HVAC Control Method for Highway Service Areas Based on Internet of Things and Neural Networks

Abstract

This invention discloses a method for HVAC control in highway service areas based on the Internet of Things (IoT) and neural networks. The method includes: dividing the building area of ​​the target service area into functional zones; deploying IoT sensors in each functional zone to acquire environmental data, HVAC equipment parameters, and pedestrian flow distribution data; collecting vehicle information at the entrances and exits of the target service area and the number of people entering and exiting each functional zone; using the pedestrian flow distribution data, vehicle information, and number of people entering and exiting the current service area and the previous service area as a first dataset, inputting it into a pedestrian flow prediction neural network to generate corresponding pedestrian flow prediction data; using the environmental data, HVAC equipment parameters, pedestrian flow distribution data, and pedestrian flow prediction data of the current service area as a second dataset, inputting it into a control strategy neural network to output control strategies for each group of HVAC equipment in the current service area within a future preset time period; and controlling each group of HVAC equipment within the future preset time period according to the control strategies.

Description

Technical Field

[0001] This invention relates to the field of HVAC control technology, and in particular to a method for HVAC control in highway service areas based on the Internet of Things and neural networks. Background Technology

[0002] With the continuous expansion of the highway network, highway service areas have become important nodes connecting cities and rural areas and promoting transportation convenience. The function of service areas is not limited to providing rest and catering facilities; they also place higher demands on the comfort of drivers and passengers. Especially with frequent passenger flow, maintaining a stable and comfortable temperature within service areas has become a significant technical challenge.

[0003] Chinese invention patent CN118729501B discloses an HVAC (Heating, Ventilation, and Air Conditioning) optimization control method based on scene recognition and demand prediction. This method collects multi-source data in real time, uses an LSTM (Long Short-Term Memory) model for scene recognition and demand prediction, and then generates an optimal control strategy to dynamically adjust the operating parameters of the HVAC system, thereby improving system efficiency, reducing energy consumption, and achieving energy conservation and emission reduction goals. Although this solution performs well in the traditional building HVAC control field, there are still several aspects that can be improved when applied to the HVAC control of highway service area buildings, specifically including the following:

[0004] 1. Limitations of Data Acquisition: Existing solutions rely solely on current data from the target building for decision-making. However, in unique environments like highway service areas, vehicle and pedestrian flows exhibit strong temporal correlations. For instance, vehicles and pedestrians entering the previous service area are typically less likely to enter the current one, and even if they do enter, their stay in the previous service area is usually shorter. Therefore, comparing data from previous and current service areas can effectively eliminate irrelevant or noisy data. Furthermore, real-time and predicted data from the previous service area can not only serve as a reference for the current service area but also improve prediction accuracy and the effectiveness of strategy adjustments, thereby enhancing the overall response efficiency of the HVAC control system.

[0005] 2. Neglect of Regional Differences: Existing HVAC equipment adjustment methods are too uniform, lacking attention to the individualized needs of different functional areas. However, highway service areas typically have multiple functional areas (such as restrooms, dining areas, and rest areas), and the characteristics and needs of people staying in these areas vary considerably. Using a single predictive and strategy adjustment method may result in insufficient adaptability of equipment adjustments, failing to optimize for the actual needs of each area. Therefore, developing personalized control strategies based on the flow of people and temperature control requirements of different functional areas is crucial to ensure the accuracy and efficiency of the system.

[0006] 3. Limited Data Variety: Existing prediction models primarily rely on time-series feature data for forecasting, making conventional LSTM networks sufficient for most scenarios. However, the complexity of highway service areas exceeds the scope of simple time-series data analysis. LSTM models that solely rely on time-series data cannot fully explore and utilize diverse data features (such as personnel behavior patterns, external environmental changes, and equipment status). To improve prediction accuracy and reliability, it is recommended to incorporate more diverse data inputs, such as environmental monitoring data, historical operating status, and user behavior patterns, and combine this with multimodal learning methods to construct more complex neural network models.

[0007] 4. Disconnect between Real-Time Prediction and HVAC System Response Delay: Existing predictive and adjustment strategies largely rely on real-time conditions, i.e., control is applied at the current point in time. However, HVAC systems have a time delay in response, especially in temperature regulation, where adjustments require time to take effect. This means that relying solely on real-time data for control may lead to a disconnect between current adjustments and expected future states, impacting system stability and comfort. Therefore, when adjusting strategies, the response delay of HVAC equipment must be fully considered, employing future-predictive control strategies rather than solely relying on current data. This approach allows for proactive system optimization and adjustment, avoiding instability issues caused by excessive or insufficient real-time adjustments. Summary of the Invention

[0008] To address the aforementioned technical problems, this invention aims to optimize the operation of HVAC equipment through intelligent control strategies, resolving issues such as slow response, low energy efficiency, and unstable comfort levels in existing highway service area HVAC control systems. By integrating multi-source data, including environmental parameters, pedestrian flow, vehicle information, and data from the previous service area, and combining this with a neural network model, the operating parameters of the HVAC equipment are dynamically predicted and optimized to achieve efficient, precise, and energy-saving HVAC control.

[0009] To achieve the above-mentioned objectives, the technical solution provided by this invention includes:

[0010] A method for HVAC control in highway service areas based on the Internet of Things and neural networks, characterized by the following steps:

[0011] S1. Divide the building area of ​​the target service area into functional zones, deploy IoT sensors in each functional zone to acquire environmental data, HVAC equipment parameters and pedestrian flow data, and group the HVAC equipment in the same building according to the functional zone in which they are located.

[0012] S2. Image acquisition devices are installed at the entrances and exits of the target service area to collect information on vehicles entering and exiting; the vehicle information includes license plate and dwell time; infrared pedestrian counters are installed at the entrances and exits of each functional area of ​​the target service area to collect the number of people entering and exiting.

[0013] S3. Take the pedestrian distribution data, vehicle information and number of people entering and leaving the current service area and the previous service area in the direction of highway traffic flow as the first dataset, input it into the pre-trained pedestrian flow prediction neural network, predict the pedestrian flow of each functional area of ​​the target service area in the future preset time period, and generate the corresponding pedestrian flow prediction data.

[0014] S4. Use the current service area's environmental data, HVAC equipment parameters, pedestrian distribution data, and pedestrian prediction data as the second dataset, input them into the pre-trained control strategy neural network, and output the control strategy of each group of HVAC equipment in the current service area for the future preset time period.

[0015] A5. Control each group of HVAC equipment within a preset time period according to the control strategy.

[0016] Preferably, step S3, before inputting the first dataset into the pre-trained pedestrian flow prediction neural network, further includes the following step:

[0017] Remove vehicle information from the previous service area where the vehicle stay time exceeds the first preset threshold among vehicles continuously entering the current service area and the previous service area in the direction of highway traffic flow. Also, delete no more than 5 people's traffic data from the pedestrian distribution data of any functional area of ​​the current service area and the number of people entering and leaving the current service area.

[0018] Preferably, the pedestrian flow prediction neural network comprises the following sequentially connected components:

[0019] The multi-input layer includes several separate input layers set according to the input data of each category of the first dataset, and outputs the feature representation of the first dataset;

[0020] Temporal convolutional networks are used to extract short-term dependency features from the feature representation through convolutional dilation.

[0021] Long Short-Term Memory (LSTM) networks are used to extract long-term dependency features from the short-term dependency features through a gating mechanism.

[0022] The output layer is used to combine the output of the Long Short-Term Memory network into the human flow prediction results.

[0023] Preferably, the control strategy neural network includes:

[0024] The temporal input layer extracts temporal features from the pedestrian distribution data and pedestrian prediction data in the second dataset based on the long short-term neural network, and outputs a multi-dimensional temporal feature vector.

[0025] The static input layer normalizes the environmental data and HVAC equipment parameters in the second data and outputs a one-dimensional numerical feature vector.

[0026] The feature fusion layer, which is connected to the temporal input layer and the static input layer respectively, is used to concatenate the temporal feature vector and the numerical feature vector into an input feature vector.

[0027] Preferably, the control strategy neural network further includes:

[0028] The reinforcement learning module connected to the feature fusion layer, wherein the reward function R of the reinforcement learning module is... t for:

[0029] R t =α1·R c +α2·R e +α3·R s +α4·R t Where α1, α2, α3, and α4 are preset weight parameters, and R c R is the comfort reward parameter. e R is an energy efficiency bonus parameter. s As a reward for temperature control stability, R t This refers to the parameters for rewarding human traffic.

[0030] Preferably, the pre-training method includes: using the length of the preset time period as the extraction window for feature data and / or input feature vectors.

[0031] Beneficial effects

[0032] 1. Intelligent control strategy: By using a neural network model to predict pedestrian flow and optimize the control strategy of HVAC equipment, this invention can automatically adjust the operating status of HVAC equipment based on real-time and predicted data of the service area, ensuring that the HVAC system can be accurately adjusted according to demand and avoiding energy waste.

[0033] 2. Reducing Data Noise and Improving Prediction Accuracy: By combining traffic and pedestrian data from the previous service area, this invention effectively eliminates irrelevant noise data, thereby improving the accuracy of the prediction model. This method can further reduce the errors that may arise from relying solely on current data and increase the reliability of prediction and control strategies.

[0034] 3. Regionalized Precision Control: Addressing the varying pedestrian flow characteristics in different functional areas within highway service areas (such as restrooms, rest areas, and dining areas), this invention proposes a regionalized HVAC control strategy. Each functional area is intelligently adjusted according to its specific needs and pedestrian flow characteristics, achieving refined and efficient control of HVAC equipment.

[0035] 4. Effective control of response delay: Considering the time delay in the response of HVAC equipment, this invention adopts a control strategy based on future prediction, rather than relying solely on real-time data. This design can effectively avoid temperature control errors caused by the lag in real-time control strategy adjustments, thereby improving system stability and comfort. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of a highway service area HVAC control method based on the Internet of Things and neural networks provided in a preferred embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram of a neural network structure for predicting pedestrian flow provided in another preferred embodiment of the present invention;

[0038] Figure 3 This is a schematic diagram of the control strategy neural network structure provided in another preferred embodiment of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described below with reference to the accompanying drawings. In the description of this invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention.

[0040] Example 1

[0041] like Figure 1 As shown, this embodiment discloses a method for HVAC control in highway service areas based on the Internet of Things and neural networks, including the following steps:

[0042] S1. Divide the building areas within the target service area according to function, and deploy IoT sensors in each functional area to acquire environmental data, HVAC equipment parameters, and pedestrian flow distribution data. Group the HVAC equipment within the same building according to their functional area. The target service area refers to all service areas to be predicted located in a specific lane direction of a highway section. The current service area is one of the target service areas.

[0043] Those skilled in the art should know that, for highway service area buildings, the interior space can be divided into several relatively independent areas according to different functions, for example:

[0044] The restroom area mainly includes men's and women's restrooms and washrooms. Its HVAC requirements are characterized by the fact that people usually stay for a short time, but due to frequent entry and exit, it is more sensitive to changes in air quality and temperature.

[0045] Entrance / Exit Area: Primarily for pedestrian access. Its HVAC requirements are characterized by high pedestrian traffic and significant external climate influence on the indoor environment, making temperature control requirements more complex.

[0046] Dining area: Its HVAC requirements are characterized by the long time people spend in the dining area and the need for good air quality and comfortable temperature, thus requiring a large amount of HVAC equipment.

[0047] Rest area: Its HVAC requirements are similar to those of a dining area, but the flow of people may be relatively small, and the temperature control requirements are relatively stable.

[0048] Corridor area: Its HVAC requirements are characterized by frequent pedestrian traffic, so the temperature control requirements may be lower, but a certain level of comfort still needs to be maintained.

[0049] IoT sensors are deployed in different functional areas to collect real-time environmental data, HVAC equipment parameters, and pedestrian flow data. These sensors include, but are not limited to:

[0050] 1. Environmental data sensors: These include temperature sensors, humidity sensors, CO2 sensors, and air quality sensors, which can monitor indoor environmental parameters in various areas in real time. This data is crucial for intelligent HVAC control, helping the system understand environmental changes in each area in real time.

[0051] 2. HVAC Equipment Parameter Sensors: These sensors are used to collect real-time operating status data of HVAC equipment, including the working status, operating power, air speed, flow rate, and other parameters of equipment such as fans, air conditioners, and heaters. This data is used for subsequent control strategy optimization.

[0052] 3. Pedestrian Flow Sensors: By deploying infrared pedestrian counters, video surveillance and analysis equipment, etc., real-time data such as pedestrian flow and dwell time in each functional area are acquired. Changes in pedestrian flow directly affect the HVAC requirements of the area; therefore, pedestrian flow distribution data is a key factor in determining control strategies.

[0053] Within each functional area, HVAC equipment is grouped according to the area's functional requirements and pedestrian traffic characteristics. Different areas require different types of HVAC equipment, therefore, during control, the relevant equipment needs to be scheduled according to the specific functional area, and HVAC equipment within the same functional area is grouped according to its area affiliation. For example, the HVAC equipment in the restroom area may focus on air exchange and temperature regulation, while the dining area requires a more efficient air conditioning system to maintain suitable temperature and humidity. Based on the needs of each area, the system can adjust the operating status of each device in real time.

[0054] S2. Image acquisition devices are installed at the entrances and exits of the target service area to collect information on vehicles entering and exiting; the vehicle information includes license plate and dwell time; infrared pedestrian counters are installed at the entrances and exits of each functional area of ​​the target service area to collect the number of people entering and exiting.

[0055] Image acquisition devices (such as high-definition cameras and license plate recognition systems) are installed at the entrances and exits of the target service area to collect vehicle information entering and leaving the service area, including license plate number, entry time, and dwell time. The vehicle information entering the service area is compared and correlated with data from the previous service area to analyze vehicle flow patterns and dwelling patterns. Typically, vehicles and personnel entering the previous service area have a low probability of entering the current service area. Even if they do enter the current service area, their dwell time in the previous service area is usually shorter than their dwell time in the current service area. This pattern indicates that vehicles or personnel who spend a long time in the previous service area are likely to choose to leave the current service area quickly, thus having a smaller impact on the current service area, and are considered "low-correlation" data. Therefore, in some preferred embodiments, preprocessing is performed in the following manner:

[0056] Vehicles continuously entering the current service area and those staying in the previous service area for more than a first preset threshold are removed. Furthermore, passenger flow data (number of people not exceeding 5) is deleted from the passenger flow distribution data in any functional area of ​​the current service area and the number of people entering and exiting the current service area. This operation effectively removes vehicles with minimal impact on passenger flow in the current service area, thereby eliminating accompanying passengers and reducing interference from irrelevant data with subsequent prediction and control strategies. It should be noted that most vehicles in highway service areas are ordinary vehicles, typically private cars or small passenger vehicles, with a passenger capacity generally not exceeding 5 people. The actual number of passengers may be lower than the capacity, but usually will not exceed 5. Since it is difficult to monitor which functional area accompanying passengers are active in (it is impossible to accurately determine whether these people have entered the dining area, rest area, or are simply staying in the passageway), this embodiment uses random deletion for data processing. On the one hand, it avoids the deviations that may be caused by fixed-pattern deletion methods, ensuring the naturalness and rationality of the deletion process. On the other hand, it retains other important data, reduces the interference of noise on data modeling, and avoids data redundancy caused by a small number of short-term visitors while ensuring that the actual flow prediction of people in the service area is not affected.

[0057] S3. Take the pedestrian flow distribution data, vehicle information and number of people entering and leaving the current service area and the previous service area in the direction of highway traffic flow as the first dataset, input it into the pre-trained pedestrian flow prediction neural network, predict the pedestrian flow of each functional area of ​​the target service area in the future preset time period, and generate the corresponding pedestrian flow prediction data.

[0058] It should be understood that, since the vehicles and people entering the target service area usually follow a certain pattern in terms of time, the behavior of people staying in the previous service area has a strong reference value for predicting the behavior of people staying in the current service area (that is, the activity patterns of people in consecutive service areas are roughly the same or similar). Combining the data from the previous and next service areas helps to reduce data noise and improve the accuracy of people flow prediction.

[0059] For neural networks used for predicting pedestrian flow, the input data has multiple dimensions and includes both long-term and short-term time-series features, such as... Figure 2 As shown, in some preferred embodiments, a pedestrian flow prediction neural network is constructed as follows, specifically including:

[0060] The multi-input layer comprises several separate input layers configured according to each category of input data in the first dataset, outputting the feature representation of the first dataset. Considering the diversity of data sources (environmental data, vehicle information, pedestrian flow data), a multi-input neural network structure is constructed. Each input can be processed through different channels to capture features of different data types. For example: the vehicle information input layer extracts features related to vehicle flow patterns; the environmental data input layer processes environmental factors affecting indoor pedestrian flow, such as temperature and humidity; and the pedestrian flow data input layer specifically processes pedestrian mobility and density information. The multi-input layer receives multiple different types of input data (such as environmental data, vehicle information, pedestrian flow distribution data, etc.) and passes each type of input data to the corresponding sub-network for processing. Each input layer independently processes its own data category, avoiding interference between different types of data and efficiently extracting features.

[0061] Temporal convolutional networks are used to extract short-term dependency features from the feature representation through convolutional dilation. The convolutional dilation operation expands the receptive field, enabling it to capture data features over long time spans with relatively low computational overhead. Compared to traditional methods, it can better handle long-term dependencies in time-series data, and is particularly suitable for processing complex time series data. Through this step, the network can identify rapid changes in the short term, such as short-term fluctuations in population movement within a functional area.

[0062] Long Short-Term Memory (LSTM) networks are used to extract long-term dependency features from short-term dependency features through gating mechanisms. They are used to process long-term dependency information in time series, such as the impact of long-term patterns like seasonal variations and holidays on pedestrian flow prediction. Gating mechanisms can maintain long-term memory by dynamically controlling the "forgetting" and "remembering" of information flow, making them particularly suitable for processing long-term pedestrian flow patterns. For example, they can handle periodic trends in special time periods such as holidays or peak periods spanning several hours or days.

[0063] The output layer combines the outputs of the Long Short-Term Memory network with the pedestrian flow prediction results. This layer is typically a fully connected layer, and its output is a probability value (for classification tasks). These prediction results are then fed as input to the downstream control policy neural network to adjust the control strategy of the HVAC equipment within the service area.

[0064] Clearly, the aforementioned neural network for predicting pedestrian flow can integrate different types of data sources and effectively handle local and long-term dependencies in time-series data.

[0065] S4. Use the environmental data, HVAC equipment parameters, pedestrian distribution data, and pedestrian prediction data of the current service area as the second dataset, input them into the pre-trained control strategy neural network, and output the control strategy of each group of HVAC equipment in the current service area for the future preset time period.

[0066] Environmental data includes temperature, humidity, air quality, wind speed, and other data for each functional area of ​​the current service area. This data helps to assess the current comfort level and provides environmental background information for control strategies.

[0067] HVAC equipment parameters include the operating status, power, and temperature setpoints of HVAC equipment in each zone. These parameters reflect the current status and equipment capabilities of the HVAC system. This data helps determine whether the operating status of each piece of equipment needs to be adjusted to achieve optimal energy efficiency and comfort.

[0068] The pedestrian flow distribution data reflects the distribution of people in each functional area. The distribution of people in different areas directly affects the temperature control requirements of those areas, so it needs to be used as input information to help the model make accurate control decisions.

[0069] Based on the previous prediction results, the pedestrian flow forecast data shows the flow of people in each functional area over a preset future time period. This data enables control strategies to anticipate future changes in pedestrian flow and adjust the operating modes of HVAC equipment in advance.

[0070] The control strategy neural network integrates all the aforementioned data to achieve intelligent control of HVAC equipment. Specifically, it receives information from the second dataset, performs calculations through several layers of neural networks, and generates a control strategy based on preset objectives (such as comfort, energy efficiency, and energy saving). Specific control strategies include, but are not limited to: the operating modes of HVAC equipment in each functional area (such as on / off status, power regulation, etc.); and specific adjustment values ​​for equipment operating time and power output.

[0071] The second dataset contains both time-series data with significant temporal dependencies (such as pedestrian flow data, temperature and humidity changes), and static data without obvious temporal dependencies (such as equipment status, environmental data). While using a single input layer for fusion can simplify the structure and increase data uniformity, it can easily lead to data loss, resulting in poor feature extraction. Furthermore, it makes it difficult to distinguish data types and perform specific optimizations for different data types. Therefore, in some preferred embodiments, such as... Figure 3 As shown, consider using a method that partially employs multiple input layers and partially directly fuses them to construct the input layer of the control policy neural network, specifically including:

[0072] The temporal input layer extracts temporal features from the pedestrian distribution data and pedestrian prediction data in the second dataset based on long short-term neural networks, and outputs a multi-dimensional temporal feature vector. The main function of this layer is to process and extract the temporal features in the input data. These temporal data reflect the dynamic changes in pedestrian flow and are crucial to the control strategy of HVAC equipment, because changes in the number of people are closely related to the regional temperature control requirements.

[0073] The static input layer normalizes the environmental data and HVAC equipment parameters in the second data set and outputs a one-dimensional numerical feature vector. These data are usually static, meaning they do not change over time, but they are crucial for the decision-making of the control strategy.

[0074] The feature fusion layer, connected to the temporal input layer and the static input layer respectively, is used to concatenate the temporal feature vector and the numerical feature vector into an input feature vector. The concatenation can be performed using a simple concatenation operation, or it can be done using weighted summation, average pooling, etc. The specific method can be optimized by those skilled in the art from existing technologies; this invention does not impose further limitations. The second dataset processed by this method, on the one hand, reduces interference between data by processing temporal and static data separately, allowing each data type to use the most suitable processing method, thereby improving model performance. On the other hand, the processing method and hyperparameters for each type of data can be adjusted individually, making the model more tunable and interpretable.

[0075] For control strategy tasks, since control involves real-time optimization and adjustment of strategies, reinforcement learning (RL) methods are considered in some preferred embodiments to further enhance the network's learning capabilities. A reinforcement learning module is designed to control HVAC equipment in various areas, enabling it to make optimal decisions under changing environmental and pedestrian conditions. The goal of the reinforcement learning module is to dynamically adjust the control strategy based on current environmental data and states to improve the overall efficiency and comfort of the system. Those skilled in the art will understand that the reward function is the most important component in reinforcement learning, determining the merits of an agent's action in a given state. A reasonable reward design can guide the reinforcement learning agent to select the correct control strategy. In other preferred embodiments, the reward function R of the reinforcement learning module... t for:

[0076] R t =α1·R c +α2·R e +α3·R s +α4·R t ; where α1, α2, α3 and α4 are preset weight parameters, and the importance of each part can be adjusted by those skilled in the art according to actual needs;

[0077] R c This is a comfort reward parameter, measuring whether the temperature in each functional area meets comfort requirements within a certain time period. In the future, the system needs to maintain the temperature within a predetermined comfort range. Large temperature fluctuations will result in a lower comfort score.

[0078] Re The energy efficiency reward parameter evaluates the energy efficiency performance of the current control strategy. The energy efficiency reward is inversely proportional to the energy consumption of HVAC equipment; that is, the higher the energy efficiency (the lower the energy consumption), the higher the reward. By analyzing the relationship between the power consumption of HVAC equipment and the temperature control requirements in future time periods, the reward system should encourage the reduction of unnecessary energy consumption.

[0079] R s To reward temperature control stability, the stability of the temperature control system is measured, i.e., whether the temperature remains within a stable range across different time periods and regions. To avoid frequent temperature fluctuations, the system should encourage stable temperature control strategies. Reinforcement learning should avoid over-adjusting the system due to short-term temperature changes.

[0080] R t The reward function should take into account the impact of pedestrian traffic on temperature control, as pedestrian traffic forecasting directly affects the control strategy of the HVAC system. Areas with high pedestrian traffic require more energy support, while areas with low pedestrian traffic should reduce energy consumption.

[0081] The reinforcement learning module is directly connected to the aforementioned feature fusion layer. The multi-dimensional input feature vectors provided by the feature fusion layer, including time-series data (such as pedestrian flow distribution data and prediction data) and static data (such as environmental data and equipment parameters), form the basis for the reinforcement learning module's decision-making. Based on this, the reinforcement learning module dynamically adjusts its control strategy according to the feedback from the received reward function.

[0082] S5. Control the HVAC equipment in each train set within a preset time period according to the control strategy. Controlling the HVAC equipment in each train set includes, but is not limited to, the following aspects:

[0083] 1. Real-time adjustment of equipment parameters: Based on the control strategy output, the operating status of HVAC equipment in each group is adjusted in real time. For example, if a large number of people are expected to flow into a certain functional area in the next 30 minutes, the control strategy may require the air conditioning equipment in that area to increase its operating power to adapt to the temperature requirements in advance.

[0084] 2. Adjust equipment operating mode: For example, when the flow of people is low and the temperature is relatively stable, the HVAC equipment in some areas may be switched to energy-saving mode (such as low power operation mode or temporary shutdown).

[0085] 3. Coordinate equipment operation: Control strategies may also require HVAC equipment in different areas to work in coordination to avoid resource waste caused by excessive operation of a single piece of equipment. For example, if the equipment in a certain area is overloaded, the system can adjust the equipment in nearby areas to share the load.

[0086] The core of this step is to translate the control strategy into specific equipment operation instructions. Through commonly used adjustment methods (automatic or manual), the HVAC equipment in each functional area is made to operate according to the optimal strategy within a preset time period. This precise control not only improves comfort and stability but also effectively reduces energy consumption, achieving the goals of energy conservation and emission reduction.

[0087] Example 2

[0088] This embodiment is based on the above embodiment one. This embodiment aims to illustrate the pre-training process of the human flow prediction neural network and the control strategy neural network in order to avoid the disconnect between real-time prediction and HVAC system response delay.

[0089] Those skilled in the art will understand that in time series data, input feature vectors are typically generated from historical data over a period of time. Specifically, a time window of a specific length is selected from the entire data sequence as input data to capture the regularity and trends within that specific time period. In this embodiment, considering a preset time period—that is, the length of the future preset time period for the predicted pedestrian traffic in each functional area of ​​the target service area and the control strategies for each group of HVAC equipment in the current service area—as the extraction window for feature data and / or input feature vectors can avoid the adjustment disconnect caused by solely relying on real-time data. Specifically, the preset time period (e.g., 30 minutes, 1 hour, etc.) is the basic time unit of the control strategy neural network. It not only covers the immediate feedback of the equipment but also spans the response latency of the HVAC system. By using this time period as the feature extraction window, the model can consider the latency of equipment adjustments during data analysis, ensuring that the adjustment results are synchronized with the system response. For example, suppose the current indoor temperature is 24°C, and based on pedestrian flow forecasts and environmental data analysis, the temperature demand in a certain functional area is expected to increase within the next 30 minutes. The control strategy neural network will then adjust the equipment operating parameters in advance based on this prediction (such as starting the air conditioner or adjusting the temperature). In this way, even if there is a time delay in equipment adjustment, the final temperature regulation effect can still match the predicted demand, thereby improving the system's stability and comfort.

[0090] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for HVAC control in highway service areas based on the Internet of Things and neural networks, characterized in that, Includes the following steps: S1. Divide the building area of ​​the target service area into functional zones, deploy IoT sensors in each functional zone to acquire environmental data, HVAC equipment parameters and pedestrian flow data, and group the HVAC equipment in the same building according to the functional zone in which they are located. S2. Image acquisition devices are installed at the entrances and exits of the target service area to collect information on vehicles entering and exiting; the vehicle information includes license plate number and dwell time. Infrared people counters are installed at the entrances and exits of each functional area of ​​the target service area to collect the number of people entering and exiting; S3. Take the pedestrian distribution data, vehicle information and number of people entering and exiting the current service area and the previous service area in the direction of highway traffic flow as the first dataset, input it into the pre-trained pedestrian flow prediction neural network, predict the pedestrian flow of each functional area of ​​the target service area in the future preset time period, and generate the corresponding pedestrian flow prediction data. S4. Use the current service area's environmental data, HVAC equipment parameters, pedestrian distribution data, and pedestrian prediction data as the second dataset, input them into the pre-trained control strategy neural network, and output the control strategy of each group of HVAC equipment in the current service area for the future preset time period. S5. Control each group of HVAC equipment within a preset time period according to the control strategy.

2. The method for HVAC control of highway service areas based on the Internet of Things and neural networks as described in claim 1, characterized in that, Before inputting the first dataset into the pre-trained pedestrian flow prediction neural network in step S3, the following steps are also included: Remove vehicle information from the previous service area where the vehicle stays for more than a first preset threshold, and delete passenger flow data of no more than 5 people from the passenger flow distribution data of any functional area of ​​the current service area and the number of people entering and leaving the current service area.

3. The method for HVAC control of highway service areas based on the Internet of Things and neural networks as described in claim 1, characterized in that, The crowd flow prediction neural network comprises the following sequentially connected components: The multi-input layer includes several separate input layers set according to the input data of each category of the first dataset, and outputs the feature representation of the first dataset; Temporal convolutional networks are used to extract short-term dependency features from the feature representation through convolutional dilation. Long Short-Term Memory (LSTM) networks are used to extract long-term dependency features from the short-term dependency features through a gating mechanism. The output layer is used to combine the output of the Long Short-Term Memory network into the human flow prediction results.

4. The method for HVAC control of highway service areas based on the Internet of Things and neural networks as described in claim 1, characterized in that, The control strategy neural network includes: The temporal input layer extracts temporal features from the pedestrian distribution data and pedestrian prediction data in the second dataset based on the long short-term neural network, and outputs a multi-dimensional temporal feature vector. The static input layer normalizes the environmental data and HVAC equipment parameters in the second data and outputs a one-dimensional numerical feature vector. The feature fusion layer, which is connected to the temporal input layer and the static input layer respectively, is used to concatenate the temporal feature vector and the numerical feature vector into an input feature vector.

5. The method for HVAC control of highway service areas based on the Internet of Things and neural networks as described in claim 4, characterized in that, The control strategy neural network also includes: The reinforcement learning module connected to the feature fusion layer, the reward function of the reinforcement learning module for: ;in, , , and The preset weight parameters, For comfort bonus parameters, For energy efficiency bonus parameters, As a reward for temperature control stability, This refers to the parameters for rewarding human traffic.

6. The method for HVAC control of highway service areas based on the Internet of Things and neural networks as described in claim 3 or 4, characterized in that, The pre-training method includes: using the length of the preset time period as the extraction window for feature data and / or input feature vectors.

Images