Temperature control optimization method and system based on recurrent neural network

Abstract

The embodiment of the present invention relates to the field of air conditioning control technology, and discloses a temperature control optimization method based on a recurrent neural network, including: obtaining terminal control information of each spatial area of ​​a target building stored in a pipeline database; obtaining target temperature information set for each spatial area in the target building, and controlling the operating status information of each device and terminal control module in the air conditioning unit according to the target control parameters; determining the regional status information of each spatial area in the target building; matching the regional status information with the target regional status, and if the match is inconsistent, determining the regional status information and the target regional status information to determine the adjustment parameters of the air conditioning unit, and controlling and adjusting the air conditioning unit of the target building according to the dynamic adjustment parameters. The embodiment of the present invention can accurately control the temperature and ensure that the temperature of each spatial area is maintained within the target range set by the user, thereby greatly improving the user's comfort.

Description

Technical Field

[0001] This invention relates to the field of air conditioning control technology, specifically to a temperature control optimization method and system based on a recurrent neural network. This application is a divisional application, with the parent application number being 202411752894.X, filed on December 2, 2024, and entitled "Optimization Method and System for Precise Room Temperature Control." Background Art

[0002] In existing air conditioning technology, most systems employ a holistic adjustment strategy. This means they typically only manage uniform temperature, humidity, and other environmental parameters for an entire building or a larger area. However, in practice, user needs vary across different spaces.

[0003] Existing air conditioning systems lack the ability to perceive changes in the spatial environment in real time. They typically adjust only according to preset programs or manual user input, but are unable to intelligently and adaptively adjust based on factors such as the actual temperature, humidity, and occupant activity within the space. This lack of perception prevents air conditioning systems from responding promptly to changes in the spatial environment, further reducing their effectiveness and energy efficiency. Therefore, designing a control solution that can optimize and adjust the system for different spaces has become a pressing technical challenge for those skilled in the art. Summary of the Invention

[0004] To address the above-mentioned drawbacks, an embodiment of the present invention discloses a temperature control optimization method based on a recurrent neural network, which can realize dynamic temperature adjustment of each room and provide a more comfortable environment.

[0005] A first aspect of an embodiment of the present invention discloses a temperature control optimization method based on a recurrent neural network, comprising:

[0006] Obtain terminal control information for each spatial area of ​​the target building stored in the pipeline database, wherein the terminal control information includes fan information, valve opening information, floor information, room number information, and pipeline point information;

[0007] Obtain target temperature information set for each spatial area in the target building, determine target control parameters for each spatial area in the target building based on the target temperature information set for each spatial area and the pre-configured initial operating parameters of the air-conditioning unit, and control the operating status information of each device in the air-conditioning unit and the terminal control module based on the target control parameters;

[0008] Determine the regional status information of each spatial area in the target building by using sensor components set in each spatial area;

[0009] The regional status information is matched with the target regional status. If the match is inconsistent, the regional status information and the target regional status information are determined to determine the adjustment parameters of the air-conditioning unit, and the air-conditioning unit of the target building is controlled, adjusted and optimized according to the dynamic adjustment parameters.

[0010] As an optional implementation manner, in the first aspect of the embodiments of the present invention, the regional state information includes a temperature adjustment curve, and the target regional state includes a target adjustment curve; and determining the regional state information of each spatial area within the target building by using sensor components disposed in each spatial area includes:

[0011] The temperature sensor at the return air outlet of each space area is used to obtain the temperature sensor signal of the corresponding space area within a set time period, and a corresponding temperature adjustment curve is generated according to the temperature sensor signal of the corresponding space area within the set time period. The temperature adjustment curve is used to represent the state of temperature drop within the set time range;

[0012] The matching of the region state information with the target region state, and determining a state difference between the region state information and the target region state information if the matches are inconsistent, includes:

[0013] The temperature adjustment curve is recorded as the first alignment sequence of length m, the target adjustment curve is recorded as the second alignment sequence of length n, and an initial distance matrix D of m*n is created. ij , where the initial distance matrix D ij Used to represent the distance between the i-th point in the first alignment sequence and the j-th point in the second alignment sequence;

[0014] Create a distance matrix D from the initial ij Cumulative distance matrix C of the same size ij , the cumulative distance matrix C ij Used to store the cumulative minimum distance from the starting position to the current position;

[0015] For the cumulative distance matrix C ij At each position in the , select the element with the smallest cumulative distance from the adjacent position and add the distance D of the current position ij As the cumulative distance of the current position; when the cumulative distance matrix C is filled ij After that, the end element C mn That is, the dynamic regularization distance between the temperature adjustment curve and the target adjustment curve;

[0016] The calculated dynamic regularization distance is compared with the pre-set dynamic distance threshold to determine the similarity between the temperature adjustment curve and the target adjustment curve. If the calculated dynamic regularization distance is greater than the set dynamic distance threshold, the two are inconsistent, and the distance difference between the two is determined based on the dynamic regularization distance.

[0017] As an optional implementation, in the first aspect of the embodiment of the present invention, when the cumulative distance matrix C is filled ij After that, the end element C mn That is, the dynamic regularization distance between the temperature adjustment curve and the target adjustment curve, and also includes:

[0018] Creating an optimal storage list, wherein the optimal storage list is used to store data points on the optimal path;

[0019] From the cumulative distance matrix C ij The end element C mn Start backtracking to the starting element C 00 ,During the backtracking process, the adjacent position with the smallest cumulative distance is selected according to the current position and added to the path list;

[0020] When tracing back to the starting element C 00 When , the optimal storage list includes all data points on the optimal path; an optimal path curve is generated according to all data points on the optimal path, and if the curvature of the optimal path curve exceeds a set value, it is determined that there is a large difference between the temperature adjustment curve and the target adjustment curve;

[0021] Generate temperature adjustment parameters for the terminal control system based on the comparison results.

[0022] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the optimization method further includes:

[0023] Acquiring infrared image video information captured by an infrared camera, extracting key frames from the infrared image video information, and sequentially stitching the key frames from the infrared image video information to obtain infrared stitching image information;

[0024] Inputting the infrared stitching image information into a pre-built infrared behavior model for identification to determine the number of people and their behavior information in the corresponding infrared image information;

[0025] The physical sensation range of the environment in which the corresponding personnel are located is determined based on the personnel quantity information and personnel behavior information. If the physical sensation range of the environment in which the corresponding personnel are located is a comfort range, the various devices in the air-conditioning unit will not be adjusted. If the comfort range of the environment in which the corresponding personnel are located is a hotter range, the ambient temperature of the corresponding space area will be controlled to be lowered. If the comfort range of the environment in which the corresponding personnel are located is a colder range, the temperature of the corresponding space area will be controlled to be increased.

[0026] As an optional implementation, in the first aspect of the embodiment of the present invention, the infrared behavior model includes an input layer, a convolution layer, a first pooling layer, an improved residual block, a second pooling layer, and a fully connected layer, wherein the convolution layer is a multi-branch homogeneous convolution layer, and an attention extraction module is provided in the improved residual block to extract channel features and spatial features of the input features; the infrared behavior model is constructed by the following steps:

[0027] The spliced ​​infrared training images are divided into a training set and a test set according to a set ratio, and the number of training sets is expanded by data augmentation; the set ratio is 7:3;

[0028] Set the relevant hyperparameters of the infrared behavior initial model and initialize the network weights; pass the training set into the infrared behavior initial model and perform forward propagation;

[0029] During the model training process, the network loss function is calculated according to the loss function formula, and then the back propagation algorithm is used to update the network weights and bias parameters to minimize the loss function. When the loss value tends to be stable, it is determined that the network model has reached a convergence state; the loss function formula is:

[0030] Among them, p(x) represents the true category of sample x, q(x) represents the predicted category of sample x obtained by the SofMax classifier, and finally the backpropagation algorithm is used to update the weights to minimize the loss function;

[0031] If the model meets the training requirements, the training of the network model is completed, and the optimal parameters of the network model are saved to obtain the infrared behavior model;

[0032] After inputting the infrared stitching image information into a pre-built infrared behavior model for identification to determine the number of people and the behavior of people in the corresponding infrared image information, the method further includes:

[0033] The personnel behavior information is matched with the set behavior strategy. If the match is consistent, the temperature is adjusted according to the matching result.

[0034] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the optimization method further includes:

[0035] When it is detected that load adjustment is required, the number of people in each space area during the corresponding time period is obtained through infrared cameras set up in each space area;

[0036] Obtain the temperature demand information of each spatial area and the spatial number information of the infrared camera within the corresponding time period;

[0037] The regional status value of each spatial area in the corresponding time period is determined based on the number of people in each spatial area in the corresponding time period, temperature demand information, and a pre-designed regional value calculation formula; wherein the regional value calculation formula is:

[0038]

[0039] Where k1 and k2 are influence coefficients, and k1 + k2 = 1, R(t) represents the number of people in the spatial area at time t, S is the floor area of ​​the spatial interval, f(t) represents the temperature demand at time t, and y is the regional state value;

[0040] Comparing the regional status value of each spatial area with a pre-set adjustment interval, if the regional status value is within the first adjustment range, using the maximum set temperature as the adjustment temperature parameter to control and adjust the air conditioning unit;

[0041] If the zone state value is within the second adjustment zone, the product of the maximum set temperature and the temperature coefficient is compared with the current set temperature. If the product of the maximum set temperature and the temperature coefficient is greater than the current set temperature, the product of the maximum set temperature and the temperature coefficient is used as the adjustment temperature parameter to adjust the control of the air-conditioning unit. If the product of the maximum set temperature and the temperature coefficient is not greater than the current set temperature, the current set temperature is used as the adjustment temperature parameter to adjust the control of the air-conditioning unit.

[0042] If the zone status value is in the third adjustment interval, the current set temperature is used as the adjustment temperature parameter to perform control adjustment of the air-conditioning unit.

[0043] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the optimization method further includes:

[0044] Obtain personnel movement information within a corresponding time period in a spatial area, and calculate the activity value of each person based on the personnel movement information and the activity analysis formula. The activity analysis formula is:

[0045]

[0046] Where ρ is the travel distance ratio, L is the travel distance in the corresponding time period, T is the duration of the corresponding time period, vt is the instantaneous speed at time t in the corresponding time period, vk is the standard travel speed, dt is the time integral, and SK is the standard travel distance;

[0047] The movement analysis difference is obtained by comparing the movement value of each person with the preset movement analysis threshold;

[0048] Inputting the movement analysis difference and the average temperature value in the spatial area into the constructed recurrent neural network model to obtain the spatial area temperature value when the absolute value of the movement analysis difference is minimum, and determining the spatial area temperature value when the absolute value of the movement analysis difference is minimum as the suitable temperature for the corresponding person in the corresponding time period;

[0049] The temperature of the corresponding space area is adjusted according to the suitable temperature for the corresponding personnel in the corresponding time period.

[0050] A second aspect of an embodiment of the present invention discloses a temperature control optimization system based on a recurrent neural network, comprising:

[0051] The first acquisition module is used to obtain the terminal control information of each spatial area of ​​the target building stored in the pipeline database, wherein the terminal control information includes fan information, valve opening information, floor information, room number information and pipeline point information;

[0052] The second acquisition module is used to obtain target temperature information set for each spatial area in the target building, determine target control parameters for each spatial area in the target building based on the target temperature information set for each spatial area and the pre-configured initial operating parameters of the air-conditioning unit, and control the operating status information of each device in the air-conditioning unit and the terminal control module according to the target control parameters;

[0053] Determination module: used to determine the regional status information of each spatial area in the target building through the sensor components set in each spatial area;

[0054] Matching module: used to match the regional status information with the target regional status. If the match is inconsistent, the regional status information and the target regional status information are determined to determine the adjustment parameters of the air-conditioning unit, and the air-conditioning unit of the target building is controlled, adjusted and optimized according to the dynamic adjustment parameters.

[0055] The third aspect of an embodiment of the present invention discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the temperature control optimization method based on a recurrent neural network disclosed in the first aspect of the embodiment of the present invention.

[0056] A fourth aspect of an embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the temperature control optimization method based on a recurrent neural network disclosed in the first aspect of an embodiment of the present invention.

[0057] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0058] The recurrent neural network-based temperature control optimization method in this embodiment of the present invention monitors the regional status information (such as actual temperature and humidity) of each spatial area in real time and matches it with the target regional status, thereby promptly detecting and correcting temperature deviations. This dynamic adjustment mechanism can avoid unnecessary energy consumption and improve the operating efficiency of the air conditioning system. Precise temperature control ensures that the temperature of each spatial area is maintained within the target range set by the user, thereby greatly improving user comfort. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0060] Figure 1 1 is a flow chart of a temperature control optimization method based on a recurrent neural network disclosed in an embodiment of the present invention;

[0061] Figure 2 1 is a schematic diagram of a process for comparing and matching cooling curves disclosed in an embodiment of the present invention;

[0062] Figure 3 This is a schematic diagram of the infrared recognition process disclosed in an embodiment of the present invention;

[0063] Figure 4 Schematic diagram of the structure of the infrared behavior model disclosed in an embodiment of the present invention;

[0064] Figure 5 is a schematic diagram of a load adjustment process disclosed in an embodiment of the present invention;

[0065] Figure 6 is a schematic diagram of the temperature adjustment process disclosed in an embodiment of the present invention; Figure 7 This is a structural diagram of a medium-wave infrared curing control system provided by an embodiment of the present invention;

[0066] Figure 8 It is a structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] It should be noted that the terms "first," "second," "third," "fourth," etc. in the description and claims of the present invention are used to distinguish different objects rather than to describe a specific order. The terms "including" and "having," as well as any variations thereof, in the embodiments of the present invention, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or apparatus.

[0069] Existing air-conditioning systems also lack the ability to perceive changes in the spatial environment in real time. They can usually only be adjusted according to preset programs or manual operations of the user, and cannot be adjusted intelligently and adaptively according to factors such as the actual temperature, humidity, and personnel activities in the space. This lack of perception makes it impossible for the air-conditioning system to respond to changes in the spatial environment in a timely manner, thereby further reducing its adjustment effect and energy efficiency. Based on this, the embodiment of the present invention discloses a temperature control optimization method, system, electronic device and storage medium based on a recurrent neural network, which can detect and correct temperature deviations in a timely manner by monitoring the regional status information (such as actual temperature, humidity, etc.) of each spatial area in real time and matching it with the target area status. This dynamic adjustment mechanism can avoid unnecessary energy consumption and improve the operating efficiency of the air-conditioning system. Accurate temperature control can ensure that the temperature of each spatial area is maintained within the target range set by the user, thereby greatly improving the user's comfort.

[0070] Example 1

[0071] See also Figure 1 , Figure 1It is a flow chart of the temperature control optimization method based on recurrent neural network disclosed in the embodiment of the present invention. Among them, the execution subject of the method described in the embodiment of the present invention is an execution subject composed of software and / or hardware, and the execution subject can receive relevant information by wired or / and wireless means, and can send certain instructions. Of course, it can also have certain processing functions and storage functions. The execution subject can control multiple devices, such as a remote physical server or cloud server and related software, or it can be a local host or server and related software that performs related operations on a device placed somewhere. In some scenarios, multiple storage devices can also be controlled, and the storage devices can be placed in the same place or different places as the devices. For example Figure 1 As shown, the temperature control optimization method based on the recurrent neural network includes the following steps:

[0072] S101: Acquire terminal control information of each spatial area of ​​a target building stored in a pipeline database, wherein the terminal control information includes fan information, valve opening information, floor information, room number information, and pipeline point information;

[0073] S102: Obtain target temperature information set for each spatial area in the target building, determine target control parameters for each spatial area in the target building based on the target temperature information set for each spatial area and pre-configured initial operating parameters of the air-conditioning unit, and control operating status information of each device and terminal control module in the air-conditioning unit based on the target control parameters;

[0074] S103: Determine regional status information of each spatial area in the target building through sensor components set in each spatial area;

[0075] S104: Match the regional status information with the target regional status. If the match is inconsistent, determine the adjustment parameters of the air-conditioning unit based on the regional status information and the target regional status information, and control and adjust the air-conditioning unit of the target building according to the dynamic adjustment parameters.

[0076] By acquiring terminal control information (such as fan information and valve opening information) for each spatial zone, the present invention can more closely understand and control the temperature control equipment in each zone. This provides basic data support for achieving precise control of room temperature.

[0077] Based on the target temperature information set for each space zone and the initial operating parameters of the air conditioning unit, the target control parameters for each zone can be calculated and set. This customized control strategy based on target temperature can significantly improve temperature control accuracy.

[0078] When regional status information differs from the target status, this embodiment of the present invention determines adjustment parameters based on the discrepancy and adjusts the operating status of the air conditioning unit in real time, helping to reduce energy waste and achieve energy conservation and emission reduction. This real-time monitoring and dynamic adjustment mechanism also enables rapid response to environmental changes or changes in user needs, ensuring that the air conditioning system's operation always meets user expectations.

[0079] More preferably, the regional status information includes a temperature adjustment curve, and the target regional status includes a target adjustment curve; and determining the regional status information of each spatial area in the target building by using sensor components disposed in each spatial area includes:

[0080] The temperature sensor at the return air outlet of each space area is used to obtain the temperature sensor signal of the corresponding space area within a set time period, and a corresponding temperature adjustment curve is generated according to the temperature sensor signal of the corresponding space area within the set time period. The temperature adjustment curve is used to represent the state of temperature drop within the set time range;

[0081] like Figure 2 As shown, matching the region state information with the target region state, and if the matches are inconsistent, determining the state difference between the region state information and the target region state information, includes:

[0082] S1041: Record the temperature adjustment curve as a first alignment sequence of length m, record the target adjustment curve as a second alignment sequence of length n, and create an m*n initial distance matrix D ij , where the initial distance matrix D ij Used to represent the distance between the i-th point in the first alignment sequence and the j-th point in the second alignment sequence;

[0083] S1042: Create a distance matrix D from the initial ij Cumulative distance matrix C of the same size ij , the cumulative distance matrix C ij Used to store the cumulative minimum distance from the starting position to the current position;

[0084] S1043: For the cumulative distance matrix C ij At each position in the , select the element with the smallest cumulative distance from the adjacent position and add the distance D of the current position ij As the cumulative distance of the current position; when the cumulative distance matrix C is filled ij After that, the end element C mn That is, the dynamic regularization distance between the temperature adjustment curve and the target adjustment curve;

[0085] S1044: The calculated dynamic regularization distance is compared with the preset dynamic distance threshold to determine the similarity between the temperature adjustment curve and the target adjustment curve. If the calculated dynamic regularization distance is greater than the set dynamic distance threshold, the two are inconsistently matched, and the distance difference between the two is determined based on the dynamic regularization distance. The embodiment of the present invention not only considers the temperature state at a single time point, but also considers the temperature change trend over time by introducing the temperature adjustment curve and the target adjustment curve. This curve-based matching method can better reflect the actual situation of temperature control than simple single-point matching, thereby improving the accuracy of matching.

[0086] By using this algorithm to calculate the dynamic regularized distance between the temperature adjustment curve and the target adjustment curve, we can effectively address the stretching and bending of the two curves on the time axis. This algorithm is particularly suitable for processing asynchronous or inconsistent time series data, improving the flexibility and accuracy of the matching.

[0087] When determining the state difference between the regional state information and the target regional state information, the degree of similarity between the two can be accurately quantified by calculating the dynamic regularization distance and comparing it with the preset dynamic distance threshold. When the dynamic regularization distance is greater than the threshold, it indicates that there is a significant difference between the current temperature control state and the target state. At this time, the specific difference size can be determined based on the dynamic regularization distance, providing a basis for subsequent adjustments. Based on accurate state difference information, more refined adjustment strategies can be formulated. For example, based on the size and direction of the difference, parameters such as the output power, fan speed, and valve opening of the air conditioning unit can be adjusted to achieve faster and more accurate temperature control. This data-driven adjustment strategy is more scientific and effective than traditional empirical adjustments.

[0088] Through precise temperature control and dynamic adjustment strategies, we ensure that the temperature in each space area is maintained within the user's desired range, thereby improving user comfort and satisfaction. In addition, due to the more refined adjustment strategy, unnecessary energy waste can be reduced, operating costs can be lowered, and the user experience can be further improved.

[0089] More preferably, when the cumulative distance matrix C is filled ij After that, the end element C mn That is, the dynamic regularization distance between the temperature adjustment curve and the target adjustment curve, and also includes:

[0090] Creating an optimal storage list, wherein the optimal storage list is used to store data points on the optimal path;

[0091] From the cumulative distance matrix C ij The end element C mn Start backtracking to the starting element C 00,During the backtracking process, the adjacent position with the smallest cumulative distance is selected according to the current position and added to the path list;

[0092] When tracing back to the starting element C 00 When , the optimal storage list includes all data points on the optimal path; an optimal path curve is generated according to all data points on the optimal path, and if the curvature of the optimal path curve exceeds a set value, it is determined that there is a large difference between the temperature adjustment curve and the target adjustment curve;

[0093] Generate temperature adjustment parameters for the terminal control system based on the comparison results.

[0094] The embodiment of the present invention can clearly track the best matching path between the temperature adjustment curve and the target adjustment curve by creating an optimal storage list to store the optimal path data points in the dynamic regularization algorithm. An optimal path curve is generated based on all the data points on the optimal path, and its curvature is evaluated. If the curvature exceeds the set value, it indicates that there is a large mismatch or deviation between the temperature adjustment curve and the target adjustment curve. After determining the state difference between the temperature adjustment curve and the target adjustment curve and the curvature of the optimal path curve, more accurate temperature adjustment parameters can be generated for the terminal control system based on this information. These parameters can more accurately reflect the difference between the current temperature state and the target state, and guide the system to make corresponding adjustments to achieve faster and more accurate temperature control.

[0095] By precisely matching the temperature adjustment curve with the target adjustment curve and generating temperature adjustment parameters based on the optimal path, the system's response speed and stability can be significantly improved. The system can more quickly identify temperature deviations and take appropriate adjustments, maintaining indoor temperature fluctuations within the set range and improving user comfort and satisfaction. Precise temperature control and dynamic adjustment strategies help reduce unnecessary energy waste. By optimizing adjustment parameters and paths, the system ensures that while meeting user needs, it minimizes energy consumption and operating costs, achieving sustainable green development.

[0096] When performing a specific comparison, you can compare the various spatial areas of the same chilled water pipeline on a floor, or you can compare curves acquired at the same location over different time periods. What are the benefits of this? Comparing the various spatial areas of the same chilled water pipeline on a floor ensures that these areas receive similar temperature control results under the same conditions. This helps reduce temperature fluctuations between different spatial areas, improves overall temperature control uniformity, and thus enhances user comfort and satisfaction.

[0097] By comparing the temperature adjustment curves of different spatial zones within the same chilled water pipeline, it is easier to identify potential system issues such as pipeline blockage, valve failure, or degraded terminal equipment performance. These issues can lead to inaccurate or inconsistent temperature control. Promptly identifying and resolving these issues can ensure stable system operation.

[0098] Comparing temperature curves collected at the same location over different time periods can help understand temperature trends and patterns over time. This helps develop more effective time-based temperature control strategies, such as adjusting air conditioning unit parameters like output power, fan speed, or valve opening to accommodate varying temperature requirements and external environmental conditions. By comparing temperature adjustment curves for different time periods, it's possible to identify time periods where temperature control is redundant or unnecessary, thereby optimizing adjustment strategies and reducing unnecessary energy waste.

[0099] Specifically, rooms near the main pipe are very cold, while cold water can't reach the rooms at the end, resulting in insufficient cooling capacity. During implementation, the equipment monitors return air temperature, and the platform records the routing of chilled water pipes in different rooms. For example, room 901 is close to the main pipe, while room 920 is at the end of the main pipe; room 1501 is close to the main pipe, while room 1520 is at the end of the main pipe. When all air conditioners are turned on, the platform determines the cooling rate of the room based on the cooling curve. If room 901 cools significantly faster than room 920, the opening of valves such as 901 and 902 is appropriately reduced, for example, to 50%, to allow cold water to reach room 920.

[0100] More preferably, Figure 3 and Figure 4 As shown, the optimization method further includes:

[0101] S105a: Acquire infrared image video information captured by an infrared camera, extract key frames from the infrared image video information, and stitch the key frames in the infrared image video information in sequence to obtain infrared stitched image information;

[0102] S105b: Inputting the infrared stitching image information into a pre-built infrared behavior model for identification to determine the number of people and their behavior information in the corresponding infrared image information;

[0103] S105c: Determine the somatosensory range of the environment in which the corresponding personnel are located based on the personnel quantity information and personnel behavior information. If the somatosensory range of the environment in which the corresponding personnel are located is a comfort range, then the various devices in the air-conditioning unit are not adjusted. If the comfort range of the environment in which the corresponding personnel are located is a hot range, then control is performed to lower the ambient temperature of the corresponding space area. If the comfort range of the environment in which the corresponding personnel are located is a cold range, then control is performed to increase the temperature of the corresponding space area.

[0104] The present invention uses infrared image and video information captured by infrared cameras to enable the system to perceive the distribution and dynamic behavior of people in indoor environments in real time. This non-contact monitoring method not only improves the convenience of data acquisition, but also avoids the discomfort or interference that traditional sensors may cause.

[0105] By extracting key frames from infrared image video and stitching them together, the system can generate complete infrared stitched image information, more accurately reflecting the temperature distribution of the indoor environment. Combined with pre-built infrared behavioral models, the system can identify the number and behavior of individuals and adjust the operating parameters of the air conditioning unit based on their actual needs and comfort levels. This behavior- and comfort-based temperature control method is more accurate and user-friendly than traditional fixed temperature settings or time-based temperature control.

[0106] Adjusting the operating parameters of air conditioners based on the number of people and their behavior can avoid unnecessary energy waste. For example, when there are few people in a room or activity is relatively quiet, the system can reduce the air conditioner's output power or shut down some equipment to save energy. Conversely, when there are more people or more activity, the system can increase the air conditioner's output power or adjust the air volume and temperature to meet occupant comfort needs. This on-demand adjustment approach helps improve energy efficiency and reduce operating costs.

[0107] By monitoring and adjusting indoor temperature in real time, the system ensures that people work and live in a comfortable environment. This behavior- and comfort-based temperature control approach not only improves user comfort but also enhances their trust and satisfaction with the system. Furthermore, the system's intelligence and automation levels are enhanced, providing users with a more convenient and intelligent user experience.

[0108] More preferably, the infrared behavior model includes an input layer, a convolution layer, a first pooling layer, an improved residual block, a second pooling layer, and a fully connected layer, wherein the convolution layer is a multi-branch homogeneous convolution layer, and an attention extraction module is provided in the improved residual block to extract channel features and spatial features of the input features; the infrared behavior model is constructed by the following steps:

[0109] The spliced ​​infrared training images are divided into a training set and a test set according to a set ratio, and the number of training sets is expanded by data augmentation; the set ratio is 7:3;

[0110] Set the relevant hyperparameters of the infrared behavior initial model and initialize the network weights; pass the training set into the infrared behavior initial model and perform forward propagation;

[0111] During the model training process, the network loss function is calculated according to the loss function formula, and then the back propagation algorithm is used to update the network weights and bias parameters to minimize the loss function. When the loss value tends to be stable, it is determined that the network model has reached a convergence state; the loss function formula is:

[0112] Among them, p(x) represents the true category of sample x, q(x) represents the predicted category of sample x obtained by the SofMax classifier, and finally the backpropagation algorithm is used to update the weights to minimize the loss function;

[0113] If the model meets the training requirements, the training of the network model is completed, and the optimal parameters of the network model are saved to obtain the infrared behavior model;

[0114] After inputting the infrared stitching image information into a pre-built infrared behavior model for identification to determine the number of people and the behavior of people in the corresponding infrared image information, the method further includes:

[0115] The personnel behavior information is matched with the set behavior strategy. If the match is consistent, the temperature is adjusted according to the matching result.

[0116] During the specific implementation, the user's behavior actions are obtained through the camera; the temperature parameters of the corresponding area are determined based on the user's behavior actions, for example, it can be identified whether the user feels the environment is cold; there can be two actions, the first is that the user has no clothes, hugs himself, and curls up; the user's behavior intention is used to characterize the user's feeling about the environment, and the second is that the corresponding user has clothes and puts on clothes; each user's feeling about the environment is different; here, dressing and undressing can also be used as an indicator to determine the user's behavior.

[0117] Specifically, based on the identified user actions, the user's behavioral intention is further analyzed. For example, if a user is naked, hugging themselves, and curling up, this may indicate that the user feels the environment is cold; while if the user is wearing clothes, this may indicate that the user feels the environment is comfortable or slightly cool but still within an acceptable range. The user's behavioral intention is converted into a representation of the environmental perception. This can be achieved by defining a series of behavioral intention labels (such as "cold," "comfortable," "hot," etc.) and associating the identified user actions with these labels.

[0118] Based on the user's representation of their perception of the environment, the temperature parameters of the corresponding area are determined. For example, if a large number of users are detected to be feeling the cold (such as hugging themselves or curling up), the temperature parameters of the area can be increased; if the user is detected to be putting on clothes, the temperature parameters may need to be adjusted or maintained according to the specific situation. A temperature adjustment strategy is formulated, including the adjustment amplitude and speed. This can be comprehensively considered based on factors such as the intensity of the user's behavioral intention, the floor space of the spatial area, and the temperature requirements of other users. The determined temperature parameters are transmitted to the air conditioning unit control system, and precise control of the spatial area temperature is achieved by adjusting parameters such as the output power and air supply volume of the air conditioning unit.

[0119] The infrared behavior model in this embodiment of the present invention uses multi-branch homogeneous convolutional layers and improved residual blocks, combined with an attention extraction module, to more effectively extract channel and spatial features from input features. This design enhances the model's ability to recognize human behavior and numbers in infrared images, improving recognition accuracy and robustness.

[0120] By dividing the stitched infrared training images into training and test sets in a 7:3 ratio and using data augmentation to expand the training set, the model can learn more diverse features and improve generalization. Furthermore, by optimizing network parameters using a loss function formula and backpropagation algorithm, the model gradually converges during training, achieving optimal performance.

[0121] After identifying the number of people and their behavior in the infrared image, the model matches this information with the pre-defined behavior policy. If a match is found, the room temperature is automatically adjusted based on the matching result. This intelligent temperature adjustment method not only improves user comfort but also reduces energy waste, achieving the goal of energy conservation and emission reduction. Through intelligent temperature adjustment, the system can adjust the operating parameters of the air conditioning unit in real time based on the actual needs and comfort levels of the users. This behavior- and comfort-based temperature control method is more user-friendly and improves user experience and satisfaction.

[0122] During the specific implementation, the overall operating parameters can be increased within a certain period of time to make the temperature in the corresponding area higher. Because this period is relatively active, the overall temperature perception of the human body will increase. For example, in the summer, when you just enter the office or just finish exercising, the temperature will be more obvious. At this time, cooling is needed. It is determined by identifying the amount of heat dissipated by the human body. The infrared camera can not only identify the user's behavior, but also the user's heat dissipation, and determine the user's behavior through heat dissipation.

[0123] During the specific implementation, each air-conditioning unit is associated with a space area, and the physical information input by the user is received to personalize each space. Because everyone's feelings are different, the above-mentioned personalized configuration method is used to provide more diverse data parameters.

[0124] like Figure 5 As shown, more preferably, the optimization method further includes:

[0125] S106a: When it is detected that load adjustment is required, the number of people in each spatial area within the corresponding time period is obtained through infrared cameras installed in each spatial area;

[0126] S106b: Obtaining temperature requirement information of each spatial area and spatial number information of the infrared camera within a corresponding time period;

[0127] S106c: Determine the regional status value of each spatial area within the corresponding time period based on the number of people in each spatial area within the corresponding time period, temperature demand information, and a pre-designed regional value calculation formula; wherein the regional value calculation formula is:

[0128]

[0129] Where k1 and k2 are influence coefficients, and k1 + k2 = 1, R(t) represents the number of people in the spatial area at time t, S is the floor area of ​​the spatial interval, f(t) represents the temperature demand at time t, and y is the regional state value;

[0130] S106d: comparing the zone status value of each spatial zone with a preset adjustment range; if the zone status value is within the first adjustment range, using the maximum set temperature as the adjustment temperature parameter to control and adjust the air conditioning unit;

[0131] S106e: If the zone state value is within the second adjustment zone, the product of the maximum set temperature and the temperature coefficient is compared with the current set temperature. If the product of the maximum set temperature and the temperature coefficient is greater than the current set temperature, the product of the maximum set temperature and the temperature coefficient is used as the adjustment temperature parameter to adjust the control of the air-conditioning unit. If the product of the maximum set temperature and the temperature coefficient is not greater than the current set temperature, the current set temperature is used as the adjustment temperature parameter to adjust the control of the air-conditioning unit.

[0132] S106f: If the zone status value is in the third adjustment range, the current set temperature is used as the adjustment temperature parameter to perform control adjustment of the air-conditioning unit.

[0133] This embodiment of the present invention uses infrared cameras to capture the number of people in each space in real time. Combined with temperature demand information and space number information, the system can more accurately understand the load demand of each space. This refined load adjustment method helps avoid excessive or insufficient cooling / heating, thereby improving energy efficiency.

[0134] The system dynamically adjusts temperature parameters by comparing zone status values ​​with pre-set adjustment ranges. This dynamic adjustment method allows for more flexible adaptation to load demands in different spatial areas and time periods, improving the accuracy and responsiveness of temperature control. By comprehensively considering the number of occupants and temperature requirements to determine temperature adjustment parameters, the system ensures that everyone works and lives in a comfortable environment. This behavior- and comfort-based temperature control approach not only improves user comfort but also enhances their trust and satisfaction with the system.

[0135] The system adjusts air conditioning unit operating parameters based on different zone status values, thus avoiding unnecessary energy waste. For example, in areas with fewer people or lower temperature requirements, the system can reduce air conditioning unit output or shut down some equipment to save energy. This on-demand adjustment helps improve energy efficiency and reduce operating costs.

[0136] In this embodiment of the present invention, the central air conditioning system control area in a building can be divided into n control zones based on user electricity usage behavior (such as peak hours and electricity usage habits) or building structure characteristics (such as room size, orientation, floor location, etc.). Each control zone has independent temperature control and adjustment capabilities, which can be flexibly adjusted according to actual needs.

[0137] When a demand response event (such as a peak in grid load, energy shortage, etc.) occurs, each control area is divided into response levels according to preset strategies and algorithms. Response levels are usually divided into three categories: strong response areas, weak response areas, and non-response areas. Strong response areas: areas where energy consumption needs to be reduced first in demand response events. Weak response areas: areas where energy consumption can be appropriately reduced in demand response events, but have a lower priority than strong response areas. Non-response areas: areas where energy consumption is not reduced or the reduction is small in demand response events. The regional value is a comprehensive indicator used to measure the response capability and priority of each control area in demand response events. The range of regional values ​​is usually set according to actual needs.

[0138] In the embodiment of the present invention, the temperature coefficient is an important parameter in the set temperature adjustment formula of the control area, which reflects the impact of the change of the regional value on the set temperature adjustment range during the demand response event.

[0139] More preferably, Figure 6 As shown, the optimization method further includes:

[0140] S107a: Obtain personnel movement information within a corresponding time period in the spatial area, and calculate the activity value of each person according to the personnel movement information and an activity analysis formula. The activity analysis formula is:

[0141]

[0142] Where ρ is the travel distance ratio, L is the travel distance in the corresponding time period, T is the duration of the corresponding time period, vt is the instantaneous speed at time t in the corresponding time period, vk is the standard travel speed, dt is the time integral, and SK is the standard travel distance;

[0143] S107b: Obtaining a movement analysis difference value by comparing each person's activity value with a preset movement analysis threshold;

[0144] S107c: Inputting the movement analysis difference and the average temperature value in the spatial area into the constructed recurrent neural network model to obtain the spatial area temperature value when the absolute value of the movement analysis difference is minimum, and determining the spatial area temperature value when the absolute value of the movement analysis difference is minimum as the suitable temperature for the corresponding person in the corresponding time period;

[0145] S107d: Adjust the temperature of the corresponding space area according to the suitable temperature for the corresponding personnel in the corresponding time period.

[0146] By introducing an activity analysis formula, this optimization method can calculate each person's activity value (AC) based on their movement information (such as distance and speed). This calculation method not only considers the distance a person travels but also incorporates changes in speed, thereby more comprehensively reflecting the activity status of a person within a spatial area.

[0147] By comparing the calculated human activity value with a preset human activity analysis threshold, the present invention can generate a motion analysis difference. This step helps identify discrepancies between human activity and preset standards, providing a basis for subsequent temperature adjustments. By inputting the motion analysis difference and the average temperature value within a spatial region into a recurrent neural network model, the temperature value for the spatial region at which the absolute value of the motion analysis difference is minimized can be predicted. This prediction reflects the optimal temperature for the corresponding human activity level within a given time period.

[0148] This embodiment of the present invention determines the optimal temperature for each person in a specific time period based on the output of the recurrent neural network model—the spatial temperature value at the point where the absolute value of the motion analysis difference is minimized. This step implements intelligent temperature regulation, automatically adjusting the temperature of a spatial area based on changes in a person's activity status. This intelligent temperature regulation not only improves personal comfort but also reduces energy consumption, achieving energy conservation and emission reduction goals.

[0149] The recurrent convolutional neural network in the embodiment of the present invention is constructed by the following steps:

[0150] Obtain the historical monitoring period's movement analysis difference and the spatial area's average temperature value, as well as the temperature-adjusted movement analysis difference and spatial area's temperature value, and construct a recurrent neural network model whose input is the historical monitoring period's movement analysis difference and spatial area's temperature value, and whose output is the spatial area's temperature value when the absolute value of the movement analysis difference is minimum.

[0151] The extracted historical monitoring time period movement analysis difference and spatial area temperature values, and the temperature-adjusted movement analysis difference and spatial area temperature values ​​are divided into a 70% parameter training set and a 30% parameter test set; the 70% parameter training set is input into the recurrent neural network model for training to obtain an initial recurrent neural network model; the 30% parameter test set is used to test the initial recurrent neural network model, and the initial recurrent neural network model with the highest judgment accuracy of the spatial area temperature value when the absolute value of the preset movement analysis difference is minimized is output as the recurrent neural network model until the corresponding training conditions are met.

[0152] The recurrent neural network-based temperature control optimization method in this embodiment of the present invention monitors the regional status information (such as actual temperature and humidity) of each spatial area in real time and matches it with the target regional status, thereby promptly detecting and correcting temperature deviations. This dynamic adjustment mechanism can avoid unnecessary energy consumption and improve the operating efficiency of the air conditioning system. Precise temperature control ensures that the temperature of each spatial area is maintained within the target range set by the user, thereby greatly improving user comfort.

[0153] Example 2

[0154] See also Figure 7 , Figure 7 FIG. 1 is a schematic diagram of the structure of the medium-wave infrared curing control system disclosed in an embodiment of the present invention. Figure 7 As shown, the medium-wave infrared curing control system may include:

[0155] The first acquisition module 21 is used to obtain the terminal control information of each spatial area of ​​the target building stored in the pipeline database, wherein the terminal control information includes fan information, valve opening information, floor information, room number information and pipeline point information;

[0156] The second acquisition module 22 is configured to obtain target temperature information set for each spatial area in the target building, determine target control parameters for each spatial area in the target building based on the target temperature information set for each spatial area and the pre-configured initial operating parameters of the air-conditioning unit, and control the operating status information of each device in the air-conditioning unit and the terminal control module based on the target control parameters;

[0157] Determination module 23: used to determine the regional status information of each spatial area in the target building through the sensor components arranged in each spatial area;

[0158] Matching module 24: used to match the regional status information with the target regional status. If the match is inconsistent, the regional status information and the target regional status information are determined to determine the adjustment parameters of the air-conditioning unit, and the air-conditioning unit of the target building is controlled and adjusted according to the dynamic adjustment parameters.

[0159] The recurrent neural network-based temperature control optimization method in this embodiment of the present invention monitors the regional status information (such as actual temperature and humidity) of each spatial area in real time and matches it with the target regional status, thereby promptly detecting and correcting temperature deviations. This dynamic adjustment mechanism can avoid unnecessary energy consumption and improve the operating efficiency of the air conditioning system. Precise temperature control ensures that the temperature of each spatial area is maintained within the target range set by the user, thereby greatly improving user comfort.

[0160] Example 3

[0161] See also Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device can be a computer, a server, etc. Of course, in certain circumstances, it can also be a smart device such as a mobile phone, a tablet computer, a monitoring terminal, and an image acquisition device with processing functions. Figure 8 As shown, the electronic device may include:

[0162] A memory 510 storing executable program code;

[0163] a processor 520 coupled to the memory 510;

[0164] The processor 520 calls the executable program code stored in the memory 510 to execute part or all of the steps in the temperature control optimization method based on the recurrent neural network in the first embodiment.

[0165] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute some or all of the steps in the temperature control optimization method based on a recurrent neural network in embodiment one.

[0166] An embodiment of the present invention further discloses a computer program product, wherein when the computer program product is run on a computer, the computer is caused to execute some or all of the steps in the temperature control optimization method based on recurrent neural network in embodiment one.

[0167] An embodiment of the present invention also discloses an application publishing platform, wherein the application publishing platform is used to publish a computer program product. When the computer program product runs on a computer, the computer executes some or all of the steps in the temperature control optimization method based on a recurrent neural network in Example 1.

[0168] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the processes does not necessarily mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0169] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of these units may be selected based on actual needs to achieve the objectives of this embodiment.

[0170] In addition, the functional units in the embodiments of the present invention may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The integrated unit may be implemented in the form of hardware or software functional units.

[0171] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a memory and includes several requests for causing a computer device (which can be a personal computer, server, or network device, specifically a processor in the computer device) to execute some or all of the steps of the method described in each embodiment of the present invention.

[0172] In the embodiments provided herein, it should be understood that "B corresponding to A" means that B is associated with A and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B based solely on A; B can also be determined based on A and / or other information.

[0173] Those skilled in the art will appreciate that some or all of the steps in the various methods of the embodiments may be performed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0174] The above is a detailed introduction to the temperature control optimization method, system, electronic device and storage medium based on the recurrent neural network disclosed in the embodiments of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A temperature control optimization method based on recurrent neural network, characterized in that: include: Obtain terminal control information for each spatial area of ​​the target building stored in the pipeline database, wherein the terminal control information includes fan information, valve opening information, floor information, room number information, and pipeline point information; Obtain target temperature information set for each spatial area in the target building, determine target control parameters for each spatial area in the target building based on the target temperature information set for each spatial area and the pre-configured initial operating parameters of the air-conditioning unit, and control the operating status information of each device in the air-conditioning unit and the terminal control module based on the target control parameters; Determining regional status information of each spatial area in a target building by using sensor components disposed in each spatial area; the regional status information includes a temperature adjustment curve, and the target regional status includes a target adjustment curve; determining regional status information of each spatial area in a target building by using sensor components disposed in each spatial area includes: The temperature sensor at the return air outlet of each space area is used to obtain the temperature sensor signal of the corresponding space area within a set time period, and a corresponding temperature adjustment curve is generated according to the temperature sensor signal of the corresponding space area within the set time period. The temperature adjustment curve is used to represent the state of temperature drop within the set time range; Matching the regional status information with the target regional status; if the matches are inconsistent, determining the regional status information and the target regional status information to determine adjustment parameters of the air-conditioning unit, and performing control adjustment optimization on the air-conditioning unit of the target building according to the adjustment parameters; The optimization method further includes: Obtain personnel movement information within a corresponding time period in a spatial area, and calculate the activity value of each person based on the personnel movement information and the activity analysis formula. The activity analysis formula is: Where ρ is the travel distance ratio, S is the travel distance in the corresponding time period, T is the duration of the corresponding time period, vt is the instantaneous speed at time t in the corresponding time period, v is the standard travel speed, dt is the time integral, and SK is the standard travel distance; The movement analysis difference is obtained by comparing the movement value of each person with the preset movement analysis threshold; Inputting the movement analysis difference and the average temperature value in the spatial area into the constructed recurrent neural network model to obtain the spatial area temperature value when the absolute value of the movement analysis difference is minimum, and determining the spatial area temperature value when the absolute value of the movement analysis difference is minimum as the suitable temperature for the corresponding person in the corresponding time period; The temperature of the corresponding space area is adjusted according to the suitable temperature for the corresponding personnel in the corresponding time period.

2. The temperature control optimization method based on recurrent neural network according to claim 1, characterized in that: The matching of the region state information with the target region state, and determining a state difference between the region state information and the target region state information if the matches are inconsistent, includes: The temperature adjustment curve is recorded as the first alignment sequence of length m, the target adjustment curve is recorded as the second alignment sequence of length n, and an initial distance matrix D of m*n is created. ij , where the initial distance matrix D ij Used to represent the distance between the i-th point in the first alignment sequence and the j-th point in the second alignment sequence; Create a distance matrix D from the initial ij Cumulative distance matrix C of the same size ij , the cumulative distance matrix C ij Used to store the cumulative minimum distance from the starting position to the current position; For the cumulative distance matrix C ij At each position in the , select the element with the smallest cumulative distance from the adjacent position and add the distance D of the current position ij As the cumulative distance of the current position; when the cumulative distance matrix C is filled ij After that, the end element C mn That is, the dynamic regularization distance between the temperature adjustment curve and the target adjustment curve; The calculated dynamic regularization distance is compared with the pre-set dynamic distance threshold to determine the similarity between the temperature adjustment curve and the target adjustment curve. If the calculated dynamic regularization distance is greater than the set dynamic distance threshold, the two are inconsistent, and the distance difference between the two is determined based on the dynamic regularization distance.

3. The temperature control optimization method based on recurrent neural network according to claim 2, characterized in that: When the cumulative distance matrix C is filled ij After that, the end element C mn That is, the dynamic regularization distance between the temperature adjustment curve and the target adjustment curve, and also includes: Creating an optimal storage list, wherein the optimal storage list is used to store data points on the optimal path; From the cumulative distance matrix C ij The end element C mn Start backtracking to the starting element C 00 ,During the backtracking process, the adjacent position with the smallest cumulative distance is selected according to the current position and added to the path list; When tracing back to the starting element C 00 When , the optimal storage list includes all data points on the optimal path; an optimal path curve is generated according to all data points on the optimal path, and if the curvature of the optimal path curve exceeds a set value, it is determined that there is a large difference between the temperature adjustment curve and the target adjustment curve; Generate temperature adjustment parameters for the terminal control system based on the comparison results.

4. The temperature control optimization method based on recurrent neural network according to claim 1, characterized in that: The optimization method further includes: Acquiring infrared image video information captured by an infrared camera, extracting key frames from the infrared image video information, and sequentially stitching the key frames from the infrared image video information to obtain infrared stitching image information; Inputting the infrared stitching image information into a pre-built infrared behavior model for identification to determine the number of people and their behavior information in the corresponding infrared image information; The physical sensation range of the environment in which the corresponding personnel are located is determined based on the personnel quantity information and personnel behavior information. If the physical sensation range of the environment in which the corresponding personnel are located is a comfort range, the various devices in the air-conditioning unit will not be adjusted. If the comfort range of the environment in which the corresponding personnel are located is a hotter range, the ambient temperature of the corresponding space area will be controlled to be lowered. If the comfort range of the environment in which the corresponding personnel are located is a colder range, the temperature of the corresponding space area will be controlled to be increased.

5. The temperature control optimization method based on recurrent neural network according to claim 4, characterized in that: The infrared behavior model includes an input layer, a convolutional layer, a first pooling layer, an improved residual block, a second pooling layer, and a fully connected layer. The convolutional layer is a multi-branch homogeneous convolutional layer. An attention extraction module is provided in the improved residual block to extract channel features and spatial features of the input features. The infrared behavior model is constructed by the following steps: The spliced ​​infrared training images are divided into a training set and a test set according to a set ratio, and the number of training sets is expanded by data augmentation; the set ratio is 7:3; Set the relevant hyperparameters of the infrared behavior initial model and initialize the network weights; pass the training set into the infrared behavior initial model and perform forward propagation; During the model training process, the network loss function is calculated according to the loss function formula, and then the back propagation algorithm is used to update the network weights and bias parameters to minimize the loss function. When the loss value tends to be stable, it is determined that the network model has reached a convergence state; the loss function formula is: Among them, p(x) represents the true category of sample x, q(x) represents the predicted category of sample x obtained by the SofMax classifier, and finally the backpropagation algorithm is used to update the weights to minimize the loss function; If the model meets the training requirements, the training of the network model is completed, and the optimal parameters of the network model are saved to obtain the infrared behavior model; After inputting the infrared stitching image information into a pre-built infrared behavior model for identification to determine the number of people and the behavior of people in the corresponding infrared image information, the method further includes: The personnel behavior information is matched with the set behavior strategy. If the match is consistent, the temperature is adjusted according to the matching result.

6. The temperature control optimization method based on recurrent neural network according to claim 1, characterized in that: The optimization method further includes: When it is detected that load adjustment is required, the number of people in each space area during the corresponding time period is obtained through infrared cameras set up in each space area; Obtain the temperature demand information of each spatial area and the spatial number information of the infrared camera within the corresponding time period; The regional status value of each spatial area in the corresponding time period is determined based on the number of people in each spatial area in the corresponding time period, temperature demand information, and a pre-designed regional value calculation formula; wherein the regional value calculation formula is: Where k1 and k2 are influence coefficients, and k1 + k2 = 1, R(t) represents the number of people in the spatial area at time t, S is the floor area of ​​the spatial interval, f(t) represents the temperature demand at time t, and y is the regional state value; Comparing the regional status value of each spatial area with a pre-set adjustment interval, if the regional status value is within the first adjustment range, using the maximum set temperature as the adjustment temperature parameter to control and adjust the air conditioning unit; If the zone state value is within the second adjustment zone, the product of the maximum set temperature and the temperature coefficient is compared with the current set temperature. If the product of the maximum set temperature and the temperature coefficient is greater than the current set temperature, the product of the maximum set temperature and the temperature coefficient is used as the adjustment temperature parameter to adjust the control of the air-conditioning unit. If the product of the maximum set temperature and the temperature coefficient is not greater than the current set temperature, the current set temperature is used as the adjustment temperature parameter to adjust the control of the air-conditioning unit. If the zone status value is in the third adjustment interval, the current set temperature is used as the adjustment temperature parameter to perform control adjustment of the air-conditioning unit.

7. A temperature control optimization system based on recurrent neural network, characterized in that: include: The first acquisition module is used to obtain the terminal control information of each spatial area of ​​the target building stored in the pipeline database, wherein the terminal control information includes fan information, valve opening information, floor information, room number information and pipeline point information; The second acquisition module is used to obtain target temperature information set for each spatial area in the target building, determine target control parameters for each spatial area in the target building based on the target temperature information set for each spatial area and the pre-configured initial operating parameters of the air-conditioning unit, and control the operating status information of each device in the air-conditioning unit and the terminal control module according to the target control parameters; Determination module: used to determine regional status information of each spatial area in the target building by using sensor components arranged in each spatial area; the regional status information includes a temperature adjustment curve, and the target area status includes a target adjustment curve; the determination of regional status information of each spatial area in the target building by using sensor components arranged in each spatial area includes: The temperature sensor at the return air outlet of each space area is used to obtain the temperature sensor signal of the corresponding space area within a set time period, and a corresponding temperature adjustment curve is generated according to the temperature sensor signal of the corresponding space area within the set time period. The temperature adjustment curve is used to represent the state of temperature drop within the set time range; Matching module: used to match the regional status information with the target regional status. If the match is inconsistent, the regional status information and the target regional status information are determined to determine the adjustment parameters of the air-conditioning unit, and control and adjust the air-conditioning unit of the target building according to the adjustment parameters; The optimization system further includes: Obtain personnel movement information within a corresponding time period in a spatial area, and calculate the activity value of each person based on the personnel movement information and the activity analysis formula. The activity analysis formula is: Where ρ is the travel distance ratio, S is the travel distance in the corresponding time period, T is the duration of the corresponding time period, vt is the instantaneous speed at time t in the corresponding time period, v is the standard travel speed, dt is the time integral, and SK is the standard travel distance; The movement analysis difference is obtained by comparing the movement value of each person with the preset movement analysis threshold; Inputting the movement analysis difference and the average temperature value in the spatial area into the constructed recurrent neural network model to obtain the spatial area temperature value when the absolute value of the movement analysis difference is minimum, and determining the spatial area temperature value when the absolute value of the movement analysis difference is minimum as the suitable temperature for the corresponding person in the corresponding time period; The temperature of the corresponding space area is adjusted according to the suitable temperature for the corresponding personnel in the corresponding time period.

8. An electronic device, characterized in that: include: a memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the temperature control optimization method based on recurrent neural network according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein the computer program enables a computer to execute the temperature control optimization method based on a recurrent neural network according to any one of claims 1 to 6.

Images