Building big data based intelligent building energy saving scheme automatic generation system

Abstract

The application discloses a building big data-based intelligent building energy-saving scheme automatic generation system and relates to the technical field of building big data; two reasoning models are constructed, a first reasoning model is constructed based on environmental data, and a second reasoning model is constructed based on personnel distribution and time data; the first and second reasoning models are complementary in performance consumption prediction; different reasoning models are selected and enabled comprehensively, the system can dynamically adjust energy efficiency prediction according to actual data quality and integrity, and the accuracy of prediction is improved; through comparison with historical data, abnormal energy consumption can be found and corrected in time, and specific energy consumption control signals are generated according to an energy-saving scheme adjustment matrix. A control execution module further controls various energy-using equipment, such as air conditioners, lighting and ventilation equipment, according to the signals, energy use of the building is optimized, and intelligent energy-saving management is realized.

Description

Technical Field

[0001] This invention relates to the field of building big data technology, and in particular to an automatic generation system for smart building energy-saving solutions based on building big data. Background Technology

[0002] With the continued growth of global energy demand and increasing focus on sustainable development, building energy efficiency management has become a crucial area for energy conservation and emission reduction. Especially in the process of modern urbanization, building energy consumption accounts for a significant proportion of total energy consumption. How to effectively improve building energy efficiency and reduce energy waste is a pressing issue for society today. To achieve intelligent building energy-saving management, more and more buildings are adopting building big data technology to monitor and optimize energy use within the building in real time. Building big data refers to multi-dimensional data collected through technologies such as sensors, IoT devices, and smart meters, covering aspects such as the building environment, equipment operation, occupant distribution, time, and energy consumption.

[0003] Although building big data and intelligent energy-saving technologies have been widely applied in some fields, current technical solutions still have the following problems and shortcomings:

[0004] First, most current building energy efficiency systems rely on sensors and smart devices to acquire various environmental and energy consumption data both inside and outside the building. However, sensors may malfunction, have missing data, or experience noise issues, leading to incomplete or inaccurate data. Furthermore, most existing systems fail to adequately address how to effectively clean, filter, and process this abnormal data, resulting in decreased data quality and consequently affecting the accuracy of energy-saving decisions.

[0005] Secondly, many existing building energy-saving solutions use models based on environmental data (such as temperature and humidity) to predict energy consumption. However, these models often overlook other key factors, such as occupant distribution and temporal variations, which have a significant impact on actual energy efficiency management. In existing technologies, most energy efficiency prediction models are static and lack real-time adaptability to dynamic factors, failing to accurately reflect the inherent energy consumption patterns of buildings, leading to inefficient and unstable energy-saving effects.

[0006] To address the aforementioned issues, it is necessary to propose an automatic generation system for smart building energy-saving solutions based on building big data. Summary of the Invention

[0007] The purpose of this invention is to solve the problems existing in the background technology and to propose an automatic generation system for smart building energy-saving solutions based on building big data.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] The intelligent building energy-saving solution automatic generation system based on building big data includes a data acquisition module, a data modeling module, a model matching module, an energy efficiency analysis module, and a control execution module.

[0010] The data acquisition module is responsible for acquiring environmental and energy consumption data in real time from energy-consuming devices and sensors inside and outside the building on each floor and in each room. It then cleans, filters, and normalizes this data to obtain environmental and energy consumption data matrices. The module also accesses a mobile client in real time to obtain personnel distribution data, resulting in a personnel distribution data vector. Furthermore, it accesses the mobile internet in real time to obtain a time data vector. The module updates the environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector at preset time intervals.

[0011] At preset time intervals, the system accesses sensors inside and outside the building on each floor and in each room to obtain environmental data, including the building's internal temperature, external temperature, internal humidity, external humidity, internal carbon dioxide concentration, and surface light intensity.

[0012] At preset time intervals, access energy-consuming devices inside and outside the building to obtain energy consumption data, including the power of air conditioning, lighting and ventilation equipment in each room on each floor.

[0013] At preset time intervals, the system accesses the mobile client to obtain personnel distribution data, including the floor number and room number of each person.

[0014] At preset time intervals, access the mobile internet to obtain time data, including the current month, date, hour, and a workday identifier indicating whether the current day is a workday.

[0015] In a preferred embodiment of the present invention, environmental and energy consumption data are cleaned, filtered, and normalized. Kalman filtering is then used to recursively estimate all environmental and energy consumption data. For each environmental and energy consumption data point, the actual value at the current moment is estimated by combining the actual observed value at the current moment with the predicted value from the previous preset time interval.

[0016] In a preferred embodiment of the present invention, all Kalman filter results are normalized to obtain the mean and standard deviation of the Kalman filter results for the following data at all timestamps: building interior temperature, building exterior temperature, building interior humidity, building exterior humidity, building interior carbon dioxide concentration, building surface illuminance, and the power of air conditioning, lighting, and ventilation equipment in each room on each floor. Each Kalman filter result is then subtracted from its corresponding mean and divided by its corresponding standard deviation to obtain the normalized results of the environmental and energy consumption data at each timestamp. This includes the normalized results for the building interior temperature, building exterior temperature, building interior humidity, building exterior humidity, building interior carbon dioxide concentration, building surface illuminance, air conditioning power, lighting equipment power, and ventilation equipment power in each room on each floor.

[0017] As a preferred embodiment of the present invention, the process of generating an environmental data matrix, an energy consumption data matrix, and a personnel distribution data vector is as follows:

[0018] An environmental data matrix for each time stamp is generated based on the normalized results of the building's internal temperature, external temperature, internal humidity, external humidity, internal carbon dioxide concentration, and surface light intensity in each room on each floor.

[0019] Based on the normalized results of the air conditioning power, lighting power, and ventilation power in each room on each floor, an energy consumption data matrix for each timestamp is generated.

[0020] Generate a personnel distribution data vector for each timestamp based on the number of people in each room on each floor:

[0021] Generate time data vectors for each timestamp based on the month, date, hour, and weekday identifier of the current time.

[0022] In a preferred embodiment of the present invention, the elements corresponding to missing data values ​​in the environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector are marked as data error symbols.

[0023] The generated environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector are sent to the data modeling module, energy efficiency analysis module, and model matching module.

[0024] The data modeling module records historical data for all environmental data matrices, energy consumption data matrices, personnel distribution data vectors, and time data vectors. It extracts correlation features from the environmental and energy consumption data matrices to establish a first inference model tracing the data from the environmental data matrix to the energy consumption data matrix. Similarly, it extracts correlation features from the personnel distribution data vector, time data vector, and energy consumption data matrix to establish a second inference model tracing the data from the personnel distribution data vector and time data vector to the energy consumption data matrix.

[0025] As a preferred embodiment of the present invention, a first inference model is established, specifically a support vector machine, including an input layer, a mapping layer, an optimization layer, and an output layer.

[0026] The input layer includes several nodes, each used to receive elements from the environmental data matrix, specifically the normalized results of environmental data from each floor and each room.

[0027] The mapping layer consists of several nodes. Each node receives two elements from its corresponding environmental data matrix and is responsible for mapping the normalized results of the environmental data of each floor and each room to a high-dimensional space through a kernel function so as to facilitate linear separation.

[0028] The optimization layer includes a preset number of nodes, which are responsible for finding the optimal separating hyperplane using an optimization algorithm based on Lagrange multipliers, and recording the normal vector and the hyperplane bias vector of the hyperplane obtained through machine learning training.

[0029] The output layer is responsible for outputting the inference results of each corresponding energy consumption data, including the final inference results of air conditioning power, lighting equipment power, and ventilation equipment power.

[0030] As a preferred embodiment of the present invention, a second inference model is established, specifically a support vector machine, including an input layer, a mapping layer, an optimization layer, and an output layer.

[0031] The input layer consists of several nodes, divided into a first part of nodes and a second part of nodes. The first part of nodes receives elements from the personnel distribution data vector, specifically the number of people in each room on each floor. The second part of nodes receives elements from the time data vector, specifically the month, date, hour, and weekday identifier of the current time.

[0032] The mapping layer consists of several nodes. Each node is responsible for receiving elements from a personnel distribution data vector and an element from a time data vector, and mapping them to a high-dimensional space for linear separation.

[0033] The optimization layer includes a preset number of nodes, which are responsible for finding the optimal separating hyperplane using an optimization algorithm based on Lagrange multipliers, and storing the normal vector and hyperplane bias vector of the hyperplane of the second inference model obtained through machine learning training.

[0034] The output layer is responsible for outputting the inference results of each corresponding energy consumption data, including the final inference results of air conditioning power, lighting equipment power, and ventilation equipment power.

[0035] The model matching module acquires the environmental data matrix and personnel distribution data vector generated in real time from the previous preset time interval and performs data integrity analysis on them respectively. Based on the data integrity analysis results, it selects to activate either the first inference model or the second inference model to obtain the final inference result of the energy consumption data matrix.

[0036] The environmental data matrix is ​​subjected to data integrity analysis, and the specific process is as follows:

[0037] Obtain the number of error symbols in the environmental data matrix generated at the previous preset time interval, and count the number of each floor. The number of corresponding data error symbols is used to obtain the first integrity verification symbol for each floor. If the number of data error symbols for a certain floor in the environmental data matrix exceeds a preset threshold, the environmental data for that floor is determined to have failed data integrity verification. If the number of floors in the environmental data matrix that have failed data integrity verification exceeds 20% of the total number of floors, the environmental data matrix generated in the previous preset time interval is determined to have failed data integrity analysis.

[0038] The data integrity analysis of the personnel distribution data vector is performed as follows:

[0039] The system retrieves the number of error symbols ("error") in the personnel distribution data vector and time data vector generated in the previous preset time interval. It then counts the number of error symbols corresponding to each floor in the personnel distribution data vector to obtain the second integrity verification symbol for each floor. If the number of error symbols corresponding to a certain floor in the personnel distribution data vector exceeds a preset threshold, the personnel distribution data for that floor is deemed to have failed data integrity verification. If the number of floors that have failed data integrity verification in the personnel distribution data vector exceeds 20% of the total number of floors, the personnel distribution data vector generated in the previous preset time interval is deemed to have failed data integrity analysis.

[0040] As a preferred embodiment of the present invention, the first inference model or the second inference model is selected to be used based on the data integrity analysis results, and the specific process is as follows:

[0041] If the environmental data matrix generated in the previous preset time interval is found to pass the data integrity analysis, but the personnel distribution data vector generated in the previous preset time interval fails the data integrity analysis, then the first inference model is activated.

[0042] If the environmental data matrix generated in the previous preset time interval fails the data integrity analysis, while the personnel distribution data vector generated in the previous preset time interval passes the data integrity analysis, then the second inference model is activated.

[0043] If both the environmental data matrix and the personnel distribution data vector generated in the previous preset time interval are found to have passed the data integrity analysis, then the summation results of the first integrity validator and the second integrity validator are further compared. If the summation result of the first integrity validator is found to be greater than the summation result of the second integrity validator, then the second inference model is activated; if the summation result of the second integrity validator is found to be greater than the summation result of the first integrity validator, then the first inference model is activated.

[0044] If it is found that neither the environmental data matrix nor the personnel distribution data vector generated in the previous preset time interval has passed the data integrity analysis, then the first inference model and the second inference model are activated simultaneously. The average values ​​of the inference results of the air conditioning power, lighting equipment power and ventilation equipment power output by the first inference model and the second inference model are calculated respectively to obtain the final inference results of the air conditioning power, lighting equipment power and ventilation equipment power.

[0045] The final inference results of the air conditioning power, lighting equipment power, and ventilation equipment power are sent to the energy efficiency analysis module.

[0046] The energy efficiency analysis module compares the energy consumption data matrix generated at the previous preset time interval with the final inference result of the energy consumption data matrix to obtain an energy consumption data adjustment matrix containing energy-saving scheme information.

[0047] By subtracting the final inference results of the air conditioning power, lighting equipment power, and ventilation equipment power from each element in the energy consumption data matrix generated at the previous preset time interval, an energy consumption data adjustment matrix containing energy-saving scheme information is obtained. This energy consumption data adjustment matrix is ​​then sent to the control execution module.

[0048] The control execution module adjusts the matrix based on energy consumption data to generate energy consumption control signals, which are then distributed to each energy-consuming device to control and execute the energy-saving scheme information.

[0049] Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the power adjustment amount of the air conditioning power of each room on each floor;

[0050] Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the power adjustment amount of the lighting equipment power in each room on each floor;

[0051] Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the power adjustment amount of the ventilation equipment power in each room on each floor.

[0052] The power adjustment of air conditioning, lighting, and ventilation equipment in each room on each floor is used as the adjustment target. PID control guides all energy-consuming equipment to adjust their power.

[0053] The beneficial effects of this invention are as follows:

[0054] This invention constructs two inference models: a first inference model (based on environmental data) and a second inference model (based on personnel distribution and time data) to provide energy consumption predictions. These inference models help the system identify energy consumption patterns under different conditions and automatically adjust energy-saving strategies. By comprehensively selecting and enabling different inference models, the system can dynamically adjust energy efficiency predictions based on the quality and completeness of actual data, improving prediction accuracy. By comparing with historical data, abnormal energy consumption is promptly detected and corrected, and specific energy consumption control signals are generated based on the adjustment matrix according to the energy-saving plan. The control execution module further uses these signals to precisely control various energy-consuming devices, such as air conditioning, lighting, and ventilation equipment, optimizing building energy use and achieving intelligent energy-saving management. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the system architecture of the intelligent building energy-saving solution automatic generation system based on building big data provided in the embodiments of the present invention;

[0056] Figure 2 This is a topological structure diagram of the first inference model provided in this embodiment of the invention;

[0057] Figure 3 This is a topological diagram of the second inference model provided in this embodiment of the invention;

[0058] Figure 4 This is a schematic diagram of environmental data provided in an embodiment of the present invention;

[0059] Figure 5 This is a schematic diagram of power adjustment provided in an embodiment of the present invention;

[0060] Figure 6 This is an energy consumption classification diagram based on personnel distribution provided in an embodiment of the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0062] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A or solution B, or a solution that simultaneously satisfies A and B. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0063] Please see Figure 1 As shown, the intelligent building energy-saving solution automatic generation system based on building big data includes a data acquisition module, a data modeling module, a model matching module, an energy efficiency analysis module, and a control execution module.

[0064] The data acquisition module is responsible for acquiring environmental and energy consumption data in real time from energy-consuming devices and sensors inside and outside the building on each floor and in each room. It then cleans, filters, and normalizes this data to obtain environmental and energy consumption data matrices. The module also accesses a mobile client in real time to obtain personnel distribution data, resulting in a personnel distribution data vector. Furthermore, it accesses the mobile internet in real time to obtain a time data vector. The module updates the environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector at preset time intervals.

[0065] At preset time intervals, the system accesses sensors inside and outside the building on each floor and in each room to obtain environmental data, including the building's internal temperature, external temperature, internal humidity, external humidity, internal carbon dioxide concentration, and surface light intensity.

[0066] At preset time intervals, access energy-consuming devices inside and outside the building to obtain energy consumption data, including the power of air conditioning, lighting and ventilation equipment in each room on each floor.

[0067] At preset time intervals, the system accesses the mobile client to obtain personnel distribution data, including the floor number and room number of each person.

[0068] At preset time intervals, access the mobile internet to obtain time data, including the current month, date, hour, and a workday identifier indicating whether the current day is a workday.

[0069] Furthermore, the environmental and energy consumption data are cleaned, filtered, and normalized. The specific process is as follows:

[0070] All environmental and energy consumption data are recursively estimated using Kalman filtering. For each environmental and energy consumption data point, the actual observed values ​​at the current moment are combined. Predicted value and the previous preset time interval To estimate the true value at the current moment The Kalman filter formula is:

[0071] ,

[0072] in For the current moment Kalman gain, This is a timestamp, representing the moment when environmental or energy consumption data was acquired. This is the timestamp of the previous preset time interval.

[0073] For all Kalman filter results Normalization was performed, and Kalman filter results were obtained for all timestamps of the following data: building interior temperature, building exterior temperature, building interior humidity, building exterior humidity, building interior carbon dioxide concentration, building surface light intensity, and the power of air conditioning, lighting, and ventilation equipment in each room on each floor. mean and standard deviation By using a preset formula:

[0074] ,

[0075] Calculate each timestamp Normalization results of environmental and energy consumption data Including each floor Each room Normalized results of building interior temperature Normalized results of building exterior temperature Normalized results of building interior humidity Normalized results of building exterior humidity Normalized results of carbon dioxide concentration inside buildings Normalized results of light intensity on building surfaces Normalized results of air conditioner power Normalized results of lighting equipment power and ventilation equipment power The normalized result.

[0076] in For floor numbering, This represents the total number of floors in the building.

[0077] Among them Number the rooms. ; This represents the total number of rooms contained in the building.

[0078] Furthermore, an environmental data matrix, an energy consumption data matrix, and a personnel distribution data vector are generated. The specific process is as follows:

[0079] Generate various timestamps Environmental data matrix:

[0080] ,

[0081] Generate various timestamps Energy consumption data matrix:

[0082] ,

[0083] Generate various timestamps Personnel distribution data vector:

[0084] ,

[0085] Among them Each floor Each room The number of people in the group.

[0086] Generate various timestamps Time data vector:

[0087] ,

[0088] Among them The current month. ;

[0089] Among them The current date. ;

[0090] Among them The current hour. ;

[0091] Among them This serves as the weekday identifier for the current time. ; This indicates that the day is a public holiday; The day indicates that it is a working day.

[0092] Furthermore, the elements corresponding to missing data values ​​in the environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector are marked with the data error symbol: error.

[0093] The generated environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector are sent to the data modeling module, energy efficiency analysis module, and model matching module.

[0094] The data modeling module records historical data for all environmental data matrices, energy consumption data matrices, personnel distribution data vectors, and time data vectors. It extracts correlation features from the environmental and energy consumption data matrices to establish a first inference model tracing the data from the environmental data matrix to the energy consumption data matrix. Similarly, it extracts correlation features from the personnel distribution data vector, time data vector, and energy consumption data matrix to establish a second inference model tracing the data from the personnel distribution data vector and time data vector to the energy consumption data matrix.

[0095] Please see Figure 2 As shown, the first inference model is established, specifically a support vector machine, which includes an input layer, a mapping layer, an optimization layer, and an output layer.

[0096] The input layer includes Each node is used to receive elements from the environmental data matrix, specifically the floor numbers. and room numbers The normalized results of the environmental data, including each floor number and room numbers Normalization results of building interior temperature Normalized results of building exterior temperature Normalized results of building interior humidity Normalized results of building exterior humidity Normalized results of carbon dioxide concentration inside buildings Normalized results of light intensity on building surfaces .

[0097] The mapping layer consists of several nodes, each receiving elements from two corresponding environmental data matrices. and It is responsible for numbering each floor using a kernel function. and room numbers The normalized environmental data is mapped to a high-dimensional space for linear separation, and its kernel function is:

[0098] ,

[0099] in, and These are elements in the two environment data matrices that represent the kernel function corresponding to the input mapping layer node, where... For elements and The square of the Euclidean distance between them.

[0100] in, The kernel function control parameter is obtained from machine learning training for the mapping layer node in the first inference model, and is used to control the width of the Gaussian distribution.

[0101] The optimization layer includes There are 10 pre-defined nodes, each responsible for finding the optimal separating hyperplane using an optimization algorithm based on Lagrange multipliers. The optimization function is:

[0102] ,

[0103] in, This refers to the node number of the optimization layer in the first inference model. This represents the permissible degree of misclassification for each node, where... and These are the normal vector and the hyperplane bias vector of the hyperplane obtained by the first inference model through machine learning training, respectively. This is the preset regularization parameter for the first inference model, used to control the degree of penalty for misclassification.

[0104] The output layer includes Each node is responsible for outputting the inference result of its corresponding energy consumption data, including the final inference result of the air conditioner power. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power .

[0105] It should be noted that the principle of the first inference model is to predict building energy consumption by establishing a mapping relationship between environmental data and energy consumption data. This model primarily predicts energy consumption data (such as the power of air conditioning, lighting, and ventilation equipment) based on environmental factors inside and outside the building, such as temperature, humidity, carbon dioxide concentration, and light intensity. The model aims to calculate the energy consumption of each device based on real-time changes in environmental data, thereby providing a basis for energy-saving decisions. By inputting environmental data, the model can help the system identify energy consumption patterns under different environmental conditions, thus achieving effective management and optimization of building energy consumption.

[0106] Please see Figure 3 As shown, a second inference model is established, specifically a support vector machine, which includes an input layer, a mapping layer, an optimization layer, and an output layer.

[0107] The input layer includes There are nodes, among which the first Each node is used to receive elements from the personnel distribution data vector, specifically the data for each floor. Each room The number of people in the data; the last four nodes are used to receive elements from the time data vector, specifically the month of the current time. Date of the current time The current hour and the current time's workday identifier .

[0108] The mapping layer consists of several nodes, each responsible for receiving an element from a personnel distribution data vector. and elements in a time data vector It is mapped to a high-dimensional space for linear separation, and its kernel function is:

[0109] ,

[0110] Among them For elements and The square of the Euclidean distance between them.

[0111] in, The kernel function control parameter, obtained from machine learning training, is used to control the width of the Gaussian distribution for the corresponding mapping layer node in the second inference model.

[0112] The optimization layer includes There are 10 pre-defined nodes, each responsible for finding the optimal separating hyperplane using an optimization algorithm based on Lagrange multipliers. The optimization function is:

[0113] ,

[0114] in, This refers to the node number of the optimization layer in the second inference model. , This represents the permissible degree of misclassification for each node, where... and Let be the normal vector and the hyperplane bias vector of the second inference model obtained through machine learning training, respectively. This is a preset regularization parameter for the second inference model, used to control the degree of penalty for misclassification.

[0115] The output layer includes Each node is responsible for outputting the inference result of its corresponding energy consumption data, including the final inference result of air conditioner power. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power .

[0116] It should be noted that the second inference model predicts building energy consumption by analyzing the relationship between personnel distribution, time factors, and energy consumption. This model takes personnel distribution (such as the number of people on different floors and in different rooms) and time data (such as month, date, hour, and whether it is a weekday) as input, and trains it using energy consumption data to generate energy consumption prediction results. The model aims to capture the impact of personnel activity and time patterns on energy consumption, helping the system dynamically adjust energy-consuming equipment within the building, especially during periods of high personnel density or low activity, to improve the building's energy efficiency and comfort.

[0117] The normal vectors and hyperplane bias vectors of the first and second inference models are obtained through machine learning training. The separating hyperplanes and decision boundaries of the first and second inference models are trained by calculating actual energy consumption and energy consumption data predicted by the models.

[0118] The model matching module acquires the environmental data matrix and personnel distribution data vector generated in real time from the previous preset time interval and performs data integrity analysis on them respectively. Based on the data integrity analysis results, it selects to activate either the first inference model or the second inference model to obtain the final inference result of the energy consumption data matrix.

[0119] The environmental data matrix is ​​subjected to data integrity analysis, and the specific process is as follows:

[0120] Obtain the number of error symbols in the environmental data matrix generated at the previous preset time interval, and count the number of each floor. The first integrity verification character is obtained by counting the number of data error symbols. If a floor number is identified in the environmental data matrix The number of corresponding data error symbols If the value exceeds a preset threshold, then the floor number is determined. The environmental data failed data integrity verification. If a floor number that failed data integrity verification is identified in the environmental data matrix... Greater than the total number of floors If 20% of the data is not passed, the environmental data matrix generated in the previous preset time interval is deemed to have failed the data integrity analysis.

[0121] The data integrity analysis of the personnel distribution data vector is performed as follows:

[0122] Obtain the number of error symbols in the personnel distribution data vector and time data vector generated at the previous preset time interval, and count the floor numbers in the personnel distribution data vector respectively. The second integrity verification symbol is obtained by counting the number of data error symbols. If a floor number is identified in the personnel distribution data vector. The number of corresponding data error symbols If the value exceeds a preset threshold, then the floor number is determined. The personnel distribution data failed data integrity verification. If a floor number that failed data integrity verification is identified in the personnel distribution data vector... Greater than the total number of floors If 20% of the data is not passed, the personnel distribution data vector generated in the previous preset time interval is determined to have failed the data integrity analysis.

[0123] Furthermore, based on the data integrity analysis results, the first inference model or the second inference model is selected for use. The specific process is as follows:

[0124] If the environmental data matrix generated in the previous preset time interval passes the data integrity analysis, but the personnel distribution data vector generated in the previous preset time interval fails the data integrity analysis, then the first inference model is activated, and the final inference result of its output air conditioning power is recorded. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power .

[0125] If the environmental data matrix generated in the previous preset time interval fails the data integrity analysis, while the personnel distribution data vector generated in the previous preset time interval passes the data integrity analysis, then the second inference model is activated, and the final inference result of its output air conditioning power is recorded. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power .

[0126] If the environmental data matrix and personnel distribution data vector generated in the previous preset time interval are both identified as having passed the data integrity analysis, then the summation results of the first integrity validator and the second integrity validator are further compared. If the summation result of the first integrity validator is identified as greater than the summation result of the second integrity validator, then the second inference model is activated, and its final inference result of the air conditioning power is recorded. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power If the summation result of the second integrity checkpoint is found to be greater than the summation result of the first integrity checkpoint, then the first inference model is activated, and the final inference result of its output air conditioner power is recorded. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power ;

[0127] If the environmental data matrix and personnel distribution data vector generated in the previous preset time interval are found to have failed the data integrity analysis, then both the first inference model and the second inference model are activated simultaneously. The average values ​​of the inference results for air conditioning power, lighting equipment power, and ventilation equipment power output by the first and second inference models are calculated to obtain the final inference result for air conditioning power. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power .

[0128] The final inference result of the obtained air conditioner power The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power Send to the energy efficiency analysis module.

[0129] It's important to note that the purpose of data integrity analysis is to ensure that the data sources used are as complete and accurate as possible, thereby reducing prediction errors caused by missing or incorrect data. In practical applications, sources such as environmental data and personnel distribution data may be incomplete or contain errors. Directly using this incomplete data for energy efficiency analysis may lead to biased energy consumption predictions. Therefore, data integrity analysis verifies the data before it enters the inference model, removing abnormal or severely missing portions to ensure the reliability of the model's input. In this way, the system can maximize the use of available complete data, improving the accuracy and reliability of predictions, thus providing a more accurate basis for subsequent energy-saving decisions.

[0130] The energy efficiency analysis module compares the energy consumption data matrix generated at the previous preset time interval with the final inference result of the energy consumption data matrix to obtain an energy consumption data adjustment matrix containing energy-saving scheme information.

[0131] The final inference result based on the comparison of air conditioner power. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power The elements in the energy consumption data matrix generated at the previous preset time interval are compared with the corresponding elements in the energy consumption data matrix generated at the previous preset time interval. The final inference result is obtained by subtracting the corresponding air conditioner power from each element in the energy consumption data matrix generated at the previous preset time interval. The final inference result of the power of lighting equipment The final inference result of the ventilation equipment power This yields an energy consumption data adjustment matrix containing energy-saving scheme information:

[0132] ,

[0133] The energy consumption data adjustment matrix is ​​sent to the control execution module.

[0134] The control execution module adjusts the matrix based on energy consumption data to generate energy consumption control signals, which are then distributed to each energy-consuming device to control and execute the energy-saving scheme information.

[0135] Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the floor numbers. Room numbers The power adjustment amount of the air conditioner power ;

[0136] Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the floor numbers. Room numbers Power adjustment amount of lighting equipment power ;

[0137] Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the floor numbers. Room numbers Power adjustment amount of ventilation equipment .

[0138] Please see Figure 4 and Figure 5 As shown, each floor is numbered. Room numbers The power adjustment amounts for air conditioning, lighting, and ventilation equipment are the target adjustment amounts, and PID control guides all energy-consuming equipment to adjust their power.

[0139] It should be noted that the overall technical solution utilizes historical data to correct abnormal energy consumption, establishes an inference model to predict building energy consumption under different environmental conditions and human activities, generates energy-saving solutions, and ultimately uses a control execution module to precisely control the energy efficiency of various energy-consuming devices within the building. This method of correction and optimization based on historical data can effectively reduce errors in energy consumption prediction, improve the building's energy utilization efficiency, and achieve intelligent energy-saving management.

[0140] It's important to further clarify that building energy consumption patterns are typically influenced by factors such as occupant distribution and the external environment. Under similar occupant distribution and environmental conditions, energy consumption generally maintains a certain degree of similarity. For example, under the same season and weather conditions, environmental parameters such as temperature and humidity within a building are roughly the same, and the occupant distribution is relatively stable. Therefore, energy consumption on the same floor or in the same room should be roughly similar under these circumstances. If the energy consumption of a particular room is significantly higher than that of other rooms, and this high energy consumption contradicts the occupant distribution and external environmental conditions of its floor, it likely indicates energy waste in that room. For example, air conditioning or lighting equipment operating at high power for extended periods, or equipment operating inefficiently, can lead to abnormally high energy consumption. By comparing and analyzing historical data, the system can identify such deviations from the norm and determine potential energy waste issues. Specifically, historical data provides the system with a baseline of "normal energy consumption" under similar environmental and occupant distribution conditions. When current energy consumption data deviates from the normal range of historical data, the system can automatically flag these abnormal energy consumption patterns and make corrections. In this way, the system can not only accurately identify and correct energy waste, but also provide a strong basis for energy-saving decisions, optimize building energy consumption management, and improve overall energy efficiency.

[0141] Please see Figure 6As shown, number each floor. Room numbers The sum of the power adjustment amounts for air conditioning, lighting, and ventilation equipment, divided by the number of people in the room, yields the room number. The energy consumption rating is generated and converted into color values ​​according to the numerical value, creating an energy consumption grading map based on personnel distribution. This personnel-based energy consumption grading map is then sent to the backend administrator.

Claims

1. An automatic generation system for smart building energy-saving solutions based on building big data, characterized in that, include: The data acquisition module is responsible for acquiring environmental data and energy consumption data in real time, and cleaning, filtering and normalizing them to obtain environmental data matrix and energy consumption data matrix. Obtain personnel distribution data vectors and time data vectors; The data modeling module establishes a first inference model from the environmental data matrix to the energy consumption data matrix; and establishes a second inference model from the personnel distribution data vector and the time data vector to the energy consumption data matrix. The first reasoning model is as follows: It includes an input layer, a mapping layer, an optimization layer, and an output layer; The input layer includes several nodes, which are used to receive elements from the environmental data matrix, specifically the normalized results of environmental data for each floor and each room. The mapping layer includes several nodes. Each node receives two elements from its corresponding environmental data matrix and is responsible for mapping the normalized results of the environmental data of each floor and each room to a high-dimensional space through a kernel function so as to facilitate linear separation. The optimization layer includes a preset number of nodes, which are responsible for finding the optimal separating hyperplane through an optimization algorithm based on Lagrange multipliers, and recording the normal vector and the hyperplane bias vector obtained by machine learning training. The output layer is responsible for outputting the inference results of each corresponding energy consumption data, including the final inference results of air conditioning power, lighting equipment power, and ventilation equipment power. The second reasoning model is as follows: It includes an input layer, a mapping layer, an optimization layer, and an output layer; The input layer includes several nodes, divided into a first part of nodes and a second part of nodes. The first part of nodes is used to receive elements from the personnel distribution data vector, specifically the number of people in each room on each floor. The second part of nodes is used to receive elements from the time data vector, specifically the month, date, hour, and weekday identifier of the current time. The mapping layer consists of several nodes. Each node is responsible for receiving elements from a personnel distribution data vector and an element from a time data vector, and mapping them to a high-dimensional space for linear separation. The optimization layer includes a preset number of nodes, which are responsible for finding the optimal separating hyperplane through an optimization algorithm based on Lagrange multipliers, and storing the normal vector and hyperplane bias vector of the hyperplane of the second inference model obtained through machine learning training. The output layer is responsible for outputting the inference results of each corresponding energy consumption data, including the final inference results of air conditioning power, lighting equipment power, and ventilation equipment power. The model matching module performs data integrity analysis on the environmental data matrix and the personnel distribution data vector respectively; based on the data integrity analysis results, it selects to enable the first inference model or the second inference model to obtain the final inference result of the energy consumption data matrix; The specific process of conducting data integrity analysis is as follows: The environmental data matrix is ​​subjected to data integrity analysis, and the specific process is as follows: Obtain the number of error symbols in the environmental data matrix generated at the previous preset time interval, and count the number of each floor. The number of data error symbols corresponds to the first integrity verification symbol for each floor. If the number of data error symbols for a certain floor in the environmental data matrix is ​​greater than a preset threshold, the environmental data for that floor is determined to have failed the data integrity verification. If the number of floors in the environmental data matrix that have failed the data integrity verification is greater than 20% of the total number of floors, the environmental data matrix generated in the previous preset time interval is determined to have failed the data integrity analysis. The data integrity analysis of the personnel distribution data vector is performed as follows: The system retrieves the number of error symbols in the personnel distribution data vector and time data vector generated in the previous preset time interval, and counts the number of error symbols corresponding to each floor in the personnel distribution data vector to obtain the second integrity verification symbol for each floor. If the number of error symbols corresponding to a certain floor in the personnel distribution data vector is greater than a preset threshold, the personnel distribution data for that floor number is determined to have failed the data integrity verification. If the number of floors that have failed the data integrity verification in the personnel distribution data vector is greater than 20% of the total number of floors, the personnel distribution data vector generated in the previous preset time interval is determined to have failed the data integrity analysis. Choose to enable either the first inference model or the second inference model based on the data integrity analysis results.

2. The automatic generation system for smart building energy-saving solutions based on building big data as described in claim 1, characterized in that, It also includes an energy efficiency analysis module and a control execution module; The energy efficiency analysis module compares the energy consumption data matrix generated at the previous preset time interval with the final inference result of the energy consumption data matrix to obtain an energy consumption data adjustment matrix containing energy-saving scheme information; The control execution module adjusts the matrix based on energy consumption data to generate energy consumption control signals, which are then distributed to each energy-consuming device to control and execute the energy-saving scheme information.

3. The automatic generation system for smart building energy-saving solutions based on building big data as described in claim 1, characterized in that, The specific process of cleaning, filtering, and normalizing real-time acquired environmental and energy consumption data is as follows: At preset time intervals, the sensors inside and outside the building on each floor and in each room are accessed to obtain environmental data, including the building's internal temperature, external temperature, internal humidity, external humidity, internal carbon dioxide concentration, and light intensity on the building surface. At preset time intervals, access energy-consuming devices inside and outside the building to obtain energy consumption data, including the power of air conditioning, lighting and ventilation equipment in each room on each floor; At preset time intervals, access the mobile client to obtain personnel distribution data, including the floor number and room number of each person. At preset time intervals, access the mobile internet to obtain time data, including the current month, date, hour, and a workday identifier indicating whether the current day is a workday; All environmental and energy consumption data are recursively estimated using Kalman filtering; for each environmental and energy consumption data point, the actual value at the current moment is estimated by combining the actual observed value at the current moment with the predicted value at the previous preset time interval. All Kalman filter results are normalized to obtain the mean and standard deviation of the Kalman filter results for the following data at all timestamps: building interior temperature, building exterior temperature, building interior humidity, building exterior humidity, building interior carbon dioxide concentration, building surface illuminance, and the power of air conditioning, lighting, and ventilation equipment in each room on each floor. Each Kalman filter result is then subtracted from its corresponding mean and divided by its corresponding standard deviation to obtain the normalized results of environmental and energy consumption data at each timestamp. This includes the normalized results for the building interior temperature, building exterior temperature, building interior humidity, building exterior humidity, building interior carbon dioxide concentration, building surface illuminance, air conditioning power, lighting equipment power, and ventilation equipment power in each room on each floor.

4. The automatic generation system for smart building energy-saving solutions based on building big data as described in claim 1, characterized in that, The specific process for generating the environmental data matrix, energy consumption data matrix, and personnel distribution data vector is as follows: An environmental data matrix for each time stamp is generated based on the normalized results of the building's internal temperature, external temperature, internal humidity, external humidity, internal carbon dioxide concentration, and surface light intensity in each room on each floor. Based on the normalized results of the air conditioning power, lighting power, and ventilation power in each room on each floor, an energy consumption data matrix for each timestamp is generated. Generate a personnel distribution data vector for each timestamp based on the number of people in each room on each floor: Generate time data vectors for each timestamp based on the month, date, hour, and weekday identifier of the current time; The elements corresponding to missing data values ​​in the environmental data matrix, energy consumption data matrix, personnel distribution data vector, and time data vector are marked with the data error symbol.

5. The automatic generation system for smart building energy-saving solutions based on building big data as described in claim 1, characterized in that, The specific process for selecting whether to use the first or second inference model based on the data integrity analysis results is as follows: If the environmental data matrix generated in the previous preset time interval is found to pass the data integrity analysis, but the personnel distribution data vector generated in the previous preset time interval fails the data integrity analysis, then the first inference model is activated. If the environmental data matrix generated in the previous preset time interval fails the data integrity analysis, while the personnel distribution data vector generated in the previous preset time interval passes the data integrity analysis, then the second inference model is activated. If the environmental data matrix and personnel distribution data vector generated in the previous preset time interval are both identified as having passed the data integrity analysis, then the summation results of the first integrity verification symbol and the second integrity verification symbol are further compared; if the summation result of the first integrity verification symbol is identified as being greater than the summation result of the second integrity verification symbol, then the second inference model is activated; if the summation result of the second integrity verification symbol is identified as being greater than the summation result of the first integrity verification symbol, then the first inference model is activated. If it is found that neither the environmental data matrix nor the personnel distribution data vector generated in the previous preset time interval has passed the data integrity analysis, then the first inference model and the second inference model are activated simultaneously. The average values ​​of the inference results of the air conditioning power, lighting equipment power and ventilation equipment power output by the first inference model and the second inference model are calculated respectively to obtain the final inference results of the air conditioning power, lighting equipment power and ventilation equipment power.

6. The automatic generation system for smart building energy-saving solutions based on building big data as described in claim 2, characterized in that, The specific process for obtaining the energy consumption data adjustment matrix containing energy-saving scheme information is as follows: Subtract the final inference results of the air conditioning power, lighting equipment power, and ventilation equipment power from each element in the energy consumption data matrix generated in the previous preset time interval to obtain an energy consumption data adjustment matrix containing energy-saving scheme information; send the energy consumption data adjustment matrix to the control execution module.

7. The automatic generation system for smart building energy-saving solutions based on building big data as described in claim 2, characterized in that, The specific process for controlling and executing energy-saving scheme information is as follows: Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the power adjustment amount of the air conditioning power of each room on each floor; Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the power adjustment amount of the lighting equipment power in each room on each floor; Obtain the energy consumption data adjustment matrix containing energy-saving scheme information, and parse its elements to obtain the power adjustment amount of the ventilation equipment power in each room on each floor; The power adjustment of air conditioning, lighting, and ventilation equipment in each room on each floor is used as the adjustment target. PID control guides all energy-consuming equipment to adjust their power.

Images