HVAC intelligent regulation and control method and system based on personnel dynamic prediction

Abstract

The invention discloses an HVAC intelligent regulation and control method and system based on personnel dynamic prediction, and the method comprises the steps: deploying a plurality of monitoring nodes in a building regulation and control region through a sensing layer, collecting the number of personnel and environment parameter data in the region in real time, and transmitting the data to a comprehensive regulation and control module according to a standard format; predicting the number of people in a short-term dimension and a historical dimension, and outputting a comprehensive predicted number of people after weight fusion; a decision-making layer calculates a target temperature set value according to the real-time number of people and the comprehensive predicted number of people in combination with the thermal comfort temperature and the energy-saving temperature, and generates a regional HVAC equipment regulation and control instruction based on a PID control algorithm; the execution layer adjusts HVAC equipment operation parameters of the corresponding area according to the regulation and control instruction, and the total load of a building air conditioner is synchronously calculated so as to adjust the power of an air conditioner host; and analyzing historical operation data to optimize regulation and control parameters, detecting abnormal data, generating a fault diagnosis report, and realizing iterative optimization.

Description

Technical Field

[0001] The present application relates to the field of intelligent building technology, and in particular to an HVAC intelligent control method and system based on dynamic prediction of personnel. Background Art

[0002] A large number of studies have shown that most HVAC (Heating, Ventilation and Air Conditioning) systems rely on a single temperature and humidity sensor or simple feedback control, without considering the number of people in the area, resulting in excessive cooling and heating of the equipment. For example, the distribution of people in different areas of a shopping mall is different at different times, but the existing technology is mostly unified, resulting in overheating or overcooling in some areas. For example, when no one is working in the office area during lunch time, the system still operates in the normal mode. According to research statistics, this rigid operation mode leads to energy waste of 20%-40%. When the number of people in the room changes suddenly, due to thermodynamic inertia, the HVAC equipment cannot respond quickly. For example, if a meeting is suddenly held in the conference room, the equipment needs to run for a period of time before the temperature is adjusted to a comfortable range, resulting in a decrease in human comfort. At the same time, the traditional system performs reactive control based on current data, lacks prediction of future demand, resulting in response delays and energy waste. For example, a few minutes before the end of the meeting, the equipment is still running at a relatively high power, and it is impossible to predict power reduction.

[0003] The existing control method is based on fixed manually set parameters, lacks a dynamic optimization mechanism, and is difficult to adapt to long-term environmental changes. For example, the flow of people in shopping malls is different on holidays and weekdays, and the rigidity of HVAC control methods leads to energy waste due to over-regulation or insufficient comfort due to insufficient regulation. In addition, traditional crowd prediction methods are mostly single prediction models, and the error between the predicted value and the actual value is large. Summary of the invention

[0004] The present application provides an HVAC intelligent control method and system based on personnel dynamic prediction, which is used to solve the technical problem that the existing HVAC equipment control method easily causes energy waste.

[0005] In one aspect, the present application provides a method for intelligent HVAC control based on personnel dynamic prediction, the method comprising the following steps:

[0006] Step S1: deploy multiple monitoring nodes in the building control area through the perception layer to collect the number of people and environmental parameter data in the area in real time, and transmit the data to the comprehensive control module in a standardized format;

[0007] Step S2: Based on the data collected by the perception layer, the prediction layer predicts the number of people in the short-term dimension and the historical dimension respectively; wherein, in the short-term dimension, the LSTM model is used to predict the number of people in the second preset time period in the future based on the data of the first preset time period in the past, and the historical dimension predicts the number of people in the same period through time clustering and linear regression models, and dynamically allocates weights according to the prediction error, and outputs the weighted fusion comprehensive prediction number;

[0008] Step S3: The decision layer calculates the target temperature setting value based on the real-time number of people and the comprehensive predicted number of people, combined with the thermal comfort temperature and the energy-saving temperature, and generates regional HVAC equipment control instructions based on the PID control algorithm;

[0009] Step S4: adjusting the HVAC equipment operating parameters of the corresponding area according to the regional HVAC equipment control instructions through the execution layer, and synchronously calculating the total air-conditioning load of the building to adjust the power of the air-conditioning host;

[0010] Step S5: Analyze historical operation data through the evaluation layer to optimize control parameters, detect abnormal data and generate fault diagnosis reports to achieve iterative optimization.

[0011] In one implementation of the present application, in step S1, each node includes a visual sensor, an infrared sensor, a CO2 sensor, and a temperature and humidity sensor; the deployment of the monitoring nodes specifically includes: evenly arranging them on the ceiling of the control area in units of preset areas to ensure full coverage; hierarchical spatial encoding of the building area, and synchronizing the clocks of each sensor through the NTP protocol to set the timestamp error.

[0012] In one implementation of the present application, in step S2, the process of short-term dimensional prediction specifically includes: performing linear interpolation of missing values ​​and elimination of outliers on sensor data; standardizing the number of personnel using the Z-score normalization method, and inputting the LSTM model for training, wherein the model structure includes an input layer, a 32-neuron LSTM layer, and a fully connected output layer; training the model with the goal of minimizing MAE, and outputting the predicted number of people in the second time period in the future after denormalization.

[0013] In one implementation of the present application, in step S2, the process of historical dimension prediction specifically includes: clustering timestamps into Monday to Sunday and holidays, and associating historical meteorological data to filter special weather periods; encoding historical data of the same category by minutes, and inputting a linear regression model to output the predicted number of people in the second time period in the future.

[0014] In one implementation of the present application, in step S2, the weight allocation formula is:

[0015]

[0016] ω 历史维度 =1-ω 短期维度

[0017] Among them, a is the short-term forecast error and b is the historical forecast error.

[0018] In one implementation of the present application, in step S3, the target temperature setting value is calculated as follows:

[0019]

[0020] Among them, T 设定 The target temperature set for the area based on energy saving and comfort goals; T 舒适 is the standard operating temperature of the functional area, which is the first-level thermal comfort level range; T 节能 N is the energy-saving temperature, which is the lower limit of the temperature range of the second-level thermal comfort level; 综合 is the combined value of the real-time number of people and the predicted number of people; k is the ratio of the trigger area adjusted to a comfortable temperature; N 基础 Design the area for capacity, or the number of people who will typically use it.

[0021] In one implementation of the present application, in step S5, the process of anomaly detection specifically includes: setting the threshold range of the number of personnel, environmental parameters and equipment operating parameters; analyzing the abnormal data characteristics through an association rule algorithm, locating the root cause of sensor failure or network interruption, and automatically generating a diagnostic report.

[0022] In one implementation of the present application, the method also includes: when the area enters the manual control mode and lasts for a third preset time period, automatically switching back to the intelligent control mode; if the unmanned state of the area continues for more than a threshold time, triggering the energy-saving temperature setting.

[0023] The present application also provides an HVAC intelligent control system based on dynamic personnel prediction, and the system includes: a perception module, which is composed of multiple monitoring nodes, each node integrates a visual sensor, an infrared sensor, a CO2 sensor and a temperature and humidity sensor, and is used to collect the number of people and environmental parameter data in real time; a prediction module, which is deployed on the edge computing node, and includes a short-term prediction unit and a historical prediction unit, and outputs a weighted fusion comprehensive predicted number of people through an LSTM model and a linear regression model respectively; a decision module, which calculates the target temperature setting value based on real-time data and prediction results, and generates control instructions through a PID algorithm; an execution module, which adjusts the operating parameters of the regional HVAC equipment according to the instructions, and dynamically adjusts the total power of the air conditioner; an evaluation module, which uses machine learning to optimize the control parameters, detect abnormal data and generate fault reports.

[0024] In one implementation of the present application, the data transmission rules of the perception module are: real-time upload is triggered when the number of people changes, and temperature and humidity data are uploaded at a fixed frequency; the transmission format is: spatial coding-timestamp-number of people-temperature-humidity, and the data is compressed and transmitted after noise reduction by Kalman filtering.

[0025] The present application provides an HVAC intelligent control method and system based on personnel dynamic prediction, which has the following beneficial effects:

[0026] (1) By combining short-term dimension prediction with historical dimension prediction, the future number of people is predicted in multiple dimensions. At the same time, the weight of the dimension is changed according to the error with the actual number of people, the predicted value of the dimension with high error is reduced, and the predicted number of people is corrected, thus solving the problem of excessive error caused by the single prediction model of the original prediction method.

[0027] (2) Based on the real-time number of people and environmental parameters, combined with future demand forecasts, the air conditioning operation mode is adjusted in advance to avoid excessive cooling or heating and improve the energy-saving efficiency of the system.

[0028] (3) A weighted comprehensive control strategy is adopted to dynamically adjust the temperature setting value according to the number of people, ensuring that high-density areas prioritize comfort and low-density areas focus on energy conservation, thus achieving on-demand adjustment.

[0029] (4) By predicting future personnel, the temperature of the area is adjusted in advance, offsetting the lag caused by real-time control of thermal inertia. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0031] Figure 1 A flow chart of an HVAC intelligent control method based on personnel dynamic prediction provided in an embodiment of the present application;

[0032] Figure 2 The overall logic flow chart provided for the embodiment of the present application;

[0033] Figure 3 A composition diagram of an HVAC intelligent control system based on personnel dynamic prediction provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0035] The embodiment of the present application provides an HVAC intelligent control method and system based on dynamic personnel prediction. The technical solution proposed in the embodiment of the present application is described in detail below with the help of the accompanying drawings.

[0036] Figure 1 A flow chart of an HVAC intelligent control method based on personnel dynamic prediction provided in an embodiment of the present application. Figure 1 As shown, the method mainly includes the following steps:

[0037] Step S1: Deploy multiple monitoring nodes in the building control area through the perception layer to collect real-time data on the number of people and environmental parameters in the area, and transmit the data to the comprehensive control module in a standardized format.

[0038] In the embodiment of the present application, the sensor deployment and installation setting is: according to the building structure, the preset area can be 1m 2 or 4m 2 (The specific value of the preset area is not limited in this manual and other values ​​may be used) as a node evenly placed in the ceiling of the control area (the control area is the control range of the controllable unit HVAC equipment) to ensure that the node fully covers the area to be controlled. Each node is pre-set with a dynamic monitoring system, which includes visual sensors (to achieve anonymous headcount), infrared sensors (to assist in calibrating the number of personnel), CO2 sensors, and temperature and humidity sensors to ensure all-round perception of the number of personnel and environmental parameters in the area, and time alignment of each sensor (all sensor clocks can be unified based on the NTP protocol or manual synchronization to ensure that the data timestamp error is ≤1s).

[0039] Furthermore, a global monitoring network is built, specifically: hierarchical spatial coding of building areas, the coding form is building-floor-measurement and control module, such as B1-F2-Z3. This coding rule cleverly implies the spatial proximity relationship, for example, the areas represented by Z3 and Z4 are adjacent to each other in space. This coding method facilitates the precise positioning of data during system operation, and is also conducive to the coordinated control between areas.

[0040] Furthermore, the data processing and transmission process is as follows: standardize and de-noise the monitoring data (such as Kalman filtering), compress and transmit the data to the integrated control system according to different transmission rules based on different regions and different sensor data. The transmission rules are as follows: (1) Number of people: changes in the number of people in the region trigger real-time upload. (2) Temperature and humidity: upload data in real time. (3) Transmission format: {spatial coding-time coding-number of people-temperature-humidity}.

[0041] Step S2: Based on the collected data of the perception layer, the prediction layer predicts the number of people in the short-term dimension and the historical dimension respectively; wherein, in the short-term dimension, the LSTM model is used to predict the number of people in the second preset time period in the future based on the data of the first preset time period in the past, and the historical dimension predicts the number of people in the same period through time clustering and linear regression models, and dynamically allocates weights according to the prediction error, and outputs the weighted fusion comprehensive prediction number.

[0042] In an embodiment of the present application, in order to reduce costs or reduce network delays caused by large amounts of data transmission, the prediction layer can be set at the edge computing node, and the number of people in each area can be predicted independently for each area.

[0043] Specifically, the data preprocessing for short-term dimension prediction is as follows: (1) extracting timestamps and personnel quantity data from the data collected by sensors. For missing data, if the data for a certain minute is missing, the linear interpolation method of the two points before and after is used to fill it. For outliers, the box plot (IQR method) is used to identify and eliminate them, and then the data is completed by interpolation. (2) The timestamp is converted into a continuous numerical feature, which is specifically encoded as minutes from 0 to 59 in chronological order. These values ​​serve as the time dimension identifier of the model input, providing an intuitive time reference for the model when analyzing the law of changes in the number of personnel over time. (3) In the data processing stage, the Z-score normalization method is used to standardize the personnel quantity data. Through this method, the personnel quantity data is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1. The purpose of this processing is to eliminate the dimensional differences between different data, improve the stability of the model training process, and avoid poor model training results due to data scale problems.

[0044] Furthermore, in the process of model construction and training, data preparation is first performed, historical data from the past three months are selected, and training samples are generated in a sliding time window manner. Each sample is input by the number of personnel sequence in the past 60 minutes (the specific value of the first preset time period is not limited in this specification, and 30 minutes or 45 minutes can also be taken) and the actual number of personnel in the next 10 minutes (the specific value of the second preset time period is not limited in this specification, and 15 minutes or 5 minutes can also be taken) as output.

[0045] Furthermore, the embodiment of the present application adopts a long short-term memory network (LSTM) model structure, wherein the input layer is responsible for receiving the personnel quantity sequence data of the past 60 minutes. LSTM layer: 1 layer of LSTM units is set, each layer contains 32 neurons, so as to fully capture the long-term dependencies in the time series. Output layer: a fully connected layer is used to process the data output by the LSTM layer and output the predicted value of the number of personnel in the next 5 minutes.

[0046] Furthermore, during model training, the optimization goal is to minimize the mean absolute error (MAE) between the predicted value and the actual value as the optimization goal of the model. Training cycle and batch: The training cycle is set to 100 rounds, and each batch of training contains 32 samples. Validation set setting: 10% of the data is reserved as the validation set to monitor overfitting during model training.

[0047] In the embodiment of the present application, the values ​​of the number of layers and the number of neurons of the LSTM neural network are only provided as a reference, and the specific number of layers and the number of neurons are not specifically limited, and other values ​​may also be used.

[0048] Furthermore, when making real-time predictions, (1) Input data: Obtain the number of people sequence for the past 60 minutes from the perception layer in real time, with the data format of N_{t-59}, N_{t-58}, …, N_t. (2) Prediction output: Input the obtained number of people sequence into the trained model, and the model outputs the predicted number of people for the next 10 minutes, with the format of N_{t+1}, N_{t+2}, …, N_{t+10}. (3) Denormalization: Since the data input into the model has been normalized by Z-score, the predicted value needs to be converted from the Z-score range back to the predicted number of people. Output the short-term prediction result, i.e., the number of people in the next 10 minutes.

[0049] It should be noted that the historical dimension database needs to be updated periodically and only retains data from the past year.

[0050] In the embodiment of the present application, the prediction of the historical dimension specifically includes:

[0051] (1) Time clustering: cluster timestamps into the same group of data sets. The data sets are divided into 8 categories, Monday to Sunday, and holidays, and coded 1-8. Priority is given to whether it is a holiday. For example, if the Dragon Boat Festival falls on Monday, the timestamp is grouped as a holiday.

[0052] (2) Data selection: For each time cluster, select the time data of each sensor in the past year. (In the early stage of system operation, if the data in the time cluster set is insufficient, the data of the past 6 months can be selected. If it is less than 6 months, the time cluster set will not be used for historical dimension prediction. If the system only runs for 4 weeks, or runs for 6 weeks but the data in the holiday cluster set is less than 20 times, then the data will only be collected and not used as historical reference).

[0053] (3) Data screening: Use the meteorological data API to obtain historical meteorological information. Based on this information, the data during special weather periods (such as rain and haze) will be filtered. The purpose is to ensure that when making historical personnel forecasts, the reference data are all from typical scenarios, avoiding abnormal factors such as special weather from interfering with the forecast results, thereby improving the accuracy of the forecast.

[0054] (4) Data compression and import: The historical number of personnel is divided into a node for every 1 minute, coded as 1-840 (assuming the system operates 14 hours a day), the number of personnel in each time period is calculated and aligned with the same code in the same group.

[0055] (5) Personnel prediction output: Take the same coded data and embed it into the linear regression model (such as 8-15, which is the 15th minute of the holiday system operation) to output the predicted number of people in the historical dimension in the next 10 minutes.

[0056] Furthermore, the predicted number of people is output based on the weights. First, the weight factors include the short-term prediction error a: the average absolute error between the predicted value in the past time window and the current actual value. The historical prediction error b: the average absolute error between the predicted value in the same period in history and the current actual value.

[0057] The weight formula is as follows:

[0058]

[0059] ω 历史维度 =1-ω 短期维度

[0060] The smaller the error, the higher the weight.

[0061] The weighted fusion formula is as follows:

[0062] Y 未来5分钟预测人数 =ω 短期维度 ×Y 短期预测人数 +ω 历史维度 ×Y 长期预测人数

[0063] Then output the predicted number of people per minute in the next 10 minutes.

[0064] Step S3: The decision layer calculates the target temperature setting value based on the real-time number of people and the comprehensive predicted number of people, combined with the thermal comfort temperature and the energy-saving temperature, and generates regional HVAC equipment control instructions based on the PID control algorithm.

[0065] In the embodiment of the present application, the core formula for temperature adjustment setting of the decision layer is as follows:

[0066]

[0067] T 设定 The target temperature set for the area based on energy saving and comfort goals; T 舒适 The standard operating temperature of the functional area is 25℃ in summer and 23℃ in winter, which is the first-level thermal comfort level range; T 节能 The energy-saving temperature can be set to 28°C in summer and 18°C ​​in winter, which is the lower limit of the temperature range of the second-level thermal comfort level; N 综合 is the combined value of the real-time number of people and the predicted number of people; k is the ratio of the trigger area adjusted to a comfortable temperature; N 基础 Design capacity or number of regular users for an area. 设定 The specific value range of the value should be within T 舒适 and T 节能 When N 综合 = k·N 基础 When the system directly executes T 舒适 .

[0068]

[0069] Among them, ω 当前 is the weight of the current number of people and temperature (can be set to 0.6), N 当前 is the current number of people, ω i The weight of the number of people predicted for the temperature in the future i-th minute; N i is the number of people in the i-th minute, N 基础 is the design capacity of the area or the number of regular users (such as the number of seats in a conference room), k is the proportion of the trigger area adjusted to a comfortable temperature (which can be set to 15%), for example, the regular use of the classroom area is N 基础 = 40 people, k = 15%, the number of people who trigger the area to adjust to a comfortable temperature k·N 基础 =6 people, then when N 综合 When ≥6 people, T 设定 =T 舒适 .

[0070]

[0071] Among them, e -i / τIt is an exponential decay weight, that is, the weight of the predicted number of people decreases rapidly in the early stage of the time window and tends to be stable in the later stage, emphasizing that the recent prediction weight is significant. τ is the exponential decay coefficient, that is, the speed at which the prediction weight decays over time. It can be set according to the thermodynamic inertia and is initially set to 3. If the thermal inertia is large, τ can be increased as appropriate.

[0072] also, is the normalization coefficient, which forces the total weight of the current number of people and the predicted number of people to be 1.

[0073] In the embodiment of the present application, a comfort compensation mechanism is also set. A manual mode can be set, that is, the temperature is manually adjusted by personnel. When the area enters the unmanned mode or the manual mode lasts for two hours, it re-enters the intelligent control mode to compensate for the comfort problem of being at the edge of the comfortable temperature for a long time when there are very few people. The regional control instruction is based on the traditional PID control method to issue a regional HVAC equipment control instruction to make the regional temperature reach T 设定 The total power of the air conditioner is calculated by summing up the loads in each area to calculate the total load of the area to be regulated, and a control command is issued to the air conditioner host according to the total load.

[0074] Step S4: The execution layer adjusts the operating parameters of the HVAC equipment in the corresponding area according to the regional HVAC equipment control instructions, and simultaneously calculates the total air-conditioning load of the building to adjust the power of the air-conditioning host.

[0075] In the embodiment of the present application, the HVAC equipment (fan frequency, valve opening, total air conditioning power) corresponding to the area is adjusted according to the decision instructions.

[0076] Step S5: Analyze historical operation data through the evaluation layer to optimize control parameters, detect abnormal data and generate fault diagnosis reports to achieve iterative optimization.

[0077] In the embodiment of the present application, the evaluation layer uses existing AI technology to set normal range thresholds for data such as the number of personnel, environmental parameters, fan frequency, valve opening, etc., detects data at all levels, and marks it as abnormal data if it exceeds the threshold. When evaluating the operating status of the system and troubleshooting, an association rule algorithm is used to conduct an in-depth analysis of the abnormal data characteristics. Through this algorithm, the source of the fault can be quickly and accurately located, such as determining whether a sensor failure or a network interruption occurs. In addition, the system will automatically generate a detailed diagnostic report to provide strong support for subsequent maintenance and system optimization.

[0078] Calculate the temperature control accuracy, energy consumption and other evaluation indicators based on the system's historical operation data, use machine learning algorithms to analyze their relationship with the control parameters, and optimize the decision-making layer's control instructions, such as adjusting the valve opening control parameters. Evaluate the personnel quantity prediction model at the prediction layer, optimize the prediction data processing method of each historical dimension, and improve the prediction accuracy.

[0079] In the embodiment of the present application, the overall control flow chart is as follows: Figure 2 shown.

[0080] The above is an HVAC intelligent control method based on personnel dynamic prediction provided by the embodiment of the present application. Based on the same inventive concept, the embodiment of the present application also provides an HVAC intelligent control system based on personnel dynamic prediction. Figure 3 A composition diagram of an HVAC intelligent control system based on personnel dynamic prediction provided in an embodiment of the present application, such as Figure 3 As shown, the system mainly includes: a perception module 301, which is composed of multiple monitoring nodes, each node integrates a visual sensor, an infrared sensor, a CO2 sensor and a temperature and humidity sensor, and is used to collect the number of people and environmental parameter data in real time; a prediction module 302, which is deployed on the edge computing node, and includes a short-term prediction unit and a historical prediction unit, and outputs a weighted fusion comprehensive predicted number of people through an LSTM model and a linear regression model respectively; a decision module 303, which calculates the target temperature setting value based on real-time data and prediction results, and generates control instructions through a PID algorithm; an execution module 304, which adjusts the operating parameters of the regional HVAC equipment according to the instructions, and dynamically adjusts the total power of the air conditioner; an evaluation module 305, which uses machine learning to optimize the control parameters, detect abnormal data and generate a fault report.

[0081] Furthermore, the data transmission rules of the perception module are as follows: real-time upload is triggered when the number of people changes, and temperature and humidity data are uploaded at a fixed frequency; the transmission format is: spatial coding-timestamp-number of people-temperature-humidity, and the data is compressed and transmitted after noise reduction by Kalman filtering.

[0082] The present application provides an HVAC intelligent control method and system based on dynamic personnel prediction. By combining short-term dimension prediction with historical dimension prediction, the future number of people is predicted in multiple dimensions. At the same time, the weight of the dimension is changed according to the error with the actual number of people, the predicted value of the dimension with higher error is reduced, and the predicted number of people is corrected, which solves the problem of excessive error caused by the single prediction model of the original prediction method. Based on the real-time number of people and environmental parameters, combined with future demand prediction, the air-conditioning operation mode is adjusted in advance to avoid excessive cooling or heating, and improve the energy-saving efficiency of the system. A weighted comprehensive control strategy is adopted to dynamically adjust the temperature setting value according to the number of people, ensuring that high-density areas give priority to comfort, and low-density areas focus on energy saving, realizing on-demand adjustment. By predicting future personnel, the temperature of the area is adjusted in advance, offsetting the lag caused by real-time control of thermal inertia.

[0083] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0084] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0085] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A HVAC intelligent control method based on personnel dynamic prediction, characterized in that: The method comprises the following steps: Step S1: deploy multiple monitoring nodes in the building control area through the perception layer to collect the number of people and environmental parameter data in the area in real time, and transmit the data to the comprehensive control module in a standardized format; Step S2: Based on the data collected by the perception layer, the prediction layer predicts the number of people in the short-term dimension and the historical dimension respectively; wherein, in the short-term dimension, the LSTM model is used to predict the number of people in the second preset time period in the future based on the data of the first preset time period in the past, and the historical dimension predicts the number of people in the same period through time clustering and linear regression models, and dynamically allocates weights according to the prediction error, and outputs the weighted fusion comprehensive prediction number; Step S3: The decision layer calculates the target temperature setting value based on the real-time number of people and the comprehensive predicted number of people, combined with the thermal comfort temperature and the energy-saving temperature, and generates regional HVAC equipment control instructions based on the PID control algorithm; Step S4: adjusting the HVAC equipment operating parameters of the corresponding area according to the regional HVAC equipment control instructions through the execution layer, and synchronously calculating the total air-conditioning load of the building to adjust the power of the air-conditioning host; Step S5: Analyze historical operation data through the evaluation layer to optimize control parameters, detect abnormal data and generate fault diagnosis reports to achieve iterative optimization.

2. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: In step S1, each node includes a visual sensor, an infrared sensor, a CO2 sensor, and a temperature and humidity sensor; the deployment of the monitoring nodes specifically includes: evenly arranging them on the ceiling of the control area in units of preset areas to ensure full coverage; hierarchical spatial encoding of the building area, and synchronizing the clocks of each sensor through the NTP protocol to set the timestamp error.

3. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: In step S2, the process of short-term dimension prediction specifically includes: Perform linear interpolation of missing values ​​and elimination of outliers on sensor data; The Z-score normalization method is used to standardize the number of personnel and input into the LSTM model training. The model structure includes an input layer, a 32-neuron LSTM layer, and a fully connected output layer; The model is trained with the goal of minimizing MAE, and the predicted number of people in the second time period in the future is output after denormalization.

4. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: In step S2, the process of historical dimension prediction specifically includes: Cluster timestamps into Monday to Sunday and holidays, and associate historical meteorological data to filter special weather periods; The historical data of the same category are encoded by minute and input into the linear regression model to output the predicted number of people in the second time period in the future.

5. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: In step S2, the weight allocation formula is: oh 历史维度 =1-h 短期维度 Among them, a is the short-term forecast error and b is the historical forecast error.

6. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: In step S3, the target temperature setting value is calculated as follows: Among them, T 设定 The target temperature set for the area based on energy saving and comfort goals; T 舒适 is the standard operating temperature of the functional area, which is the first-level thermal comfort level range; T 节能 N is the energy-saving temperature, which is the lower limit of the temperature range of the second-level thermal comfort level; 综合 is the combined value of the real-time number of people and the predicted number of people; k is the ratio of the trigger area adjusted to a comfortable temperature; N 基础 Design the area for capacity, or the number of people who will typically use it.

7. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: In step S5, the abnormality detection process specifically includes: Set threshold ranges for the number of personnel, environmental parameters, and equipment operating parameters; The association rule algorithm is used to analyze abnormal data characteristics, locate the root cause of sensor failure or network interruption, and automatically generate a diagnostic report.

8. The HVAC intelligent control method based on personnel dynamic prediction according to claim 1 is characterized in that: The method also includes: when the area enters the manual control mode and lasts for a third preset time period, automatically switching back to the intelligent control mode; if the unmanned state of the area lasts for more than a threshold time, triggering the energy-saving temperature setting.

9. An HVAC intelligent control system based on personnel dynamic prediction, characterized in that: The system comprises: The perception module is composed of multiple monitoring nodes. Each node integrates visual sensors, infrared sensors, CO2 sensors, and temperature and humidity sensors to collect data on the number of people and environmental parameters in real time. The prediction module is deployed on the edge computing node and includes a short-term prediction unit and a historical prediction unit. It outputs the weighted fusion comprehensive prediction number through the LSTM model and the linear regression model respectively. The decision module calculates the target temperature setting value based on real-time data and prediction results, and generates control instructions through the PID algorithm; The execution module adjusts the operating parameters of the regional HVAC equipment according to the instructions and dynamically adjusts the total power of the air conditioner; The evaluation module uses machine learning to optimize control parameters, detect abnormal data and generate fault reports.

10. The HVAC intelligent control system based on personnel dynamic prediction according to claim 9, characterized in that: The data transmission rules of the perception module are: real-time upload is triggered when the number of people changes, and temperature and humidity data are uploaded at a fixed frequency; the transmission format is: spatial coding-timestamp-number of people-temperature-humidity, and the data is compressed and transmitted after noise reduction by Kalman filtering.

Images