Intelligent campus energy-saving control method and system

Abstract

The invention provides an intelligent campus energy-saving control method and system. The method comprises the following steps: acquiring campus environment data and people flow data by using a sensor network and uploading the data to a central control system; constructing a feature project according to the feature vector set, performing enhancement analysis, and identifying a current teaching activity type; generating a personalized environment adjustment strategy according to the learning mode and the environment parameters, sending the personalized environment adjustment strategy to a central control system, and commanding each device to carry out specific parameter adjustment; distributing the environment adjustment strategy to different scenes of the campus, carrying out adaptive adjustment, introducing an equipment response model to carry out fault detection in real time, and introducing a regularization energy consumption optimization item to carry out energy-saving adjustment of the equipment; user feedback is collected, and user experience is optimized. Through an intelligent means, higher comfort and lower energy consumption are brought to the campus environment, and powerful support is provided for sustainable development of the education field.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart campus, and in particular relates to a smart campus energy-saving control method and system. Background Art

[0002] With the global emphasis on sustainable development, campus energy management has become an important challenge for higher education institutions. The campus environment is usually composed of multiple buildings and facilities, including classrooms, laboratories, libraries, and sports fields. Energy consumption mainly comes from lighting, air conditioning, heating, and equipment use. Existing energy management systems usually adopt fixed adjustment strategies and cannot flexibly adapt to different teaching activities and usage needs. This approach has led to energy waste to some extent and failed to make full use of natural light and ventilation, especially when there are large fluctuations in different courses (such as laboratory classes, lectures, or discussion classes) and student numbers. Traditional systems also lack the ability to respond to environmental changes (such as weather changes and crowd density) in real time, making it difficult to achieve accurate energy efficiency optimization, resulting in inefficient equipment operation and increased operating costs. In addition, most campus energy management systems are relatively weak in data analysis capabilities and lack in-depth mining and analysis of historical usage patterns, making it impossible to provide personalized energy-saving suggestions and real-time feedback. These shortcomings have greatly reduced the effectiveness of existing systems in improving campus energy efficiency and cannot meet the flexibility and sustainable development needs of modern educational environments. Summary of the invention

[0003] The purpose of the present invention is to design a smart campus energy-saving control method and system, which realizes dynamic and intelligent energy consumption management by combining learning pattern recognition and microclimate monitoring.

[0004] In order to achieve the above object, a first aspect of the present invention provides a smart campus energy-saving control method, the method comprising:

[0005] The sensor network is used to collect campus environment data and human flow data and upload them to the central control system to form a multidimensional array. The trend analysis of each time period is calculated based on the multidimensional array. The data after trend analysis is output into the format required by the decision support system to form environmental adaptation suggestions for different learning modes. The environmental adaptation suggestions are formed into a feature vector set.

[0006] Construct feature engineering and perform enhanced analysis based on the feature vector set to identify the current teaching activity type and generate each learning mode and its corresponding environmental control parameters, wherein the control parameters include: the optimal temperature range, lighting intensity and air circulation volume adapted to the course;

[0007] Generate personalized environmental adjustment strategies based on learning patterns and environmental parameters to adjust environmental control parameters, generate final environmental adjustment strategies, and send them to the central control system to instruct each device to make specific parameter adjustments;

[0008] The central control system distributes environmental adjustment strategies to different scenarios on campus, and makes adaptive adjustments to environmental adjustment strategies based on the environmental parameter requirements, learning mode, and real-time traffic flow of each scenario. It also introduces equipment response models for real-time fault detection and regularized energy consumption optimization items for energy-saving adjustment of equipment.

[0009] Collect user feedback and optimize user experience.

[0010] Furthermore, in the process of using the sensor network to collect campus environment data and human flow data and uploading them to the central control system, the collected campus environment data and human flow data need to be pre-processed and integrated into a multi-dimensional array;

[0011] The step of calculating the trend analysis of each time period based on the multidimensional array includes the following steps:

[0012] The preprocessed data is smoothed using an adaptive smoothing algorithm, as shown below:

[0013] D snooth (t)=α(t)·D(t)+(1-α(t))·D smooth (t-1)

[0014] Among them, D smooth (t) is the smoothed data, D(t) is the data point at the current time t, is the adaptive smoothing factor, k is the slope parameter, and t0 is the inflection point of data change;

[0015] The sliding window method is used to calculate the trend changes in each time period, and the window size is set to W, which is expressed as follows:

[0016]

[0017] Among them, T trend (t) is the trend of the current window, λ is the regularization parameter, Reg(D smooth ) is a regularization term for trend changes.

[0018] Furthermore, the step of constructing feature engineering based on the feature vector set and performing enhanced analysis includes the following steps:

[0019] Perform deep feature extraction on the feature vector set to construct a deep feature vector, wherein the deep feature vector includes the environmental parameters corresponding to the course, the standard deviation and maximum value after preprocessing, and a heat map based on classroom activity data, which is used to reflect the flow density of people in the classroom at different time periods;

[0020] Design an adaptive dynamic clustering algorithm to divide course patterns according to the multi-dimensional characteristics of feature vectors;

[0021] According to the division results, the learning mode of each type of course is automatically identified and the corresponding environmental adjustment mode is generated.

[0022] Further, the heat map based on classroom activity data is a heat map generated by cross-analysis of crowd counter data and time series to model the intensity of activities within a specific course time period;

[0023] In the process of deep feature extraction, a regularization term is introduced to avoid interference of noise and outliers on feature extraction;

[0024] The adaptive dynamic clustering algorithm is based on the improved K-means clustering algorithm, and introduces a penalty term to control the dispersion of the course mode, which is expressed as follows:

[0025]

[0026] Among them, C j is the jth class, μ j is the characteristic mean of this type of course; Var(H activity (t)) is based on the volatility of the flow heat map, γ is the adjustment parameter to ensure that courses with high flow changes are independently classified; Penalty (C) is a penalty term for uneven sample distribution in course clustering to prevent excessive concentration of a certain type of course, resulting in poor generalization ability;

[0027] The penalty term Penalty (C) for uneven sample distribution in course clustering is expressed as follows:

[0028]

[0029] This ensures that the distribution of various courses is relatively even and avoids the situation where the course model is too monotonous.

[0030] Furthermore, the step of generating a personalized environment adjustment strategy according to the learning mode and the environment parameters to adjust the environment control parameters includes the following steps:

[0031] A specific set of environmental adjustment targets is set for each learning mode, and these targets are calculated in real time through the environmental adjustment function to obtain the corresponding environmental target value;

[0032] Construct an energy efficiency optimization model and environmental regulation strategy to constrain and ensure a balance between comfort and energy consumption;

[0033] The adaptive feedback control mechanism designed based on the real-time data feedback of the sensor compares the difference between the current environmental parameters and the target set values, and adjusts the output of the regulating device to approach the target value.

[0034] Furthermore, the energy efficiency optimization model is expressed as follows:

[0035] The comprehensive energy consumption function E(t) is used to represent the total energy consumption of the system at time t:

[0036]

[0037] Among them, P HVAC (t) is the real-time power consumption of the air conditioning system; P lighting (t) is the real-time power consumption of the lighting system; P ventilation (t) is the real-time power consumption of the ventilation system; N is the number of all controlled areas in the campus;

[0038] At the same time, the optimization objective is defined as:

[0039] min(E(t))subject to:T min ≤T set (M i ,t)≤T max

[0040] Among them, T min and T max It is the upper and lower limits of the temperature in the comfort range, ensuring that the temperature adjustment does not exceed the reasonable comfort range;

[0041] In order to avoid extreme fluctuations in energy consumption, especially when the population density changes rapidly, a regularization term is introduced to control the smoothness of the system regulation. smooth (t), expressed as follows:

[0042]

[0043] Among them, λ4 is the weight of the regularization term.

[0044] Furthermore, the device response model is expressed as follows:

[0045] R device (t) = f(X actual (t)-X set (M i ,t))+λ5·Error(t)

[0046] Among them, X actual(t) is the actual output parameter of the device at present; Error(t) is the operating state error of the device; λ5 is the penalty coefficient of fault detection;

[0047] When R device (t) When the preset threshold is exceeded, the system automatically triggers the fault handling mechanism:

[0048] Select the following response options based on the device type and operating status:

[0049] Switch to backup equipment;

[0050] Regenerate the adjustment strategy to adapt to the current equipment status;

[0051] The introduction of regularized energy consumption optimization items to perform energy-saving regulation of equipment is expressed as follows:

[0052]

[0053] Where E(t) is the total energy consumption of the system at time t; is the rate of change of the energy consumption of the equipment, and λ6 controls the smoothness of the regulation change.

[0054] Furthermore, the step of collecting user feedback and optimizing user experience includes the following steps:

[0055] Collect subjective data and objective data in real time as feedback data. The subjective data is collected through the user feedback system, and the objective data is the actual parameters of the device. The feedback data is pre-processed to analyze the difference between the subjective data and the customer data, and evaluate the effect of the current adjustment strategy. If the difference is greater than the preset threshold, it means that the current adjustment does not achieve the ideal user experience, then return to automatically re-collect data for optimization;

[0056] Build an adaptive feedback control model to update the device settings in real time based on user feedback data and current environmental parameters;

[0057] After executing the updated set value, the real-time output of the equipment is monitored again through the sensor, and the output of the equipment is fine-tuned according to the difference between the actual environmental parameters and the set value;

[0058] According to historical user feedback and adjustment parameters, the adjustment strategy is optimized through machine learning technology. At the same time, energy-saving constraint regularization terms are added during the optimization process to dynamically adjust the relationship between user feedback weight and energy consumption.

[0059] Furthermore, the adaptive feedback control model updates the setting value of the device in real time according to the user feedback data and the current environmental parameters, as shown below:

[0060] X opt (t+1)=Xfinal (t)-η·ΔS(t)·W

[0061] Among them, X opy (t+1) is the optimal setting value of the next device; η is the feedback sensitivity adjustment coefficient, which controls the impact of feedback on the adjustment; ΔS(t) is the current feedback difference, which represents the deviation between the user's subjective score and the system target value; W is a weight matrix, which reflects the different impacts of different adjustment variables on user satisfaction;

[0062] The weight matrix W is a dynamically updated weight matrix, and the update is expressed as follows:

[0063]

[0064] Among them, w ij (t) is the adjustment parameter X at time t j The weight of ; β is the learning rate of weight update; is the sensitivity gradient of the feedback difference ΔS(t) to a specific regulation parameter;

[0065] The adjustment strategy is optimized by machine learning technology based on historical user feedback and adjustment parameters. In this process, the dynamic change law of the weight matrix W is trained so that the adjustment strategy can predict the adjustment demand in advance in different seasons, course types and scenarios with changes in the number of people. The learning formula of the weight matrix is ​​expressed as follows:

[0066]

[0067] Among them, w ij (t+T) is the adjustment parameter X at the next time T. j The weight of ; γ is the long-term learning rate, which is the learning amplitude of the control system to historical feedback; is the sensitivity of the feedback difference to the specific adjustment parameters at the next k moments.

[0068] In a second aspect of the present invention, a smart campus energy-saving control system is provided, which is applied to a smart campus energy-saving control method, and the system includes:

[0069] A data collection module (1) is used to collect campus environment data and human flow data using a sensor network and upload them to a central control system to form a multidimensional array, and calculate trend analysis of each time period based on the multidimensional array, output the data after trend analysis into a format required by a decision support system, form environmental adaptation suggestions for different learning modes, and form the environmental adaptation suggestions into a feature vector set;

[0070] A data analysis module (2) is used to construct feature engineering and perform enhanced analysis based on the feature vector set, identify the current teaching activity type, generate each learning mode and its corresponding environmental control parameters, and the environmental control parameters include: an optimal temperature range, lighting intensity and air circulation volume adapted to the course;

[0071] The energy-saving strategy generation module (3) is used to generate a personalized environment adjustment strategy according to the learning mode and the environment parameters to adjust the environment control parameters, generate a final environment adjustment strategy, and send it to the central control system to instruct each device to make specific parameter adjustments;

[0072] Energy-saving control optimization module (4), used for the central control system to distribute environmental adjustment strategies to different scenarios on campus, adaptively adjust the environmental adjustment strategies according to the requirements of each scenario for environmental parameters, learning mode and real-time flow of people, introduce equipment response model to detect faults in real time, and introduce regularized energy consumption optimization items to adjust the energy saving of equipment;

[0073] The smart campus system optimization module (5) is used to collect user feedback and optimize the user experience.

[0074] The beneficial technical effects of the present invention are at least as follows:

[0075] First, the system can analyze the characteristics of different courses (such as lectures, discussions, and experiments) in real time, and automatically adjust the classroom lighting, temperature, ventilation and other equipment settings according to the learning mode, so as to maximize comfort and energy utilization. Secondly, the integrated microclimate sensor monitors the temperature, humidity and air quality in the classroom in real time, and intelligently adjusts the equipment according to environmental changes, especially when the climatic conditions are not good, by optimizing the use of ventilation and air conditioning to reduce unnecessary energy consumption. In addition, the system also has data analysis and feedback functions, and uses machine learning algorithms to continuously optimize energy management strategies to form personalized energy-saving solutions. These innovations not only solve the problems of insufficient flexibility, untimely response and energy waste of traditional systems, but also effectively improve the overall energy efficiency of the campus and promote the goal of sustainable development. Through intelligent means, the present invention will bring higher comfort and lower energy consumption to the campus environment, and provide strong support for the sustainable development of educational institutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0077] Figure 1 This is a flow chart of a smart campus energy-saving control method of the present invention.

[0078] Figure 2 This is a framework diagram of a smart campus energy-saving control system of the present invention. DETAILED DESCRIPTION

[0079] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0080] In one or more embodiments, Figure 1 As shown, a smart campus energy-saving control method is disclosed, and the method includes the following steps S1-S5:

[0081] S1. Use the sensor network to collect campus environment data and crowd flow data and upload them to the central control system to form a multidimensional array. Calculate the trend analysis of each time period based on the multidimensional array, output the data after trend analysis into the format required by the decision support system, and form environmental adaptation suggestions for different learning modes. The environmental adaptation suggestions are combined into a feature vector set.

[0082] Specifically, select appropriate sensor types on campus, including temperature and humidity sensors (TH), CO2 sensors (C), people flow counters (P), etc. These sensors should be placed in key locations such as classrooms, corridors, and libraries to ensure comprehensive monitoring. The key formula for sensor placement is:

[0083]

[0084] Where R represents the effective monitoring range; N is the number of sensors; D i is the effective monitoring radius of the ith sensor (meters).

[0085] By optimizing the layout of sensors, the overlap of each monitoring area is minimized to obtain high-quality environmental data.

[0086] Furthermore, a wireless network-based data collection system is designed, using the Internet of Things (IoT) technology to enable sensors to upload data in real time. Each sensor uploads the current data record to the central control system at regular intervals (such as 1 minute). The data recording formula of the sensor is:

[0087] S(t)=TH(t)+C(t)+P(t)

[0088] Among them, S(t) represents the total data record at time t; TH(t) is the output of the temperature and humidity sensor at time t; C(t) is the output of the CO2 sensor at time t; P(t) is the output of the pedestrian flow counter at time t; this formula ensures that the outputs of all sensors are integrated each time the data is uploaded to form a complete environmental monitoring data set.

[0089] Furthermore, during the data collection process, the rationality of the data is verified in real time, and a custom anomaly detection algorithm (called a "dynamic threshold detection algorithm") is used to identify possible erroneous data. The algorithm sets a dynamic threshold based on the statistical characteristics of historical data:

[0090] T max =μ+k·σ

[0091] Among them, T max is the dynamic threshold; μ is the mean of historical data; σ is the standard deviation of historical data; k is the set safety factor (usually 2 or 3). When the sensor output data exceeds the dynamic threshold, the data is marked as abnormal and not recorded to ensure the accuracy of subsequent analysis data.

[0092] Furthermore, cloud database technology is used to store environmental data uploaded by sensors in real time and integrate data from different sensors. Data is stored with timestamps as indexes to facilitate subsequent retrieval and analysis. Data storage structure example:

[0093] The table structure contains fields: timestamp, sensor ID, sensor type, monitoring value

[0094] This ensures that subsequent steps can be analyzed based on time series and provides a complete and accurate basis for data preprocessing.

[0095] According to the above, the trend analysis of each time period is calculated according to the multidimensional array, and the data after the trend analysis is output into the format required by the decision support system to form environmental adaptation suggestions for different learning modes, and the environmental adaptation suggestions are composed into a feature vector set, including:

[0096] The obtained environmental monitoring data S(t) is integrated into a multidimensional array, represented as D = {D1, D2, ..., D m}, where D i Represents the measurement record of the i-th sensor at time t. This dataset covers a variety of environmental indicators such as temperature, humidity, light intensity, and CO2 concentration, and is stored in a structured manner for subsequent processing. Through data formatting, each record is standardized to a uniform time interval (such as every minute), and linear interpolation is used to fill in possible gaps in the time series.

[0097] Furthermore, for missing values, the “weighted interpolation method” is used for processing, and the formula is:

[0098]

[0099] Among them, D i (t) is the interpolation result of the i-th sensor at time t; is the weight coefficient, p is the adjustment parameter, which usually takes a value between 1 and 2 and is used to emphasize the impact of historical data from the current time; n is the number of historical time points considered; this method can effectively reduce the impact of data missing due to sensor failure or environmental factors on the overall data set, ensuring that subsequent analysis is based on a complete data set.

[0100] Furthermore, the “adaptive smoothing algorithm” is used to smooth the processed data, and the formula is:

[0101] D smooth (t)=α(t)·D(t)+(1-α(t))·D smooth (t-1)

[0102] Among them, D smooth (t) is the smoothed data; D(t) is the current data point; is the adaptive smoothing factor, k is the slope parameter, and t0 is the inflection point of data change; this smoothing method can dynamically adjust the degree of smoothing according to different stages of data change, so as to better adapt to different types of environmental fluctuations.

[0103] Furthermore, the sliding window method is used to calculate the trend change in each time period, and the window size is set to W. The calculation formula is:

[0104]

[0105] Among them, T trend (t) is the trend of the current window; λ is the regularization parameter, Reg(D smooth ) is a regular term for trend changes, such as the penalty term introduced by environmental interference or abnormal equipment fluctuations; this trend analysis method can not only capture the changing trend of data, but also identify potential abnormal fluctuations, which is particularly important in campus energy-saving control because equipment failures or external environmental changes can be discovered in a timely manner.

[0106] Furthermore, the trend-analyzed data is output in the format required by the decision support system to form a set of environmental adaptation suggestions for different learning modes. For example, different lighting and air conditioning setting suggestions are generated for the environmental adjustment requirements of laboratory classes and lecture classes to achieve energy consumption optimization. The output data structure is R = {R1, R2, ..., R k}, where Rj represents the energy efficiency optimization suggestion for the jth class of courses. This provides the basis for the subsequent intelligent control system.

[0107] S2. Construct feature engineering and perform enhanced analysis based on the feature vector set, identify the current teaching activity type, generate each learning mode and its corresponding environmental control parameters, and the environmental control parameters include: the optimal temperature range, lighting intensity and air circulation suitable for the course.

[0108] Specifically, first, the input data R = {R1, R2, ..., R k} Perform deep feature extraction and construct feature vector F j , containing key environmental and curriculum activity features. Each F j is the characteristic vector of a certain type of course, and its specific form is:

[0109] F j ={T trend (t),σ(D smooth ),Max(D smooth ),H activity (t)}

[0110] Among them, T trend (t) is the overall change trend of temperature, humidity, CO2, and light corresponding to the j-th course; σ(D smooth ) is the standard deviation of the smoothed data, which is used to capture the fluctuation of environmental parameters; Max(D smooth ) is the maximum value of the smoothed data, reflecting the peak value of the environmental conditions; H activiyy (t) is a heat map based on classroom activity data, reflecting the crowd density in the classroom at different time periods.

[0111] Among them, the innovative feature H activity (t): This feature models the activity intensity during a specific course period by generating a heat map from the cross analysis of the crowd counter data and the time series. The formula is:

[0112]

[0113] Among them, P(t) is the number of people in the classroom at time t; A is the area of ​​the classroom. This feature captures the dynamic changes of student activities during the course and provides an effective basis for identifying different types of teaching modes.

[0114] Furthermore, in order to avoid the interference of noise and outliers on feature extraction, the present invention adds regularization processing in the feature extraction process:

[0115] F j ′=F j +λ·Reg(Fj )

[0116] Among them, λ is a tuning parameter that controls the strength of regularization; Reg(F j ) is the regularization term of the feature, which limits the extreme changes of the feature value and ensures the robustness of the result.

[0117] Furthermore, in order to classify different course learning modes, the present invention proposes an adaptive dynamic clustering algorithm to divide course modes according to the multidimensional characteristics of feature vectors. The algorithm is based on the improved K-means clustering algorithm and introduces a penalty term to control the dispersion of course modes. The objective function is:

[0118]

[0119] Among them, C j is the jth class, μ j is the characteristic mean of this type of course; Var(H activity (t)) is based on the volatility of the flow heat map, γ is the adjustment parameter to ensure that courses with high flow changes are independently classified; Penalty (C) is a penalty term for uneven sample distribution in course clustering to prevent excessive concentration of a certain type of course, resulting in poor generalization ability.

[0120] Furthermore, the present invention introduces a penalty term Penalty (C) to limit the overly dense sample distribution in certain categories and avoid overfitting of pattern classification results. Specifically, it is as follows:

[0121]

[0122] This ensures that the distribution of various courses is relatively even and avoids the situation where the course model is too monotonous.

[0123] Furthermore, based on the dynamic clustering results, the system automatically identifies the learning mode of each type of course and generates the corresponding environment adjustment mode. The output is M = {M1, M2, ..., M n}, where each M i Represents a certain learning mode and its corresponding environmental control parameters (such as air conditioning temperature, lighting intensity, etc.).

[0124] Specific outputs for each learning mode include:

[0125] The best temperature range for this course (based on T trend (t) calculation);

[0126] Lighting intensity (based on crowd flow data and activity heat map H activity (t)decision);

[0127] Air flow (based on CO2 concentration and environmental fluctuations σ(D smooth ) to make adjustments).

[0128] S3. Generate a personalized environmental adjustment strategy based on the learning model and environmental parameters to adjust the environmental control parameters, generate the final environmental adjustment strategy, and send it to the central control system to instruct each device to make specific parameter adjustments.

[0129] Specifically, the learning pattern set M identified by the dynamic clustering algorithm is M={M1,M2,...,M n}, where each M i Represents a learning mode (such as lecture, discussion, experiment, etc.) and contains environmental characteristic data in this mode, such as temperature trend (T trend ), humidity fluctuation (σ humidity ), CO2 concentration change (ΔCO2), and activity heat map (H activity ).

[0130] The learning mode set is generated by processing sensor data and pattern recognition, forming the environmental requirements and adjustment targets corresponding to each mode. The environmental adjustment strategy will generate dynamic adjustment schemes based on these input data to achieve intelligent management for different learning scenarios.

[0131] Furthermore, for each learning mode M i Corresponding to a specific set of environmental adjustment targets, including temperature setting (T set ), lighting intensity setting (L set ), ventilation intensity setting (V set ) etc. The present invention calculates these targets in real time through the environmental adjustment function. First, according to the requirements of each mode, the temperature adjustment formula is defined:

[0132] T set (M i ,t)=T base +α·ΔT trend (M i ,t)+λ1·f weather (t)

[0133] Among them, T set (M i ,t) indicates that in the learning mode M i and target temperature at time t; T base is the default base temperature, generally set according to season and building type; ΔT trend (M i ,t) is the mode M i The temperature trend changes in the temperature; α is the temperature adjustment coefficient, which determines the sensitivity of the adjustment; λ1·fweather (t) is a dynamic adjustment term based on external weather conditions, where λ1 controls the magnitude of weather influence, and f weather (t) is a function of real-time weather data (e.g., outdoor temperature, humidity), ensuring that the regulation scheme takes weather changes into account.

[0134] For lighting intensity setting, the formula is as follows:

[0135] L set (M i ,t)=L base +β·H activity (t)+λ2·f ambient (t)

[0136] Among them, L set (M i ,t) is the time t in mode M i The lighting intensity in base is the default base lighting intensity; H activity (t) is the heat map of human flow, which adjusts the lighting intensity according to the real-time population density; β is the adjustment coefficient, which is used to control the sensitivity of lighting; λ2·f ambient (t) is the compensation term for external light, f ambient (t) Based on the natural light intensity obtained by sensors, ensure that artificial lighting is appropriately reduced when there is sufficient sunlight.

[0137] For the setting of ventilation intensity, the present invention adopts the following formula:

[0138] V set (M i ,t)=V base +γ·ΔCO2(M i ,t)+λ3·f occupancy (t)

[0139] Among them, V set (M i ,t) is the mode M at time t i Ventilation intensity in; V base is the default basic ventilation rate; ΔCO2(M i ,t) is the CO2 concentration change in step 3, which is used to adjust the ventilation volume according to the air quality; γ is the ventilation adjustment coefficient, which controls the amplitude of the ventilation speed change; λ3·f occupancy (t) is a dynamic adjustment item based on population density, f occupancy (t) is the crowd counting data monitored by the sensor in real time, which is used to adaptively adjust the ventilation volume according to the changes in the number of people in the classroom.

[0140] Through this dynamic adjustment formula design, the adjustment target of each learning mode is not only customized according to teaching needs, but also can adapt to changes in the external environment and the number of internal people in real time, thereby achieving efficient environmental control.

[0141] Furthermore, in order to achieve energy saving while ensuring comfort, the system introduces an energy efficiency optimization model to constrain the environmental adjustment strategy. The present invention designs a comprehensive energy consumption function E(t) to represent the total energy consumption of the system at time t:

[0142]

[0143] Among them, P HVAC (t) is the real-time power consumption of the air conditioning system; P lighting (t) is the real-time power consumption of the lighting system; P ventilation (t) is the real-time power consumption of the ventilation system; N is the number of all controlled areas in the campus.

[0144] The system will introduce optimization constraints for the above energy consumption model to ensure a balance between comfort and energy consumption. Define the optimization goal:

[0145] min(E(t))subject to:T min ≤T set (M i ,t)≤T max

[0146] T min and T max It is the upper and lower limits of the temperature in the comfort range, ensuring that the temperature adjustment does not exceed the reasonable comfort range;

[0147] Similar constraints are applied to lighting intensity and ventilation rates, ensuring that user comfort is not compromised while optimizing energy consumption.

[0148] Furthermore, in order to avoid extreme fluctuations in energy consumption, especially when the population density changes rapidly, the present invention also introduces a regularization term to control the smoothness of the system adjustment:

[0149]

[0150] This regularization term ensures that the change of system energy consumption at any time is gradual by controlling the severity of the change of energy consumption over time, avoiding sudden high energy consumption behavior during the regulation process.

[0151] Due to the large fluctuations in environmental variables and the number of people, the environmental adjustment strategy needs to have adaptive capabilities. The designed adaptive feedback control mechanism is based on the real-time data feedback of the sensor, compares the difference between the current environmental parameters and the target set values, and gradually approaches the target value by adjusting the output of the adjustment device. The feedback deviation control formula is defined as:

[0152] ΔX(t)=X current (t)-X set (M i ,t)

[0153] Where X(t) represents environmental parameters (such as temperature, lighting, ventilation), X current (t) is the current environmental value collected by the sensor in real time; X set (M i ,t) is the target setting value. The deviation ΔX(t) will be used to adjust the output value of the regulating device. The specific adjustment formula is:

[0154] X adjust (t+1)=X current (t)-K·ΔX(t)

[0155] Among them, K is the feedback control gain coefficient, which determines the speed of system response.

[0156] This feedback mechanism ensures that the system can respond to external environmental changes in real time, such as sudden fluctuations in traffic or weather changes. The difference between the actual value and the target value after each adjustment will be gradually reduced until the system stabilizes in the optimal environmental state.

[0157] The system generates the final environmental adjustment strategy and sends it to the central control system to instruct each device (such as air conditioning, lighting, and ventilation systems) to adjust specific parameters. The format of the environmental adjustment command is C(t) = {T set ,L set ,V set}, the command will be updated periodically to ensure that the device settings at each moment are consistent with the real-time environment requirements.

[0158] S4. The central control system distributes the environmental adjustment strategy to different scenarios on campus, and adaptively adjusts the environmental adjustment strategy according to the requirements of each scenario for environmental parameters, learning mode and real-time traffic flow. It also introduces the equipment response model for real-time fault detection and introduces regularized energy consumption optimization items for energy-saving adjustment of equipment.

[0159] Specifically, the input of this step is the environmental adjustment strategy C(t) generated in step S3 = {T set ,L set ,V setThis is a dynamic adjustment target generated by combining the specific needs of different learning modes (such as lectures, lab classes, and discussion classes) in the smart campus and real-time environmental data (such as temperature, lighting, and CO2 concentration).

[0160] The input data C(t) is distributed by the central control system to the equipment management systems in the campus, such as air conditioning, lighting and ventilation systems. Each subsystem processes T set (temperature setting value), L set (lighting setpoint) and V set (ventilation intensity setpoints) to execute equipment adjustment commands. These setpoints are customized according to real-time learning patterns and their needs, ensuring that environmental adjustments are adapted to the actual needs of different areas on campus.

[0161] Furthermore, different areas of the campus (classrooms, laboratories, libraries, etc.) have different environmental adjustment requirements. For example, lectures and discussion classes have different requirements for temperature and lighting, and experimental classes have higher requirements for ventilation. The control instructions of specific equipment in the adjustment strategy are adaptively adjusted according to the current learning mode and real-time flow of people in each area.

[0162] For air conditioning systems, the control signal I HVAC (t) contain the following:

[0163] I HVAC (t) = {Turn On / Off, T set (M i ,t),Mode,Fan Speed}

[0164] Among them, Turn On / Off determines the start and stop of the air conditioner; T set (M i ,t) indicates mode M i The target temperature corresponding to time t; Mode indicates the air conditioning mode (cooling or heating); Fan Speed ​​is calculated based on the real-time ventilation demand V set (M i ,t)Adjust the fan speed.

[0165] Control signal of lighting system I lighting (t) is:

[0166] I lighting (t) = {Turn On / Off, L set (M i ,t),Brightness Level}

[0167] Among them, L set (M i,t) Dynamically adjust according to course activities and natural light intensity to ensure both energy saving and providing an appropriate lighting environment.

[0168] Control signal of ventilation system I ventilation (t) is:

[0169] I ventilation (t) = {Turn On / Off, V set (M i ,t),Fan Speed}

[0170] By monitoring the CO2 concentration in the classroom and analyzing the flow of people, the ventilation system is adjusted to ensure that the air quality meets the standards. These equipment control signals are sent to the corresponding equipment in real time through the central control system to ensure accurate implementation of environmental adjustments in different areas.

[0171] Furthermore, in a smart campus, the usage of classrooms, laboratories and public areas changes at any time, such as the number of students, the course status, the natural environment (such as weather changes), etc. Traditional regulation systems often find it difficult to respond to these changes in real time, resulting in energy waste or an uncomfortable environment. The feedback mechanism of the present invention is intended to solve this problem.

[0172] The system will make feedback adjustments to the device output based on the real-time data provided by the sensor (such as current temperature, lighting intensity, CO2 concentration, etc.). The feedback formula is:

[0173] ΔX(t)=X current (t)-X set (M i ,t)

[0174] Among them, X current (t) is the current environmental state detected by the sensor; X set (M i ,t) is the target value for the current learning mode. The feedback adjustment formula is:

[0175] X adjust (t+1)=X current (t)-K·ΔX(t)

[0176] This formula adjusts the device output based on the feedback to ensure that the environmental state gradually approaches the target value X. set (M i ,t). The feedback control mechanism enables the system to automatically adjust the device status according to real-time changes, such as an increase in the number of students or changes in external temperature, solving the problem of traditional systems not being able to respond to emergencies in a timely manner.

[0177] Furthermore, in a smart campus, equipment failure (such as air conditioning failure to cool normally, insufficient air volume in the ventilation system) may cause environmental regulation failure, affecting teaching activities and student experience. Traditional energy management systems lack real-time fault detection and processing capabilities. The introduction of the equipment response model R device (t) to monitor the device status:

[0178] R device (t) = f(X actual (t)-X set (M i ,t))+λ5·Error(t)

[0179] Among them, X actual (t) is the actual output parameter of the device at present; Error(t) is the operating state error of the device, such as device failure, response delay, etc.; λ5 is the penalty coefficient of fault detection.

[0180] When R device (t) When the preset threshold is exceeded, the system automatically triggers the fault handling mechanism. Depending on the device type and operating status, the system can choose the following response plans:

[0181] Switch to backup equipment (such as turning on backup air conditioning or adjusting equipment load in adjacent areas);

[0182] Regenerate the adjustment strategy to adapt to the current available device status.

[0183] This mechanism ensures the reliability and continuity of environmental regulation. Even if equipment fails, the system can respond quickly and adjust the regulation strategy to ensure the stability of the teaching environment.

[0184] Furthermore, energy-saving optimization of energy management systems is one of the key goals of smart campuses. However, during the adjustment process, too fast or too slow adjustments may cause frequent start and stop of equipment, resulting in unnecessary increase in energy consumption and affecting students' learning experience.

[0185] The system introduces regularized energy consumption optimization items to ensure that the adjustment process can achieve energy saving goals while avoiding excessive energy consumption caused by frequent equipment adjustments. The optimization formula is:

[0186]

[0187] Where E(t) is the total energy consumption of the system at time t; is the rate of change of the energy consumption of the equipment, and λ6 controls the smoothness of the regulation change. This formula ensures that the regulation process is as smooth as possible, avoids large fluctuations in equipment load, and reduces the high energy consumption caused by frequent start and stop of equipment during the regulation process.

[0188] Furthermore, the environment in a smart campus changes frequently, especially changes in the number of students, weather changes, switching of teaching tasks, etc., and traditional systems are difficult to respond flexibly. The present invention ensures that environmental regulation can adapt to various changes in real time through a loop feedback mechanism and strategy regeneration.

[0189] At the end of each adjustment cycle, the system periodically checks the difference between the current environmental state and the target setting, and gradually approaches the target value through feedback control. If a large deviation is found (such as a sudden increase in the number of students), the system will recalculate the adjustment strategy and enter the next round of adjustment. This mechanism is seamlessly connected with the adjustment strategy generation in step S3, ensuring that the environmental adjustment system is highly flexible, adapts to the ever-changing campus environment, and achieves a stable and energy-saving adjustment effect.

[0190] S5. Collect user feedback and optimize user experience.

[0191] Specifically, the present invention combines subjective and objective data to achieve continuous optimization. Subjective data is collected through a user feedback system, and users rate the current environment based on their own feelings through mobile phone apps, touch screens in classrooms, etc. The variables collected by user feedback include:

[0192] F u (t) = {S temp (t),S light (t),S vent (t)}

[0193] Among them, S temp (t) represents the satisfaction rating of the current temperature; S light (t) represents the satisfaction rating of the current lighting intensity; S vent (t) represents the satisfaction score for the current air quality and ventilation intensity. By combining these subjective feedbacks with the actual parameters of the equipment, a more targeted feedback data set is formed for subsequent optimization.

[0194] Furthermore, user feedback data F u (t) After collection, the system first cleans the data to remove invalid or abnormal feedback values, such as feedback time that does not match the current class, or feedback data that is obviously abnormal (such as extreme scores). The cleaned feedback data is F u ′(t). Multi-dimensional feedback difference evaluation:

[0195] The system evaluates the effectiveness of the current adjustment strategy by analyzing the difference between user feedback and device output. The difference evaluation formula is:

[0196]

[0197] Where N is the number of users who provide feedback; Sactual (t) is the actual rating of user feedback (S temp (t),S light (t),S vent (t)); S target It is the ideal user satisfaction target value (such as 4.5 or 5 points).

[0198] If ΔS(t) exceeds a certain threshold δ S , indicating that the current adjustment does not achieve the ideal user experience, the system will automatically start the next optimization process. This difference evaluation mechanism combines subjective data with objective data to quickly discover problems in the current adjustment (such as high temperature, insufficient lighting, etc.).

[0199] Furthermore, in order to dynamically adjust the adjustment parameters, the system designs an adaptive feedback enhancement optimization model, called the Enhanced Feedback Adaptive Model (EFAM). This model updates the device's set values ​​in real time based on user feedback data and current environmental parameters. The specific adjustment formula is:

[0200] X opt (t+1)=X final (t)-η·ΔS(t)·W

[0201] Among them, X opt (t+1) is the optimal setting value of the next equipment (such as temperature, lighting, ventilation); η is the feedback sensitivity adjustment coefficient, which controls the impact of feedback on the adjustment; ΔS(t) is the current feedback difference, which represents the deviation between the user's subjective score and the system target value; W is a weight matrix, which reflects the different effects of different adjustment variables (such as temperature, lighting, ventilation) on user satisfaction.

[0202] Furthermore, in order to make the model more innovative, the weight matrix W is not a fixed value, but is dynamically adjusted by learning from historical feedback data. ij The impact on different areas and different course activities is different. For example, in lecture mode, the weight of temperature on user experience is greater than ventilation, while in experimental mode, ventilation has a higher weight. The weight matrix update formula is:

[0203]

[0204] Among them, w ij (t) is the adjustment parameter X at time t j The weight of ; β is the learning rate of weight update; is the sensitivity gradient of the feedback difference ΔS(t) to a specific adjustment parameter. This dynamic weight adjustment mechanism enables the system to adaptively optimize the weight distribution of each adjustment parameter based on long-term feedback data, thereby achieving more accurate adjustment optimization in different scenarios.

[0205] Furthermore, the regulation system is based on X opt (t+1) After adjusting the equipment settings, the real-time output of the equipment (temperature, lighting, ventilation intensity, etc.) is monitored again through sensors, and the actual environmental parameters X are used to current (t) and the set value X opt The system will periodically calculate the adjusted device output deviation:

[0206] ΔX(t+1)=X current (t+1)-X opt (t+1)

[0207] Among them, ΔX(t+1) is the deviation between the current environmental state and the optimized setting value; X current (t+1) is the device output value monitored in real time by the sensor. Through this closed-loop feedback control, the system can gradually approach the ideal environmental state expected by the user through multiple feedbacks and adjustments. Each round of feedback will regenerate the adjustment strategy, and through continuous adjustment, the system can maintain a balance between stability and flexibility.

[0208] Furthermore, in the process of long-term data accumulation, the system continuously optimizes the adjustment strategy through machine learning technology. The system trains the dynamic change law of the weight matrix W based on historical user feedback and adjustment parameters, so that the system can predict the adjustment needs in advance in different seasons, course types and scenarios with changes in the number of people. The weight learning formula is:

[0209]

[0210] Among them, w ij (t+T) is the adjustment parameter X at the next time T. j The weight of ; γ is the long-term learning rate, which is the learning amplitude of the control system to historical feedback; is the sensitivity of the feedback difference to a specific adjustment parameter at the next k moments. This long-term learning mechanism can not only optimize feedback in the short term, but also gradually improve the adjustment effect of the system in the long term, making the system more accurate and intelligent as more feedback is accumulated.

[0211] Furthermore, in order to achieve a balance between energy saving and user experience during the continuous optimization of the system, an energy-saving constraint regularization term is designed to dynamically adjust the relationship between user feedback weight and energy consumption. By balancing user experience and energy consumption, the system can limit energy consumption while ensuring user satisfaction. The dynamic balance formula is:

[0212] X opt (t+1)=X final (t+1)-λ E ·E smooth (t)+η·ΔS(t)

[0213] Among them, λ E is the energy consumption constraint coefficient, which controls the energy consumption weight in the regulation process; E smooth (t) is the rate of change of system energy consumption; η·ΔS(t) is the adjustment correction term based on user feedback. This formula ensures that the system can not only reduce energy consumption during the optimization process, but also improve user satisfaction, achieving the energy-saving goal of environmental regulation in smart campuses.

[0214] This step proposes an enhanced feedback-based adaptive optimization model (EFAM) by combining user feedback with real-time environmental data, and continuously optimizes the system adjustment strategy through dynamic weight adjustment and long-term learning mechanism. This solution effectively responds to complex and changing environmental requirements in smart campus scenarios and solves the problem of lack of user feedback and long-term optimization capabilities in traditional systems. By introducing a dynamic balance energy-saving mechanism, the system achieves effective energy consumption control while maintaining comfort, significantly improving the overall user experience and energy-saving efficiency.

[0215] In one or more embodiments, Figure 2 As shown, a smart campus energy-saving control system is disclosed, which is applied to a smart campus energy-saving control method. The system includes:

[0216] A data collection module (1) is used to collect campus environment data and human flow data using a sensor network and upload them to a central control system to form a multidimensional array, and calculate trend analysis of each time period based on the multidimensional array, output the data after trend analysis into a format required by a decision support system, form environmental adaptation suggestions for different learning modes, and form the environmental adaptation suggestions into a feature vector set;

[0217] A data analysis module (2) is used to construct feature engineering and perform enhanced analysis based on the feature vector set, identify the current teaching activity type, generate each learning mode and its corresponding environmental control parameters, and the environmental control parameters include: an optimal temperature range, lighting intensity and air circulation volume adapted to the course;

[0218] The energy-saving strategy generation module (3) is used to generate a personalized environment adjustment strategy according to the learning mode and the environment parameters to adjust the environment control parameters, generate a final environment adjustment strategy, and send it to the central control system to instruct each device to make specific parameter adjustments;

[0219] Energy-saving control optimization module (4), used for the central control system to distribute environmental adjustment strategies to different scenarios on campus, adaptively adjust the environmental adjustment strategies according to the requirements of each scenario for environmental parameters, learning mode and real-time flow of people, introduce equipment response model to detect faults in real time, and introduce regularized energy consumption optimization items to adjust the energy saving of equipment;

[0220] The smart campus system optimization module (5) is used to collect user feedback and optimize the user experience.

[0221] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0222] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0223] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A smart campus energy-saving control method, characterized in that: The method comprises: The sensor network is used to collect campus environment data and human flow data and upload them to the central control system to form a multidimensional array. The trend analysis of each time period is calculated based on the multidimensional array. The data after trend analysis is output into the format required by the decision support system to form environmental adaptation suggestions for different learning modes. The environmental adaptation suggestions are formed into a feature vector set. Construct feature engineering and perform enhanced analysis based on the feature vector set to identify the current teaching activity type and generate each learning mode and its corresponding environmental control parameters, wherein the environmental control parameters include: the optimal temperature range, lighting intensity and air circulation volume adapted to the course; Generate personalized environmental adjustment strategies based on learning patterns and environmental parameters to adjust environmental control parameters, generate final environmental adjustment strategies, and send them to the central control system to instruct each device to make specific parameter adjustments; The central control system distributes environmental adjustment strategies to different scenarios on campus, and makes adaptive adjustments to environmental adjustment strategies based on the environmental parameter requirements, learning mode, and real-time traffic flow of each scenario. It also introduces equipment response models for real-time fault detection and regularized energy consumption optimization items for energy-saving adjustment of equipment. Collect user feedback and optimize user experience.

2. A smart campus energy-saving control method according to claim 1, characterized in that: In the process of using the sensor network to collect campus environment data and human flow data and uploading them to the central control system, it is necessary to pre-process the collected campus environment data and human flow data and integrate them into a multi-dimensional array; The step of calculating the trend analysis of each time period based on the multidimensional array includes the following steps: The preprocessed data is smoothed using an adaptive smoothing algorithm, as shown below: D smooth (t)=α(t)·D(t)+(1-α(t))·D smooth (t-1) Among them, D smooth (t) is the smoothed data, D(t) is the data point at the current time t, is the adaptive smoothing factor, k is the slope parameter, and t0 is the inflection point of data change; The sliding window method is used to calculate the trend changes in each time period, and the window size is set to W, which is expressed as follows: Among them, T trend (t) is the trend of the current window, λ is the regularization parameter, Reg(D smooth ) is a regularization term for trend changes.

3. A smart campus energy-saving control method according to claim 1, characterized in that: The step of constructing feature engineering and performing enhanced analysis based on the feature vector set includes the following steps: Perform deep feature extraction on the feature vector set to construct a deep feature vector, wherein the deep feature vector includes the environmental parameters corresponding to the course, the standard deviation and maximum value after preprocessing, and a heat map based on classroom activity data, which is used to reflect the flow density of people in the classroom at different time periods; Design an adaptive dynamic clustering algorithm to divide course patterns according to the multi-dimensional characteristics of feature vectors; According to the division results, the learning mode of each type of course is automatically identified and the corresponding environmental adjustment mode is generated.

4. A smart campus energy-saving control method according to claim 3, characterized in that: The heat map based on classroom activity data is a heat map generated by cross-analysis of crowd counter data and time series, which models the activity intensity within a specific course time period; In the process of deep feature extraction, a regularization term is introduced to avoid interference of noise and outliers on feature extraction; The adaptive dynamic clustering algorithm is based on the improved K-means clustering algorithm, and introduces a penalty term to control the dispersion of the course mode, which is expressed as follows: Among them, C j is the jth class, μ j is the characteristic mean of this type of course; Var(H activity (t)) is based on the volatility of the flow heat map, γ is the adjustment parameter to ensure that courses with high flow changes are independently classified; Penalty (C) is a penalty term for uneven sample distribution in course clustering to prevent excessive concentration of a certain type of course, resulting in poor generalization ability; The penalty term Penalty (C) for uneven sample distribution in course clustering is expressed as follows: This ensures that the distribution of various courses is relatively even and avoids the situation where the course model is too monotonous.

5. The smart campus energy-saving control method according to claim 1 is characterized in that: The step of generating a personalized environment adjustment strategy based on the learning mode and the environment parameters to adjust the environment control parameters includes the following steps: A specific set of environmental adjustment targets is set for each learning mode, and these targets are calculated in real time through the environmental adjustment function to obtain the corresponding environmental target value; Construct an energy efficiency optimization model and environmental regulation strategy to constrain and ensure a balance between comfort and energy consumption; The adaptive feedback control mechanism designed based on the real-time data feedback of the sensor compares the difference between the current environmental parameters and the target set values, and adjusts the output of the regulating device to approach the target value.

6. A smart campus energy-saving control method according to claim 5, characterized in that: The energy efficiency optimization model is expressed as follows: The comprehensive energy consumption function E(t) is used to represent the total energy consumption of the system at time t: Among them, P HVAC (t) is the real-time power consumption of the air conditioning system; P lighting (t) is the real-time power consumption of the lighting system; P ventilation (t) is the real-time power consumption of the ventilation system; N is the number of all controlled areas in the campus; At the same time, the optimization objective is defined as: min(E(t))subject to:T min ≤T set (M i ,t)≤T max Among them, T min and T max It is the upper and lower limits of the temperature in the comfort range, ensuring that the temperature adjustment does not exceed the reasonable comfort range; In order to avoid extreme fluctuations in energy consumption, especially when the population density changes rapidly, a regularization term is introduced to control the smoothness of the system regulation. smooth (t), expressed as follows: Among them, λ4 is the weight of the regularization term.

7. A smart campus energy-saving control method according to claim 1, characterized in that: The device response model is expressed as follows: R device (t)=f(X actual (t)-X set (M i ,t))+λ5·Error(t) Among them, X actual (t) is the actual output parameter of the device at present; Error(t) is the operating state error of the device; λ5 is the penalty coefficient of fault detection; When R device (t) When the preset threshold is exceeded, the system automatically triggers the fault handling mechanism: Select the following response options based on the device type and operating status: Switch to backup equipment; Regenerate the adjustment strategy to adapt to the current equipment status; The introduction of regularized energy consumption optimization items to perform energy-saving regulation of equipment is expressed as follows: Where E(t) is the total energy consumption of the system at time t; is the rate of change of the energy consumption of the equipment, and λ6 controls the smoothness of the regulation change.

8. The smart campus energy-saving control method according to claim 1 is characterized in that: The step of collecting user feedback and optimizing user experience includes the following steps: Collect subjective data and objective data in real time as feedback data. The subjective data is collected through the user feedback system, and the objective data is the actual parameters of the device. The feedback data is pre-processed to analyze the difference between the subjective data and the customer data, and evaluate the effect of the current adjustment strategy. If the difference is greater than the preset threshold, it means that the current adjustment does not achieve the ideal user experience, then return to automatically re-collect data for optimization; Build an adaptive feedback control model to update the device settings in real time based on user feedback data and current environmental parameters; After executing the updated set value, the real-time output of the equipment is monitored again through the sensor, and the output of the equipment is fine-tuned according to the difference between the actual environmental parameters and the set value; According to historical user feedback and adjustment parameters, the adjustment strategy is optimized through machine learning technology. At the same time, energy-saving constraint regularization terms are added during the optimization process to dynamically adjust the relationship between user feedback weight and energy consumption.

9. A smart campus energy-saving control method according to claim 8, characterized in that: The adaptive feedback control model updates the set value of the device in real time according to the user feedback data and the current environmental parameters, as shown below: X opt (t+1)=X final (t)-η·ΔS(t)·W Among them, X opt (t+1) is the optimal setting value of the next device; η is the feedback sensitivity adjustment coefficient, which controls the impact of feedback on the adjustment; ΔS(t) is the current feedback difference, which represents the deviation between the user's subjective score and the system target value; W is a weight matrix, which reflects the different impacts of different adjustment variables on user satisfaction; The weight matrix W is a dynamically updated weight matrix, and the update is expressed as follows: Among them, w ij (t) is the adjustment parameter X at time t j The weight of ; β is the learning rate of weight update; is the sensitivity gradient of the feedback difference ΔS(t) to a specific regulation parameter; The adjustment strategy is optimized by machine learning technology based on historical user feedback and adjustment parameters. In this process, the dynamic change law of the weight matrix W is trained so that the adjustment strategy can predict the adjustment demand in advance in different seasons, course types and scenarios with changes in the number of people. The learning formula of the weight matrix is ​​expressed as follows: Among them, w ij (t+T) is the adjustment parameter X at the next time T. j The weight of ; γ is the long-term learning rate, which is the learning amplitude of the control system to historical feedback; is the sensitivity of the feedback difference to the specific adjustment parameters at the next k moments.

10. A smart campus energy-saving control system, characterized in that: A smart campus energy-saving control method applied to claim 1, the system comprising: A data collection module (1) is used to collect campus environment data and human flow data using a sensor network and upload them to a central control system to form a multidimensional array, and calculate trend analysis of each time period based on the multidimensional array, output the data after trend analysis into a format required by a decision support system, form environmental adaptation suggestions for different learning modes, and form the environmental adaptation suggestions into a feature vector set; A data analysis module (2) is used to construct feature engineering and perform enhanced analysis based on the feature vector set, identify the current teaching activity type, generate each learning mode and its corresponding environmental control parameters, and the environmental control parameters include: an optimal temperature range, lighting intensity and air circulation volume adapted to the course; The energy-saving strategy generation module (3) is used to generate a personalized environment adjustment strategy according to the learning mode and the environment parameters to adjust the environment control parameters, generate a final environment adjustment strategy, and send it to the central control system to instruct each device to make specific parameter adjustments; Energy-saving control optimization module (4), used for the central control system to distribute environmental adjustment strategies to different scenarios on campus, adaptively adjust the environmental adjustment strategies according to the requirements of each scenario for environmental parameters, learning mode and real-time flow of people, introduce equipment response model to detect faults in real time, and introduce regularized energy consumption optimization items to adjust the energy saving of equipment; The smart campus system optimization module (5) is used to collect user feedback and optimize the user experience.

Images