operation quality index during feature analysis, and the load adaptability characteristic data group generates a load adaptability index during feature analysis.

[0077] The equipment performance characteristic data group is mainly used to reflect the stability and health status of the single energy equipment during operation. These characteristic data can directly reflect whether there are performance anomalies or potential failures in the operation of the equipment. The equipment performance characteristic data group can comprehensively evaluate the operating stability of the single equipment and help identify whether there is performance degradation or potential failure of the equipment during operation.

[0078] The load adaptability characteristic data set is used to reflect the responsiveness and adaptability of the equipment in response to load changes. These characteristics can reveal whether the equipment responds sensitively to dynamic loads and whether the adjustment is smooth. The load adaptability characteristic data set can accurately reflect the performance of the equipment in a dynamic load environment and help determine whether the equipment can efficiently cope with load fluctuations.

[0079] Equipment operation quality index: The higher the value, the stronger the equipment operation stability and the better the performance. The lower the value, the risk of performance degradation or failure of the equipment. Load adaptability index: The higher the value, the stronger the equipment's ability to respond to load changes. The lower the value, the poorer the equipment's adaptability to dynamic load changes and the need for load adjustment optimization.

[0080] Feature extraction of single energy equipment that shows early signs of abnormal energy consumption can fully reveal the operating status of the equipment from two aspects: equipment operation quality and load adaptability. By extracting the equipment performance feature data group and the load adaptability feature data group, the equipment operation quality index and the load adaptability index are generated respectively, providing reliable basic data support for subsequent classification, optimization reasoning and strategy formulation. This approach ensures the accuracy and efficiency of energy management, and at the same time lays a solid foundation for improving equipment operation quality and regional energy management efficiency.

[0081] The logic for obtaining the equipment operation quality index is:

[0082] The following assessments are made based on the equipment performance characteristic data set:

[0083] To evaluate load stability, the calculation formula is:

[0084] N represents the number of data points in the time window, L i represents the load value at the i-th moment, L i-1 Represents the load value at the i-1th moment, Mean Lrepresents the mean of the load data, and LSI represents the load stability coefficient; this formula calculates the sum of the load value changes between each two consecutive moments, normalized to the percentage of relative load change. By calculating the relative change between load values, the stability of the equipment load can be more accurately reflected. Unlike simple standard deviation or mean comparison, this method can reflect short-term changes in load and help identify instantaneous load fluctuations, especially in systems with small or relatively stable loads, which may cause greater impacts.

[0085] To evaluate the temperature stability, the calculation formula is:

[0086] Max T is the maximum value of the temperature data in the time window, Min T is the minimum value of the temperature data in the time window, Mean T The TSI is the average value of the temperature data within the time window. The formula reflects the percentage of the temperature fluctuation range (the difference between the maximum and minimum values) relative to the average temperature. Unlike the standard deviation, this formula focuses on the range of temperature fluctuations rather than the degree of fluctuation. The purpose of this is to capture extreme fluctuations in temperature, especially when the temperature changes drastically. Too high or too low a temperature of the equipment may have a serious impact on its operation, and the standard deviation may not be enough to reflect such extreme conditions.

[0087] To evaluate vibration stability, the calculation formula is:

[0088] V i is the vibration value at the i-th moment in the time window, Mean V is the mean value of the vibration data in the time window, N is the number of data points in the time window, and VSI is the vibration stability coefficient; this formula calculates the root mean square deviation of the vibration data and standardizes it into a relative proportion. Using the root mean square deviation instead of a simple standard deviation or mean can more accurately capture the sharp fluctuations in vibration data, especially when there are shocks or large fluctuations in the equipment. The root mean square deviation is particularly sensitive to signals with large fluctuations, so it can effectively reflect the mechanical stability of the equipment.

[0089] Energy efficiency evaluation is performed, and the calculation formula is:

[0090] Output Work is the workload of a single energy device in the time window, Energy Consumption is the energy consumption of a single energy device in the time window, EEI is the average operating power of a single energy device in the time window, and the formula calculates the energy efficiency index and introduces an exponential decay factor To penalize the energy efficiency evaluation when the power is low, by introducing the exponential decay factor, the impact of the load change of the equipment on the energy efficiency can be considered. When the equipment power is low, the energy efficiency performance may be poor. Therefore, this exponential decay mechanism can better adjust the energy efficiency evaluation results, making the energy efficiency evaluation more reasonable when the load is light.

[0091] Integrate the evaluation results of each feature data in the equipment performance feature data group into a unified scoring mechanism, specifically:

[0092] DPQI stands for Device Performance Quality Index. This processing method integrates the mutual influence between the evaluation results of various characteristics of the equipment, taking into account not only the contribution of each characteristic evaluation result, but also the nonlinear relationship between them. Through the calculation of cube roots, the excessive amplification of the comprehensive results by extreme values ​​is reduced, making the final performance quality index more balanced.

[0093] The logic for obtaining the load adaptability index is:

[0094] Based on the load adaptability characteristic data set, the evaluation is conducted from three aspects: the response capability of the single energy equipment to load fluctuations, the flexibility of load regulation, and the load balancing:

[0095] Load Adaptation Response Capability Assessment:

[0096]

[0097] L t Represents the load value at time t, Mean L represents the mean value of the load data in the time window, N represents the number of data points in the time window, Max L Indicates the maximum value of the load data in the time window, Min L Indicates the minimum value of the load data in the time window, sign(x) represents the sign function, returns 1 if x is greater than 0, returns -1 if x is less than 0, otherwise returns 0, LARI represents the responsiveness value; this formula calculates the deviation between the load value and the load mean at each moment, and also takes the sign (positive or negative) of the deviation into account. In this way, it can be determined whether the load is continuously rising or falling, and how the device adapts to these changes. The sign function sign(x) can help determine the direction of load change, thereby reflecting the device's responsiveness to load changes in different directions.

[0098] By comparing the absolute value of the deviation at each moment with the total absolute deviation of the load data, the intensity ratio of the device response is obtained. If the ratio is close to 1, it means that the device has a very strong adaptive response when the load changes, changes quickly and can follow the load fluctuation well; if it is close to 0, it means that the device response is weak and it is difficult to adapt to load changes.

[0099] The difference between the maximum and minimum values ​​of the load is used as a normalization factor to measure the extreme fluctuation range of the load. This part can examine the responsiveness of the equipment under extreme load conditions. If the load variation range is small, the equipment has strong adaptability and good responsiveness; if the load fluctuation range is large, the equipment has poor adaptability and the responsiveness value will be adjusted down accordingly.

[0100] By considering both the direction and intensity of load changes, the response speed and adaptability of the equipment to load fluctuations can be more comprehensively reflected. Compared with traditional standard deviation or fluctuation analysis, this formula adds an assessment of the responsiveness to the direction of load changes, making the load change pattern have a greater impact on the evaluation results. The difference between the maximum and minimum values ​​of load fluctuations reflects the challenge of extreme fluctuations to the adaptability of the equipment, which helps to evaluate the performance of the equipment under extreme load conditions.

[0101] Load Regulation Sensitivity Evaluation:

[0102] L n is the load value at the nth moment, N is the number of data points in the time window, Δt is the time interval for load measurement, and LRSI is the regulation sensitivity value; the load regulation sensitivity reflects the response speed of the device to load changes. By calculating the load change rate and normalizing it with the average value of the load, the sensitivity of the device in the face of load changes can be effectively evaluated. If the device can respond quickly to load changes, the sensitivity is high. Otherwise, it indicates that the device has a weak ability to adapt to load fluctuations. The calculation method used here is different from the common fluctuation calculation method. It focuses on the rate of load change over time, which is particularly critical for high-frequency load changes.

[0103] Load adaptability balance evaluation:

[0104] L kis the load value at the kth moment, and LABI represents the load adaptation value; first, the deviation of the load from the mean at each moment is calculated, and the deviation is normalized with the mean. This can reflect the relative amplitude of the load fluctuation. The average of these deviations is calculated to obtain the overall fluctuation of the load within the time window. If the device maintains a relatively stable load state during this period, the deviation is small, otherwise the deviation is large. In order to further judge the adaptability of the device within the load range, the distance between the maximum and minimum load values ​​and the mean is also considered. This helps to evaluate whether the device can restore balance in a short time when the load fluctuates greatly. If the extreme fluctuation of the load is large and the recovery time is long, it indicates that the load adaptability is poor. This formula combines the relative fluctuation degree and the extreme load recovery ability, and more comprehensively reflects the balance degree of the load adaptability of the device, emphasizing the relative deviation of the load from the average value and the adaptability of the device to the load fluctuation range. The load adaptation value LABI not only examines the relative stability of the device during the load fluctuation process, but also introduces the difference between the maximum and minimum load values ​​and the mean to further examine the adaptability of the device to extreme load fluctuations.

[0105] The calculation formula of load adaptability index is:

[0106] log represents the natural logarithm function, and LAI represents the load adaptability index. The square of the load regulation sensitivity is used to reflect the greater impact of sensitivity on load adaptability, because the regulation sensitivity plays an important role in the adaptability of the equipment. Too low sensitivity will significantly reduce the load adaptability, so its impact needs to be amplified. Introducing the log function to LABI can moderately reduce the excessive balance, so that the contribution of this feature shows a decreasing trend. Excessive balance may not always be a good thing, and its impact on the final index needs to be controlled. LARI and LRSI are directly multiplied to reflect the interaction between the equipment's load change response capability and regulation speed. If either of the two is low, the final result will be significantly reduced. Adding log(1+LABI) to the denominator means that when the balance is high (such as small load fluctuations), the adaptability of the equipment will naturally increase, but it will not increase indefinitely, which is more in line with the actual situation.

[0107] The machine learning model is a convolutional neural network model. The equipment operation quality index and load adaptability index are used as input data of the convolutional neural network model, and the output result of the convolutional neural network model is 0 or 1. When the output result is 0, it indicates that the single energy equipment is classified as low-quality operation equipment, and when the output result is 1, it indicates that the single energy equipment is classified as high-quality operation equipment.

[0108] Equipment operation quality index: reflects the operation stability, energy efficiency and overall performance health of single energy equipment. The higher the value, the more stable and healthier the equipment operation status. Load adaptability index: reflects the adaptability of single energy equipment to dynamic load changes. The higher the value, the faster the equipment can respond to load changes and return to a stable state.

[0109] The two indexes comprehensively reflect the operating status of single energy equipment and describe the equipment performance from the two dimensions of static stability and dynamic adaptability. The input data dimension is small, which is convenient for efficient processing and ensures the full expression of information. Input layer of the convolutional neural network model: The input data is the equipment operation quality index and the load adaptability index, in the form of a two-dimensional vector [Q, A], where Q represents the operation quality index and A represents the load adaptability index. Convolution layer: Apply a one-dimensional convolution operation to extract the correlation pattern between the input data. For example, a high operation quality index but a low load adaptability index may indicate that the equipment operation status needs to be improved. Activation function: Use a nonlinear activation function (such as ReLU) to perform nonlinear transformation on the features to increase the expressive power of the model. Pooling layer: Pool the convolved features to reduce the data dimension while retaining important information. Fully connected layer: Input the pooled features into the fully connected layer to map the high-dimensional features into a probability distribution for classification. Finally, a single neuron output layer (using the Sigmoid activation function) generates a classification result of 0 or 1. By using the convolutional neural network model to take the equipment operation quality index and load adaptability index as input, accurate classification of single energy equipment can be achieved. The classification result is 0 or 1, which is used to indicate whether the equipment needs to be optimized. The intelligence and efficiency of the model provide a solid foundation for the overall optimization of the energy management area, while greatly improving the intelligence level of equipment management and resource allocation efficiency. This classification method based on machine learning is an important tool for modern energy management.

[0110] Optimizing type inference means:

[0111] The equipment operation quality index, load adaptability index, and the ratio of the number of single energy equipment classified as low-quality operation equipment to the total number of all energy equipment of preset types are taken as input variables, and the optimization type of energy equipment in the energy management area is taken as the output variable. The input variables are fuzzified and the values ​​of the input variables are converted into fuzzy sets. The output variables are fuzzified and the output variables are converted into fuzzy sets. Fuzzy rules are formulated to describe the adaptation degree of the optimization type of energy equipment under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to obtain the optimization type of each energy equipment in the energy management area. Then the optimization types of all energy equipment are summarized to obtain the optimization strategy form.

[0112] Fuzzification of input variables: Convert the values ​​of equipment operation quality index, load adaptability index and proportion of low-quality equipment into fuzzy sets: Equipment operation quality index: low, medium, high. Load adaptability index: weak, general, strong. Proportion of low-quality equipment: low, medium, high.

[0113] The fuzzification process uses membership functions, such as triangular or trapezoidal membership functions, to map values ​​to the degree of fuzzy sets. For example, if the equipment operation quality index is 0.7, it may be mapped by the membership function to a medium membership of 0.8 and a high membership of 0.2.

[0114] The output variables (optimized type of energy equipment) are also fuzzified, for example:

[0115] Single device optimization: Only optimize individual devices.

[0116] Multi-dimensional optimization of single equipment: equipment stability optimization + load scheduling optimization.

[0117] Overall regional optimization: Collaborative optimization of energy equipment in the entire region.

[0118] The fuzzification of output variables expresses the optimization strategy as a fuzzy set of different optimization types, which facilitates fuzzy reasoning based on input variables.

[0119] Formulation of fuzzy rules: Fuzzy rules describe the logical relationship between the combination of different input variables (operation quality index, load adaptability index and proportion of low-quality equipment) and the output variable (optimization type).

[0120] Example rules: Rule 1: If the operating quality index is high and the load adaptability index is strong, and the proportion of low-quality equipment is low, then the optimization type is no optimization required.

[0121] Rule 2: If the operation quality index is medium and the load adaptability index is weak, and the proportion of low-quality equipment is medium, then the optimization type is single-equipment optimization.

[0122] Rule 3: If the operation quality index is low and the load adaptability index is weak, and the proportion of low-quality equipment is high, then the optimization type is regional overall optimization. Significance of the rule: Fuzzy rules link different combinations of input variables with the degree of adaptability of specific optimization types, making it easier to make optimization decisions quickly in complex and changing operating environments.

[0123] Steps of fuzzy reasoning: Fuzzification of input variables: Convert the input data (equipment operation quality index, load adaptability index and proportion of low-quality equipment) into the membership degree of the corresponding fuzzy set.

[0124] Fuzzy rule matching: According to the membership of the input variables, it is matched with the preset fuzzy rules to generate the matching strength of each rule. The matching strength reflects the degree of adaptability of the current input variable combination to each rule.

[0125] Reasoning and fuzzy output: According to the matching strength of fuzzy rules, the fuzzy set of fuzzy output variables (optimization type) is calculated.

[0126] Defuzzification: Convert the fuzzy output set into a specific optimization type. For example, the energy equipment optimization type is: when the ratio of the number of low-quality operating equipment to the total number of all energy equipment of the preset type is low, the individual energy equipment is optimized separately, which is manifested as: equipment stability optimization of individual energy equipment, load scheduling optimization of individual energy equipment, or equipment stability and load scheduling optimization of individual energy equipment. When the ratio of the number of low-quality operating equipment to the total number of all energy equipment of the preset type is high, the energy equipment in the energy management area is coordinated and optimized for the equipment group.

[0127] If the operating quality and load adaptability index of most equipment perform well, but some equipment has problems (that is, the proportion of low-quality equipment is low), the problems are more likely to be concentrated on individual equipment, so it is priority to optimize these individual equipment (such as equipment stability optimization or load scheduling optimization).

[0128] If the proportion of low-quality equipment is high, this usually indicates that the problem is not limited to individual equipment, but that the overall operating status of regional energy equipment is poor. This situation requires optimization of the entire energy area, involving coordinated optimization operations of multiple devices.

[0129] The proportion of low-quality equipment directly determines the focus and scope of optimization: When the proportion is low (optimization for individual equipment): If the proportion of low-quality equipment is low, it means that most equipment is running well, and only a few equipment may have poor operating quality due to specific reasons (such as overload or performance degradation). In this case, it is more efficient to optimize the stability of individual equipment, optimize load scheduling, or a combination of the two. Taking targeted measures for individual problems can reduce costs and quickly restore equipment performance.

[0130] When the proportion is high (for overall optimization of the region): If the proportion of low-quality devices is high, it indicates that the problem with the equipment is universal. In this case, optimizing individual devices alone may not be able to effectively solve the problem, and systematic optimization of the entire region is required, such as coordinated optimization of the device group.

[0131] When the problem only exists in a small number of devices, directly optimizing these devices can quickly produce results while saving resources and costs. When the proportion of low-quality devices is high, the problem is more likely to be caused by systemic reasons, such as unreasonable load distribution strategies within the region. At this time, single-unit optimization is no longer sufficient, and overall optimization is needed to improve regional operating efficiency.

[0132] Adaptability of multi-level optimization strategies: The division of optimization types can ensure the rationality of resource allocation. For example: low proportion means single device optimization: for device stability and load scheduling. High proportion means overall regional optimization: collaborative scheduling optimization between devices. Determine the priority optimized devices and optimization types through reasoning to ensure that the optimization resources are concentrated and allocated to the most needed places to avoid waste. Optimization type reasoning ensures full coverage of optimization logic from single devices to the entire region, which helps to comprehensively improve the performance and load balancing of devices in the energy management area.

[0133] Embodiment 2: An energy management system based on the Internet of Things and big data, comprising:

[0134] The data collection module obtains the operating data of all energy equipment of preset types in the energy management area in real time through the Internet of Things, and obtains the operating data sets of multiple single energy equipment

[0135] The single-unit analysis module performs single-unit analysis based on the operating data set of the single energy equipment to identify whether there are early signs of abnormal energy consumption during the operation of the single energy equipment;

[0136] The feature extraction module extracts features from individual energy devices that show early signs of abnormal energy consumption;

[0137] The feature analysis module performs feature analysis operations based on the features extracted by the feature extraction module to obtain feature analysis results;

[0138] The equipment classification module transmits the results of feature analysis to the pre-trained machine learning model to classify single energy equipment into high-quality operating equipment and low-quality operating equipment;

[0139] The optimization reasoning module performs optimization type reasoning on the results of feature analysis, the number of low-quality operating devices classified as single energy devices, and the total number of all energy devices of preset types in the energy management area to generate an optimization strategy form.

[0140] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0141] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0142] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0143] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0144] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An energy management method based on the Internet of Things and big data, characterized in that: The following steps are involved: Through the Internet of Things, real-time operation data of all energy equipment of preset types in the energy management area are obtained to obtain operation data sets of multiple single energy equipment; Conduct single-unit analysis based on the operating data set of single energy equipment to identify early signs of abnormal energy consumption during the operation of single energy equipment; Perform feature extraction and feature analysis on single energy devices that show early signs of abnormal energy consumption, obtain feature analysis results and transmit them to a pre-trained machine learning model to classify single energy devices into high-quality operating devices and low-quality operating devices; The results of feature analysis in the energy management area, the number of low-quality operating devices classified as individual energy devices, and the total number of all energy devices of preset types are used together for optimization type reasoning to generate an optimization strategy form.

2. The energy management method based on the Internet of Things and big data according to claim 1 is characterized in that: Monomer analysis refers to: In a fixed time window, a time series of operating load data is obtained from the operating data set of the single energy device. The operating load data in the time series is the load value of the single energy device at each time point. In order to obtain the central trend of the load, the average value of all load values ​​in the fixed time window is calculated. In order to measure the degree of load fluctuation, the standard deviation of all load values ​​in the fixed time window is calculated, and then the absolute value of the difference between the average value and the preset benchmark value is divided by the standard deviation. The result of the calculation is the abnormal factor AF.

3. The energy management method based on Internet of Things and big data according to claim 2 is characterized in that: The logic for identifying early signs of abnormal energy consumption during the operation of a single energy device is: The abnormal factor AF is compared with the preset standard threshold. If the abnormal factor AF is greater than the preset standard threshold, an abnormal signal is generated. If the abnormal factor AF is less than or equal to the preset standard threshold, a normal signal is generated. When the normal signal is generated, it is determined that there are no early signs of abnormal energy consumption during the operation of the single energy device. At this time, the single energy device is defaulted to a high-quality operating device. When the abnormal signal is generated, it is determined that there are early signs of abnormal energy consumption during the operation of the single energy device.

4. The energy management method based on the Internet of Things and big data according to claim 3 is characterized in that: Feature extraction of individual energy devices that show early signs of abnormal energy consumption refers to: In view of the operation quality and load adaptability of single energy equipment, we extract the equipment performance characteristic data group that focuses on reflecting the stability of the single energy equipment during operation, and the load adaptability characteristic data group that focuses on the response ability of the single energy equipment to load changes. The equipment performance characteristic data group generates an equipment operation quality index during feature analysis, and the load adaptability characteristic data group generates a load adaptability index during feature analysis.

5. The energy management method based on the Internet of Things and big data according to claim 4 is characterized in that: The logic for obtaining the equipment operation quality index is: The following assessments are made based on the equipment performance characteristic data set: To evaluate load stability, the calculation formula is: N represents the number of data points in the time window, L i represents the load value at the i-th moment, L i-1 Represents the load value at the i-1th moment, Mean L represents the mean value of load data, LSI represents the load stability coefficient; To evaluate the temperature stability, the calculation formula is: Max T is the maximum value of the temperature data in the time window, Min T is the minimum value of the temperature data in the time window, Mean T is the mean of the temperature data in the time window, TSI represents the temperature stability coefficient; To evaluate vibration stability, the calculation formula is: V i is the vibration value at the i-th moment in the time window, Mean V is the mean value of the vibration data in the time window, N is the number of data points in the time window, and VSI is the vibration stability coefficient; Energy efficiency evaluation is performed, and the calculation formula is: Output Work is the workload of a single energy device in the time window, Energy Consumption is the energy consumption of a single energy device in the time window, is the average operating power of a single energy device in the time window, and EEI represents the energy efficiency coefficient; Integrate the evaluation results of each feature data in the equipment performance feature data group into a unified scoring mechanism, specifically: DPQI stands for Device Performance Quality Index.

6. The energy management method based on the Internet of Things and big data according to claim 5 is characterized in that: The logic for obtaining the load adaptability index is: Based on the load adaptability characteristic data set, the evaluation is conducted from three aspects: the response capability of the single energy equipment to load fluctuations, the flexibility of load regulation, and the load balancing: Load Adaptation Response Capability Assessment: L t Represents the load value at time t, Mean L represents the mean value of the load data in the time window, N represents the number of data points in the time window, Max L Indicates the maximum value of the load data in the time window, Min L Indicates the minimum value of the load data in the time window, sign(x) indicates the sign function, returns 1 if x is greater than 0, returns -1 if x is less than 0, otherwise returns 0, LARI indicates the responsiveness value; Load Regulation Sensitivity Evaluation: L n is the load value at the nth moment, N is the number of data points in the time window, Δt is the time interval of load measurement, and LRSI is the regulation sensitivity value; Load adaptability balance evaluation: L k is the load value at the kth moment, LABI represents the load adaptation value; The calculation formula of load adaptability index is: log represents natural logarithm function, and LAI represents load adaptability index.

7. The energy management method based on Internet of Things and big data according to claim 6 is characterized in that: The machine learning model is a convolutional neural network model. The equipment operation quality index and load adaptability index are used as input data of the convolutional neural network model, and the output result of the convolutional neural network model is 0 or 1. When the output result is 0, it indicates that the single energy equipment is classified as low-quality operation equipment, and when the output result is 1, it indicates that the single energy equipment is classified as high-quality operation equipment.

8. The energy management method based on Internet of Things and big data according to claim 7 is characterized in that: Optimizing type inference means: The equipment operation quality index, load adaptability index, and the ratio of the number of single energy equipment classified as low-quality operation equipment to the total number of all energy equipment of preset types are taken as input variables, and the optimization type of energy equipment in the energy management area is taken as the output variable. The input variables are fuzzified and the values ​​of the input variables are converted into fuzzy sets. The output variables are fuzzified and the output variables are converted into fuzzy sets. Fuzzy rules are formulated to describe the adaptation degree of the optimization type of energy equipment under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to obtain the optimization type of each energy equipment in the energy management area. Then the optimization types of all energy equipment are summarized to obtain the optimization strategy form.

9. An energy management system based on the Internet of Things and big data, based on the energy management method based on the Internet of Things and big data according to any one of claims 1 to 8, characterized in that: include: The data collection module obtains the operating data of all energy equipment of preset types in the energy management area in real time through the Internet of Things, and obtains the operating data sets of multiple single energy equipment The single-unit analysis module performs single-unit analysis based on the operating data set of the single energy equipment to identify whether there are early signs of abnormal energy consumption during the operation of the single energy equipment; The feature extraction module extracts features from individual energy devices that show early signs of abnormal energy consumption; The feature analysis module performs feature analysis operations based on the features extracted by the feature extraction module to obtain feature analysis results; The equipment classification module transmits the results of feature analysis to the pre-trained machine learning model to classify single energy equipment into high-quality operating equipment and low-quality operating equipment; The optimization reasoning module performs optimization type reasoning on the results of feature analysis, the number of low-quality operating devices classified as single energy devices, and the total number of all energy devices of preset types in the energy management area to generate an optimization strategy form.

Images