Energy management system and method based on intelligent Internet of Things
Through the smart IoT energy management system, the prediction model is dynamically adjusted by LSTM model and island impact factor, the problems of energy demand prediction deviation and slow island detection in the existing technology are solved, the accuracy and response speed of energy management are improved, and the Efficient use of energy.
Patent Information
- Application Number
- CN202410996303.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing IoT energy management system fails to fully consider weather changes and user behavior patterns when predicting energy demand, resulting in a large deviation from the actual situation. In the siloed island event, there is a lack of effective mechanisms to adjust the prediction model, which reduces the prediction. The accuracy and practicality of the energy network affect the stable operation of the energy network.
Adopt the energy management system based on the smart Internet of Things, including real-time data acquisition module, island detection module, dynamic energy prediction module and intelligent decision-making module, data is collected in real time through IoT sensors, and the comprehensive island detection model is constructed using the LSTM model to extract the island feature vector. , set thresholds for island detection, and dynamically adjust the energy prediction model based on the island detection results, combine historical data, weather and user behavior to predict energy supply and demand, and introduce island impact factors and time decay factors to improve prediction accuracy.
It improves the efficiency, accuracy and response speed of energy management, enhances the accuracy and timeliness of island detection, dynamically adjusts the energy prediction model to predict energy supply and demand more accurately, and achieves efficient energy utilization and conservation.
Smart Images

Figure CN118941026A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of energy management of the Internet of Things, and specifically refers to an energy management system and method based on the smart Internet of Things. Background Art
[0002] With the continuous growth of global energy demand and the increasingly tight energy supply, energy management has become a focus of attention, and the island phenomenon has become an important factor affecting the stability and reliability of energy systems;
[0003] However, the existing energy management system of the Internet of Things still has certain defects. When predicting energy demand, the existing energy management system only relies on historical data without fully considering factors such as weather changes and user behavior patterns, resulting in a large deviation between the prediction results and the actual situation. Especially in emergencies such as island events, there is a lack of effective mechanisms to adjust the prediction model, which reduces the accuracy and practicality of the prediction. When detecting island phenomena, it relies on static threshold detection, which is prone to false alarms or missed alarms, affecting the stable operation of the energy network. In addition, the response speed of island detection is slow, and early warning and emergency response cannot be triggered in time. The real-time monitoring data, prediction results and island detection information are not fully utilized, resulting in the formulated energy management strategy not being optimized enough and unable to be dynamically adjusted to adapt to the real-time changes of the energy network, resulting in energy waste. Therefore, an energy management system and method based on the smart Internet of Things are proposed. Summary of the invention
[0004] The purpose of the present invention is to provide an energy management system and method based on a smart Internet of Things to solve the problems raised in the above background technology.
[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an energy management system and method based on the smart Internet of Things, comprising a real-time data acquisition module, an island detection module, a dynamic energy prediction module, an intelligent decision-making module and a feedback adjustment module;
[0006] The real-time data acquisition module collects and monitors data on energy production and consumption in real time through IoT sensors, and pre-processes and integrates the data through edge computing;
[0007] The island detection module is used to monitor and detect the island phenomenon of energy;
[0008] The dynamic energy prediction module predicts energy supply and demand based on multiple factors and dynamically adjusts the prediction model based on the island detection results;
[0009] The specific operation method of the island detection module includes the following steps:
[0010] A1. Collect data from the energy network in real time through the real-time data collection module;
[0011] A2. Extracting island feature vectors related to island detection based on the collected data;
[0012] A3. Build a comprehensive island detection model and train the comprehensive island detection model through the extracted island feature vector;
[0013] A4. Set an island threshold θ;
[0014] A5. Make a judgment based on the output result of the comprehensive island detection model and the set threshold θ.
[0015] Wherein, A2, assuming that the feature vectors related to island detection include energy flow, pressure, temperature and flow, construct an island feature vector F including the standardized feature vector, and the island feature vector F is expressed as F=[E(t), P(t), T(t), Q(t)], where E(t) represents energy flow, P(t) represents pressure, T(t) represents temperature, and Q(t) represents flow;
[0016] A3, let F t is the island eigenvector at time t, y t To synthesize the output of the island detection model at time t, it is implemented through the LSTM model, and the input gate implementation formula is:
[0017] i t =σ(W xi F t +W hi h t-1 +W ci c t-1 +b i ),
[0018] In the formula, i t represents the output of the input gate at time t, W xi Represents the input data F t The weight matrix to the input gate, W hi Represents the hidden state h at the previous moment t-1 The weight matrix to the input gate, b i Represents the bias vector of the input gate;
[0019] The forget gate implementation formula is:
[0020] f t =σ(W xf F t +W hf h t-1 +W cf c t-1 +b f ),
[0021] In the formula, ft represents the output of the forget gate at time t, W xf Represents the input data F t To the weight matrix of the forget gate, W hf Represents the hidden state h at the previous moment t-1 To the weight matrix of the forget gate, b f Represents the bias vector of the forget gate;
[0022] The cell state implementation formula is:
[0023] c t =f t *c t-1 +i t *tanh(W xc F t +W hc h t-1 +b c ),
[0024] In the formula, c t represents the cell state at time t, c t-1 represents the cell state at the previous moment, tanh(W xc F t +W hc h t-1 +b c ) represents the candidate cell state, W xc Represents the input vector X t to the weight matrix of the candidate cell state, W hc Represents the hidden state h at the previous moment t-1 to the weight matrix of the candidate cell state, b c Bias vector representing the candidate cell state;
[0025] The output gate implementation formula is:
[0026] o t =σ(W xo F t +W ho h t-1 +b o ),
[0027] In the formula, o t represents the output of the forget gate at time t, W xo Represents the input data F t The weight matrix to the output gate, W ho Represents the hidden state h at the previous moment t-1 The weight matrix to the output gate, b o Represents the bias vector of the output gate;
[0028] The hidden state implementation formula is:
[0029] h t =o t *tanh(c t ),
[0030] In the formula, h t represents the hidden state at time t;
[0031] The output formula of island detection is expressed as:
[0032] y t =W y h t +b y ,
[0033] In the formula, y t represents the predicted output at time t, the probability of an island event occurring, W y Represents the hidden state h t The weight matrix to the output layer, b y Represents the bias vector of the output layer.
[0034] Among them, the A5 is implemented by comparing the island detection output result with the island threshold θ, and the formula is:
[0035]
[0036] In the formula, when y t >θ when z t Returns 1, indicating an island warning occurs, otherwise z t Returns 0, indicating no island warning.
[0037] The island detection module builds a comprehensive island detection model and uses advanced algorithms such as LSTM to monitor the island phenomenon in the energy network in real time. This helps to identify and respond to abnormal situations in the energy network in a timely manner.
[0038] The dynamic energy forecasting module is based on the energy demand forecasting model M based on historical data H, weather W and user behavior U. t = 0, no islanding phenomenon is detected and energy demand forecasting is performed normally. The energy forecasting formula is:
[0039] P(t)=M(H,W,U;t),
[0040] In the formula, P(t) represents the energy demand forecast at time t, H represents historical data, W represents weather, and U represents user behavior.
[0041] Among them, the dynamic energy prediction module, when z t= 1, the island phenomenon is detected. An island impact factor I(t) is set to represent the impact of the island phenomenon on energy demand at time t. The island impact factor is based on the size A of the island area, the duration T, and the topological structure D of the energy network. The island impact factor is expressed as I(t) = w A A(t)+w T T(t)+w c D c (C), w A ,w T ,w C Represents the weight coefficient. The island effect gradually weakens over time, and the time attenuation factor δ(t) = e is introduced. -λt , combined with the attenuation factor, is expressed as:
[0042] I(t)=(w A A(t)+w T T(t)+w c D c (C))·δ(t).
[0043] The dynamic energy prediction module dynamically adjusts the prediction model based on the islanding influence factor, and the prediction result of the prediction model adjusted according to the islanding influence factor is expressed as α(t)=1+β·I(t), where α(t) represents the adjustment coefficient and the degree of adjustment of the prediction result of the prediction model by the islanding influence factor;
[0044] The prediction model adjusted according to the island impact factor is expressed as:
[0045] P′(t)=α(t)·M(H, W, U; t).
[0046] Among them, the intelligent decision-making module formulates the optimal energy management strategy and dynamically adjusts the energy production, storage and distribution plan based on the real-time monitoring data, the prediction results provided by the dynamic energy prediction module and the island detection results.
[0047] The dynamic energy forecasting module predicts energy supply and demand based on historical data, weather, user behavior, island detection results and other factors, and dynamically adjusts the forecasting model. This forecasting method is closer to the actual situation and can more accurately predict future energy demand.
[0048] The feedback adjustment module is used to monitor the execution status and effect of the intelligent decision-making module and collect feedback information, and dynamically adjust and optimize the strategy according to the actual operation status and changes in the island status.
[0049] Among them, the energy management method based on the smart Internet of Things is characterized by comprising the following steps;
[0050] S1. Collect energy production and consumption data in real time based on IoT sensors, and pre-process and integrate the data through edge computing;
[0051] S2. By extracting feature vectors related to island detection, the model is trained to identify potential island phenomena, and a threshold is set. When the probability of an island event output by the model exceeds the set threshold, an island warning is triggered;
[0052] S3, when no islanding phenomenon is detected, the energy demand forecasting model based on historical data, weather and user behavior performs normal forecasting;
[0053] S4. When an island phenomenon is detected, an island impact factor is introduced. Based on the size, duration and network topology of the island area, the prediction model is adjusted to reflect the island impact.
[0054] S5. Combining real-time monitoring data, dynamic energy forecast results and island detection warning, the intelligent decision-making module formulates the optimal energy management strategy;
[0055] S6. Monitor the execution status of the intelligent decision-making module, collect feedback information, and dynamically adjust and optimize the energy management strategy according to the actual operation status and changes in the island status.
[0056] Compared with the prior art, the present invention has the following beneficial effects:
[0057] 1. In order to solve the problems of real-time performance, prediction accuracy, and effective detection of island phenomena in existing energy management systems and improve the efficiency, accuracy, and response speed of energy management, the present invention uses a comprehensive island detection model of LSTM to extract island feature vectors for training, and sets thresholds for island detection and early warning, thereby improving the accuracy and timeliness of island detection. The dynamic energy prediction model is based on historical data, weather, user behavior, and island detection results, and a dynamic energy demand prediction model is constructed. The island influencing factor and time attenuation factor are introduced for dynamic adjustment, thereby improving the accuracy and real-time performance of energy demand prediction.
[0058] 2. The present invention can monitor and detect the island phenomenon in the energy network in real time through the island detection module, and train the island feature vector through the comprehensive island detection model, thereby improving the accuracy and timeliness of island detection. At the same time, the energy prediction model is dynamically adjusted according to the island detection results to more accurately predict the energy supply and demand situation and take corresponding countermeasures;
[0059] 3. The present invention uses a dynamic energy prediction module to predict energy supply and demand based on historical data, weather, user behavior and other factors, and considers the impact of islanding on energy demand. By introducing islanding influence factors and time attenuation factors, the prediction results are more in line with the actual situation, thereby improving the effectiveness of energy management;
[0060] 4. The present invention can formulate the optimal energy management strategy according to the real-time monitoring data, prediction results and island detection results through the intelligent decision-making module, dynamically adjust the energy production, storage and distribution plan, and realize the efficient utilization and conservation of energy. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 This is an operation flow chart of the energy management method based on the smart Internet of Things of the present invention;
[0062] Figure 2 This is a schematic diagram of the structure of an energy management system based on the smart Internet of Things of the present invention;
[0063] Figure 3 This is an operation flow chart of the island detection module of the energy management system based on the smart Internet of Things of the present invention;
[0064] Figure 4 This is an operation flow chart of the dynamic energy prediction module of the energy management system based on the smart Internet of Things of the present invention. DETAILED DESCRIPTION
[0065] 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.
[0066] Example
[0067] See also Figure 1-Figure 4 As shown, the present invention provides a technical solution: including a real-time data acquisition module, an island detection module, a dynamic energy prediction module, an intelligent decision-making module and a feedback adjustment module;
[0068] The real-time data acquisition module collects and monitors data on energy production and consumption in real time through IoT sensors, and pre-processes and integrates the data through edge computing;
[0069] The island detection module is used to monitor and detect the island phenomenon of energy;
[0070] The dynamic energy prediction module predicts energy supply and demand based on multiple factors and dynamically adjusts the prediction model based on the island detection results;
[0071] The specific operation method of the island detection module includes the following steps:
[0072] A1. Collect data from the energy network in real time through the real-time data collection module;
[0073] A2. Extracting island feature vectors related to island detection based on the collected data;
[0074] A3. Build a comprehensive island detection model and train the comprehensive island detection model through the extracted island feature vector;
[0075] A4. Set an island threshold θ;
[0076] A5. Make a judgment based on the output result of the comprehensive island detection model and the set threshold θ.
[0077] Wherein, A2, assuming that the feature vectors related to island detection include energy flow, pressure, temperature and flow, construct an island feature vector F including the standardized feature vector, and the island feature vector F is expressed as F=[E(t), P(t), T(t), Q(t)], where E(t) represents energy flow, P(t) represents pressure, T(t) represents temperature, and Q(t) represents flow;
[0078] A3, let F t is the island eigenvector at time t, y t To synthesize the output of the island detection model at time t, it is implemented through the LSTM model, and the input gate implementation formula is:
[0079] i t =σ(W xi F t +W hi h t-1 +W ci c t-1 +b i ),
[0080] In the formula, i t represents the output of the input gate at time t, W xi Represents the input data F t The weight matrix to the input gate, W hi Represents the hidden state h at the previous moment t-1 The weight matrix to the input gate, b i Represents the bias vector of the input gate;
[0081] The forget gate implementation formula is:
[0082] f t =σ(W xf F t +W hf h t-1 +W cf c t-1 +b f ),
[0083] In the formula, ft represents the output of the forget gate at time t, W xf Represents the input data F t To the weight matrix of the forget gate, W hf Represents the hidden state h at the previous moment t-1 To the weight matrix of the forget gate, b f Represents the bias vector of the forget gate;
[0084] The cell state implementation formula is:
[0085] c t =f t *c t-1 +i t *tanh(W xc F t +W hc h t-1 +b c ),
[0086] In the formula, c t represents the cell state at time t, c t-1 represents the cell state at the previous moment, tanh(W xc F t +W hc h t-1 +b c ) represents the candidate cell state, W xc Represents the input vector X t to the weight matrix of the candidate cell state, W hc Represents the hidden state h at the previous moment t-1 to the weight matrix of the candidate cell state, b c Bias vector representing the candidate cell state;
[0087] The output gate implementation formula is:
[0088] o t =σ(W xo F t +W ho h t-1 +b o ),
[0089] In the formula, o t represents the output of the forget gate at time t, W xo Represents the input data F t The weight matrix to the output gate, W ho Represents the hidden state h at the previous moment t-1 The weight matrix to the output gate, b o Represents the bias vector of the output gate;
[0090] The hidden state implementation formula is:
[0091] h t =o t *tanh(c t ),
[0092] In the formula, h t represents the hidden state at time t;
[0093] The output formula of island detection is expressed as:
[0094] y t =W y h t +b y ,
[0095] In the formula, y t represents the predicted output at time t, the probability of an island event occurring, W y Represents the hidden state h t The weight matrix to the output layer, b y Represents the bias vector of the output layer.
[0096] Among them, the A5 is implemented by comparing the island detection output result with the island threshold θ, and the formula is:
[0097]
[0098] In the formula, when y t >θ when z t Returns 1, indicating an island warning occurs, otherwise z t Returns 0, indicating no island warning.
[0099] The island detection module builds a comprehensive island detection model and uses advanced algorithms such as LSTM to monitor the island phenomenon in the energy network in real time. This helps to identify and respond to abnormal situations in the energy network in a timely manner.
[0100] The dynamic energy forecasting module is based on the energy demand forecasting model M based on historical data H, weather W and user behavior U. t = 0, no islanding phenomenon is detected and energy demand forecasting is performed normally. The energy forecasting formula is:
[0101] P(t)=M(H,W,U;t),
[0102] In the formula, P(t) represents the energy demand forecast at time t, H represents historical data, W represents weather, and U represents user behavior.
[0103] Among them, the dynamic energy prediction module, when z t=1, the island phenomenon is detected. An island impact factor I(t) is set to represent the impact of the island phenomenon on energy demand at time t. The island impact factor is based on the size of the island area, the duration, and the topological structure of the energy network. The island impact factor is expressed as I(t)=w A A(t)+w T T(t)+w c D c (C), w A ,w T ,w C Represents the weight coefficient. The island effect gradually weakens over time, and the time attenuation factor δ(t) = e is introduced. -λt , combined with the attenuation factor, is expressed as:
[0104] I(t)=(w A A(t)+w T T(t)+w c D c (C))·δ(t).
[0105] The dynamic energy prediction module dynamically adjusts the prediction model based on the islanding influence factor, and the prediction result of the prediction model adjusted according to the islanding influence factor is expressed as α(t)=1+β·I(t), where α(t) represents the adjustment coefficient and the degree of adjustment of the prediction result of the prediction model by the islanding influence factor;
[0106] The prediction model adjusted according to the island impact factor is expressed as:
[0107] P′(t)=α(t)·M(H, W, U; t).
[0108] Among them, the intelligent decision-making module formulates the optimal energy management strategy and dynamically adjusts the energy production, storage and distribution plan based on the real-time monitoring data, the prediction results provided by the dynamic energy prediction module and the island detection results.
[0109] The dynamic energy forecasting module predicts energy supply and demand based on historical data, weather, user behavior, island detection results and other factors, and dynamically adjusts the forecasting model. This forecasting method is closer to the actual situation and can more accurately predict future energy demand.
[0110] The feedback adjustment module is used to monitor the execution status and effect of the intelligent decision-making module and collect feedback information, and dynamically adjust and optimize the strategy according to the actual operation status and changes in the island status.
[0111] Among them, the energy management method based on the smart Internet of Things is characterized by comprising the following steps;
[0112] S1. Collect energy production and consumption data in real time based on IoT sensors, and pre-process and integrate the data through edge computing;
[0113] S2. By extracting feature vectors related to island detection, the model is trained to identify potential island phenomena, and a threshold is set. When the probability of an island event output by the model exceeds the set threshold, an island warning is triggered;
[0114] S3, when no islanding phenomenon is detected, the energy demand forecasting model based on historical data, weather and user behavior performs normal forecasting;
[0115] S4. When an island phenomenon is detected, an island impact factor is introduced. Based on the size, duration and network topology of the island area, the prediction model is adjusted to reflect the island impact.
[0116] S5. Combining real-time monitoring data, dynamic energy forecast results and island detection warning, the intelligent decision-making module formulates the optimal energy management strategy;
[0117] S6. Monitor the execution status of the intelligent decision-making module, collect feedback information, and dynamically adjust and optimize the energy management strategy according to the actual operation status and changes in the island status.
[0118] In this example, specifically: A2, assuming that the feature vectors related to island detection include energy flow, pressure, temperature and flow, construct an island feature vector F including the standardized feature vector, and the island feature vector F is expressed as F = [E(t), P(t), T(t), Q(t)], where E(t) represents energy flow, P(t) represents pressure, T(t) represents temperature, and Q(t) represents flow;
[0119] A3, let F t is the island eigenvector at time t, y t To synthesize the output of the island detection model at time t, it is implemented through the LSTM model, and the input gate implementation formula is:
[0120] i t =σ(W xi F t +W hi h t-1 +W ci c t-1 +b i ),
[0121] In the formula, i t represents the output of the input gate at time t, W xi Represents the input data F t The weight matrix to the input gate, W hi Represents the hidden state h at the previous moment t-1The weight matrix to the input gate, b i Represents the bias vector of the input gate;
[0122] The forget gate implementation formula is:
[0123] f t =σ(W xf F t +W hf h t-1 +W cf c t-1 +b f ),
[0124] In the formula, f t represents the output of the forget gate at time t, W xf Represents the input data F t To the weight matrix of the forget gate, W hf Represents the hidden state h at the previous moment t-1 To the weight matrix of the forget gate, b f Represents the bias vector of the forget gate;
[0125] The cell state implementation formula is:
[0126] c t =f t *c t-1 +i t *tanh(W xc F t +W hc h t-1 +b c ),
[0127] In the formula, c t represents the cell state at time t, c t-1 represents the cell state at the previous moment, tanh(W xc F t +W hc h t-1 +b c ) represents the candidate cell state, W xc Represents the input vector X t to the weight matrix of the candidate cell state, W hc Represents the hidden state h at the previous moment t-1 to the weight matrix of the candidate cell state, b c Bias vector representing the candidate cell state;
[0128] The output gate implementation formula is:
[0129] o t =σ(W xo F t +Who h t-1 +b o ),
[0130] In the formula, o t represents the output of the forget gate at time t, W xo Represents the input data F t The weight matrix to the output gate, W ho Represents the hidden state h at the previous moment t-1 The weight matrix to the output gate, b o Represents the bias vector of the output gate;
[0131] The hidden state implementation formula is:
[0132] h t =o t *tanh(c t ),
[0133] In the formula, h t represents the hidden state at time t;
[0134] The island detection output formula is expressed as:
[0135] y t =W y h t +b y ,
[0136] In the formula, y t represents the predicted output at time t, the probability of an island event occurring, W y Represents the hidden state h t The weight matrix to the output layer, b y Represents the bias vector of the output layer.
[0137] In this example, specifically: the dynamic energy forecasting module assumes an energy demand forecasting model M based on historical data H, weather W and user behavior U. t = 0, no islanding phenomenon is detected and energy demand forecasting is performed normally. The energy forecasting formula is:
[0138] P(t)=M(H,W,U;t),
[0139] In the formula, P(t) represents the energy demand forecast at time t, H represents historical data, W represents weather, and U represents user behavior.
[0140] Among them, the dynamic energy prediction module, when z t=1, the island phenomenon is detected. An island impact factor I(t) is set to represent the impact of the island phenomenon on energy demand at time t. The island impact factor is based on the size of the island area, the duration, and the topological structure of the energy network. The island impact factor is expressed as I(t)=w A A(t)+w T T(t)+w c D c (C), w A ,w T ,w C Represents the weight coefficient. The island effect gradually weakens over time, and the time attenuation factor δ(t) = e is introduced. - λt, combined with the attenuation factor, is expressed as:
[0141] I(t)=(w A A(t)+w T T(t)+w c D c (C))·δ(t).
[0142] The dynamic energy prediction module dynamically adjusts the prediction model based on the islanding influence factor, and the prediction result of the prediction model adjusted according to the islanding influence factor is expressed as α(t)=1+β·I(t), where α(t) represents the adjustment coefficient and the degree of adjustment of the prediction result of the prediction model by the islanding influence factor;
[0143] The prediction model adjusted according to the island impact factor is expressed as:
[0144] P′(t)=α(t)·M(H, W, U; t).
[0145] Among them, the intelligent decision-making module formulates the optimal energy management strategy and dynamically adjusts the energy production, storage and distribution plan based on the real-time monitoring data, the prediction results provided by the dynamic energy prediction module and the island detection results.
[0146] Working principle: IoT sensors are used to collect and monitor data on energy production and consumption in real time, and edge computing is used for preliminary preprocessing and fusion;
[0147] The island detection module extracts feature vectors related to island detection from the collected data and performs standardization. It uses LSTM to build and train a comprehensive island detection model. It trains feature vectors and sets an island threshold. It determines whether an island warning occurs based on the comparison between the model output and the threshold. When no island phenomenon is detected, the prediction model M is used to predict energy demand based on historical data, weather and user behavior. When an island phenomenon is detected, an island impact factor is introduced to indicate the degree of impact of the island phenomenon on energy demand. The island impact factor is based on the size of the island area, duration and energy network. The topological structure is constructed, and the time attenuation factor is considered to reflect the change of the islanding effect over time. The output result of the prediction model is adjusted according to the islanding effect factor. The intelligent decision-making module comprehensively considers multiple factors based on the real-time monitoring data, the prediction results of the dynamic energy prediction module and the islanding detection results to formulate the optimal energy management strategy. The energy production, storage and distribution plans are dynamically adjusted according to the strategy to cope with the changes in energy supply and demand and the islanding phenomenon. The feedback adjustment module is responsible for monitoring the status and effect of the execution of the intelligent decision-making module, and collecting feedback information. According to the actual operation situation and the changes in the islanding status, the energy management strategy is dynamically adjusted and optimized.
[0148] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
[0149] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.
Claims
1. Energy management system based on smart Internet of Things, characterized by: It includes real-time data acquisition module, island detection module, dynamic energy prediction module, intelligent decision-making module and feedback adjustment module; The real-time data acquisition module collects and monitors data on energy production and consumption in real time through IoT sensors, and pre-processes and integrates the data through edge computing; The island detection module is used to monitor and detect the island phenomenon of energy; The dynamic energy prediction module predicts energy supply and demand based on multiple factors and dynamically adjusts the prediction model based on the island detection results; The specific operation method of the island detection module includes the following steps: A1. Collect data from the energy network in real time through the real-time data collection module; A2. Extracting island feature vectors related to island detection based on the collected data; A3. Build a comprehensive island detection model and train the comprehensive island detection model through the extracted island feature vector; A4. Set an island threshold θ; A5. Make a judgment based on the output result of the comprehensive island detection model and the set threshold θ.
2. The energy management system based on the smart Internet of Things according to claim 1 is characterized in that: In A2, assuming that the feature vectors related to island detection include energy flow, pressure, temperature and flow, an island feature vector F is constructed including the standardized feature vector, and the island feature vector F is expressed as F=[E(t), P(t), T(t), Q(t)], where E(t) represents energy flow, P(t) represents pressure, T(t) represents temperature, and Q(t) represents flow; A3, let F t is the island eigenvector at time t, y t To synthesize the output of the island detection model at time t, the LSTM model is used to implement the input gate implementation formula: i t =σ(W xi F t +W hi h t-1 +W ci c t-1 +b i ), In the formula, i t represents the output of the input gate at time t, W xi Represents the input data F t The weight matrix to the input gate, W hi Represents the hidden state h at the previous moment t-1 The weight matrix to the input gate, b i Represents the bias vector of the input gate; The forget gate implementation formula is: f t =σ(W xf F t +W hf h t-1 +W cf c t-1 +b f ), In the formula, f t represents the output of the forget gate at time t, W xf Represents the input data F t To the weight matrix of the forget gate, W hf Represents the hidden state h at the previous moment t-1 To the weight matrix of the forget gate, b f Represents the bias vector of the forget gate; The cell state implementation formula is: c t =f t *c t-1 +i t *tanh(W xc F t +W hc h t-1 +b c ), In the formula, c t represents the cell state at time t, c t-1 represents the cell state at the previous moment, tanh(W xc F t +W hc h t-1 +b c ) represents the candidate cell state, W xc Represents the input vector X t to the weight matrix of the candidate cell state, W hc Represents the hidden state h at the previous moment t-1 to the weight matrix of the candidate cell state, b c Bias vector representing the candidate cell state; The output gate implementation formula is: o t =σ(W xo F t +W ho h t-1 +b o ), In the formula, o t represents the output of the forget gate at time t, W xo Represents the input data F t The weight matrix to the output gate, W ho Represents the hidden state h at the previous moment t-1 The weight matrix to the output gate, b o Represents the bias vector of the output gate; The hidden state implementation formula is: h t =o t *tanh(c t ), In the formula, h t represents the hidden state at time t; The output formula of island detection is expressed as: y t =W y h t +b y , In the formula, y t represents the predicted output at time t, the probability of an island event occurring, W y Represents the hidden state h t The weight matrix to the output layer, b y Represents the bias vector of the output layer.
3. The energy management system based on the smart Internet of Things according to claim 2 is characterized in that: A5, based on the island detection output result and the island threshold θ comparison detection, the implementation formula is: In the formula, when y t >θ when z t Returns 1, indicating an island warning occurs, otherwise z t Returns 0, indicating no island warning.
4. The energy management system based on the smart Internet of Things according to claim 1 is characterized in that: The dynamic energy forecasting module assumes an energy demand forecasting model M based on historical data H, weather W and user behavior U. t = 0, no islanding phenomenon is detected and energy demand forecasting is performed normally. The energy forecasting formula is: P(t)=M(H,W,U;t), In the formula, P(t) represents the energy demand forecast at time t, H represents historical data, W represents weather, and U represents user behavior.
5. The energy management system based on the smart Internet of Things according to claim 4 is characterized in that: The dynamic energy prediction module, when z t = 1, the island phenomenon is detected. An island impact factor I(t) is set to represent the impact of the island phenomenon on energy demand at time t. The island impact factor is based on the size A of the island area, the duration T, and the topological structure D of the energy network. The island impact factor is expressed as I(t) = w A A(t)+w T T(t)+w c D c (C), w A ,w T ,w C Represents the weight coefficient. The island effect gradually weakens over time, and the time attenuation factor δ(t) = e is introduced. -λt , combined with the attenuation factor, is expressed as: I(t)=(w A A(t)+w T T(t)+w c D c (C))·δ(t)。 6. The energy management system based on the smart Internet of Things according to claim 5 is characterized in that: The dynamic energy prediction module dynamically adjusts the prediction model based on the islanding influence factor, and the prediction result of the prediction model adjusted according to the islanding influence factor is expressed as α(t)=1+β·I(t), where α(t) represents the adjustment coefficient and the degree of adjustment of the prediction result of the prediction model by the islanding influence factor; The prediction model adjusted according to the island impact factor is expressed as: P'(t)=α(t)·M(H, W, U; t).
7. The energy management system based on the smart Internet of Things according to claim 1 is characterized in that: The intelligent decision-making module formulates the optimal energy management strategy and dynamically adjusts the energy production, storage and distribution plan based on the real-time monitoring data, the prediction results provided by the dynamic energy prediction module and the island detection results.
8. The energy management system based on the smart Internet of Things according to claim 1 is characterized in that: The feedback adjustment module is used to monitor the execution status and effect of the intelligent decision-making module and collect feedback information, and dynamically adjust and optimize the strategy according to the actual operation situation and changes in the island status.
9. The energy management method based on the smart Internet of Things is characterized by: The steps include: S1. Collect energy production and consumption data in real time based on IoT sensors, and pre-process and integrate the data through edge computing; S2. By extracting feature vectors related to island detection, the model is trained to identify potential island phenomena, and a threshold is set. When the probability of an island event output by the model exceeds the set threshold, an island warning is triggered; S3, when no islanding phenomenon is detected, the energy demand forecasting model based on historical data, weather and user behavior performs normal forecasting; S4. When an island phenomenon is detected, an island impact factor is introduced. Based on the size, duration and network topology of the island area, the prediction model is adjusted to reflect the island impact. S5. Combining real-time monitoring data, dynamic energy forecast results and island detection warning, the intelligent decision-making module formulates the optimal energy management strategy; S6. Monitor the execution status of the intelligent decision-making module, collect feedback information, and dynamically adjust and optimize the energy management strategy according to the actual operation status and changes in the island status.
Citation Information
Patent Citations
Energy management platform based on Internet of Things
CN117933660A
Energy storage converter control method and system of adaptive multi-parameter island detection algorithm
CN118214047A
Cited By
Intelligent computing center energy consumption optimization method and system based on Internet of Things
CN121257862A
Energy Consumption Optimization Method and System for Smart Computing Centers Based on the Internet of Things
CN121257862B
