Space-intention-based conference room device linkage control method, device and medium

Abstract

This invention relates to the field of intelligent office technology, and in particular to a method, device, and medium for the coordinated control of conference room equipment based on spatial intent. The method includes: constructing structured information text based on a monitoring dataset of the target conference room during a historical monitoring period; inputting the data into a preset large model to obtain a sequence of spatial intent states of the target conference room; extracting typical feature centers of each spatial intent state; obtaining the optimal control parameters of each device under each spatial intent state through multi-objective constraint optimization; determining the current spatial intent state based on the matching degree between real-time monitoring data and each typical feature center; assigning control priorities to each device in real time; sorting several device control commands according to control priorities; generating final control commands by combining the optimal control parameters of each device; and issuing them to the corresponding devices for coordinated execution. This invention can achieve adaptive parameter optimization and precise coordinated control of multiple devices, significantly reducing energy consumption while ensuring comfort.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent office, and particularly to a method, device and medium for linkage control of conference room devices based on spatial intention. Background Art

[0002] With the rapid development of Internet of Things and artificial intelligence technologies, the linkage control of multiple devices in intelligent office spaces has become an important means to improve office comfort and energy efficiency. Traditional control methods for office space devices usually adopt timing control based on pre-determined information or trigger control based on a single sensor (such as human infrared induction). For example, conference room air conditioners, lights and other devices are automatically turned on according to the pre-scheduled meeting time, or lighting is turned on when personnel movement is detected.

[0003] Although the above methods have improved the level of intelligent control to a certain extent, there are still the following deficiencies: the pre-determined information cannot reflect the real-time usage status. For example, a meeting may start earlier or be cancelled temporarily. Simply relying on pre-determined information for device control will lead to energy waste or a decrease in comfort. If a sensor trigger method is introduced, it is difficult to perceive complex scenarios. Only through human infrared induction, the fine-grained status of the office space cannot be distinguished, which is prone to misjudgment. And when adjusting device parameters, it is adjusted according to fixed set parameters and cannot be adjusted according to real-time environmental dynamic factors, resulting in difficulty in balancing comfort and energy consumption. Summary of the Invention

[0004] In view of the above technical problems, the present invention provides a method, device and medium for linkage control of conference room devices based on spatial intention, which realizes adaptive parameter optimization and precise linkage control of multiple devices by integrating historical data rules and fine-grained real-time state perception, and significantly reduces energy consumption on the premise of ensuring comfort.

[0005] According to the first aspect of the present invention, a method for linkage control of conference room devices based on spatial intention is provided, including the following steps: S1, preprocess the monitoring data set of the target conference room in the obtained historical monitoring period and perform index statistics in multiple preset dimensions to construct a structured information text of the target conference room; the monitoring data set includes pre-determined information, environmental sensor information, device usage information and personnel situation information corresponding to the target conference room.

[0006] S2, embed the structured information text into a preset prompt word template and then input it into a preset large model, drive the preset large model to perform inference on the spatial intention state sequence of the target conference room, and extract the typical feature centers of each spatial intention state based on the inference result.

[0007] S3, calculate the optimal control parameters of each device in each spatial intention state by using a multi-objective constraint optimization method; the multi-objectives at least include energy consumption and comfort.

[0008] S4. Collect real-time monitoring data of the target conference room and determine the current spatial intent state based on the matching degree between the real-time monitoring data and each typical feature center. Based on the current spatial intent state, real-time personnel information and equipment usage information extracted from the monitoring dataset, dynamically assign control priorities to each device in real time.

[0009] S5 sorts the control commands of several devices according to the control priority of each device, generates the final control command by combining the optimal control parameters of each device corresponding to the current spatial intention state, and sends it to the corresponding device for coordinated execution.

[0010] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, wherein at least one instruction or at least one program is stored therein, the at least one instruction or the at least one program being loaded and executed by a processor to implement the above-described spatial intent-based conference room equipment linkage control method.

[0011] According to a third aspect of the present invention, an electronic device is provided, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0012] The present invention has at least the following beneficial effects: This invention provides a spatial intent-based method for the coordinated control of conference room equipment. First, based on the monitoring dataset of the target conference room during historical monitoring periods, and combined with statistical indicators, a structured information text is constructed, covering a comprehensive representation of conference room usage patterns, providing a high-quality and information-rich data foundation for subsequent processing. Then, the structured information text is embedded into a preset prompt word template and input into a preset large model, achieving a reliable mapping from data to business semantics. This automatically mines the implicit conference room usage patterns in historical monitoring data, accurately inferring spatial intent states with finer granularity, and extracting the typical feature centers of each spatial intent state to provide a stable matching benchmark for real-time identification of subsequent states. Furthermore, multi-objective constraint optimization is used to obtain a balance between energy consumption and comfort under each spatial intent state. The system optimizes the control parameters of each device, avoiding energy waste and insufficient comfort caused by fixed parameters. Then, based on the matching degree between real-time monitoring data and typical feature centers, it determines the current spatial intention state, meeting the millisecond-level real-time control response requirements. Combined with multi-dimensional calculation results, it assigns control priorities to each device in real time, ensuring priority response of high-importance devices and adapting to real-time changes brought about by personnel flow, thus achieving reasonable allocation of control resources. Finally, it sorts several instructions according to control priority, generates the final control instructions based on the optimal control parameters of each device, and sends them to the corresponding devices for coordinated execution. Through adaptive parameter optimization and precise linkage control of multiple devices, the intelligent control level of the office space is significantly improved, and overall energy consumption is effectively reduced while ensuring comfort. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart of a conference room equipment linkage control method based on spatial intent provided in an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] This invention provides a method for coordinated control of conference room equipment based on spatial intent, such as... Figure 1As shown, the method includes the following steps: S1 involves preprocessing the acquired historical monitoring dataset of the target meeting room and statistically analyzing multiple pre-defined dimensions to construct a structured information text for the target meeting room. This structured information text can be understood as including the preprocessed monitoring dataset and the extracted indicators across various dimensions. The historical monitoring period can be the previous week or month. The target meeting room can also be replaced with the target office space, especially office spaces with strong usage patterns.

[0017] The monitoring dataset includes reservation information, environmental sensor information, equipment usage information, and personnel information for the target meeting room.

[0018] Specifically, the preprocessing of the monitoring dataset includes the following steps: Align each type of monitoring data in the monitoring dataset with the same sampling period on the time axis. For example, use a sampling period of 1 minute to ensure that the time dimension of all types of monitoring data is consistent.

[0019] For missing monitoring data, linear interpolation is used to complete it. For example, missing environmental sensor data.

[0020] For outliers, the 3σ principle is used for identification and correction, specifically: X i =(aX i-1 +bX i+1 ) / (a+b), where X i Let X be the correction value at time i. i-1 and X i+1 Let be the monitoring data values ​​at the time preceding and following time i, respectively, and let a and b be preset dynamic weights such that a + b = 1. For example, when the mean of the environmental sensor data is μ, the standard deviation is σ, and |X| satisfies... i If -μ|>3σ, then determine X i This is an outlier.

[0021] Specifically, the monitoring dataset is subjected to statistical analysis of multiple preset dimensions of indicators, including: For the aforementioned reservation information, the set of reserved time intervals for the target meeting room within the historical monitoring period is extracted. Based on the set of reserved time intervals, the total reservation duration and reservation compactness are calculated, and the meeting topic, number of participants, and required equipment list corresponding to each reserved time interval are recorded. The reservation compactness is the ratio of the product of the total reservation duration and the number of reserved time intervals to the total duration of the historical monitoring period.

[0022] For the environmental sensor information, the comprehensive air quality index and its associated preset basic characteristic indicators are calculated. The environmental sensor information includes real-time carbon dioxide concentration, dust concentration, real-time temperature, and real-time humidity. The preset basic characteristic indicators associated with the comprehensive air quality index include the average rate of change of carbon dioxide concentration, peak dust concentration, temperature fluctuation range, and humidity standard deviation. The comprehensive air quality index is calculated by dividing each of the above preset basic characteristic indicators by its corresponding preset maximum value to obtain normalized values, and then performing a weighted summation.

[0023] For the aforementioned device usage information, the usage frequency, average single usage duration, usage preference weights under different meeting themes, and usage efficiency indicators for each device are statistically analyzed. Device usage information includes operating parameters, usage duration, and usage frequency for each device under different meeting themes; devices refer to switches, lights, external speakers, air conditioners, electric curtains, and display devices such as televisions and projectors. The usage preference weight for a device under different meeting themes is the proportion of the device's usage frequency under that specific meeting theme to the total usage frequency of that device across all meeting themes.

[0024] Specifically, the equipment utilization efficiency index η is: Where ε is the average duration of a single device use, K is the number of different meeting topics, and N is the average duration of a single device use. d ω represents the number of times the d-th meeting topic is scheduled. d T0 represents the usage preference weight of the device under the d-th meeting topic, and T0 represents the total duration of the historical monitoring period.

[0025] The personnel information includes statistics on actual arrival rate, entry and exit frequency, and real-time personnel density. This personnel information includes the number of people, entry and exit frequency, and the actual number of people entering each meeting topic during historical monitoring periods.

[0026] Specifically, the real-time population density is as follows: D i =(G i +f i ×ζ×G) / G0, where, D i Let G be the real-time population density at time i. i Let f be the actual number of people in the target meeting room at time i, ζ be the preset influence coefficient, and f be the number of people in the meeting room at time i. i The formula defines the frequency of personnel entry and exit within a preset time period before and after time i, and G0 represents the rated capacity of the target meeting room. This formula combines real-time personnel numbers with short-term entry and exit frequencies to more accurately reflect the personnel density of the meeting room at different times, providing a precise basis for real-time environmental adjustments.

[0027] The above-mentioned approach, through multi-dimensional statistical analysis and structured processing of historical monitoring data, comprehensively represents the usage patterns of the conference room and transforms heterogeneous data into quantitative indicators with physical and business semantics. This provides a high-quality and information-rich data foundation for subsequent large-scale model inference and multi-objective optimization, enabling the identification of spatial intent states and the solution of optimal control parameters to take into account both historical patterns and real-time changes, thereby improving the accuracy of large-scale model inference and optimal control parameter calculation.

[0028] S2 involves embedding structured information text into a preset prompt word template and inputting it into a preset large model. This drives the preset large model to perform spatial intent state sequence reasoning for the target conference room and extracts typical feature centers for each spatial intent state based on the reasoning results. This can be understood as follows: the preset prompt word template includes a preset set of spatial intent states and reasoning requirement statements. The preset large model can be a large model based on GPT-4 or LLaMA3 architecture, which, by understanding time-series monitoring data and supplementing it with statistical multi-dimensional indicators, infers fine-grained spatial intent state sequences.

[0029] Specifically, the spatial intent state set includes idle, preparing, meeting, rest / interval, timed out, and cleanup.

[0030] Furthermore, the typical feature center of the spatial intent state refers to the mean vector of various types of monitoring data in the historical monitoring dataset under the current spatial intent state, which is used to characterize the typical environmental pattern of the current spatial intent state.

[0031] As mentioned above, traditional methods only statistically analyze the historical usage status of meeting rooms based on predetermined information, which often results in errors compared to the actual status and fails to reveal more granular states. This application utilizes a large model to perform semantic understanding on structured information text, automatically mining the hidden patterns of meeting room usage in historical monitoring data, generating a continuous spatial intent state sequence, achieving a reliable mapping from multi-source heterogeneous data to high-dimensional semantic states, improving the inference accuracy of fine-grained meeting room states, and extracting typical feature centers of each state based on the inference results, providing a stable matching benchmark for the real-time identification of subsequent states.

[0032] In one specific embodiment, after generating the spatial intent state sequence, the following steps are also included: S201, Obtain a predefined directed graph of state transitions; the nodes of the directed graph of state transitions are a preset set of spatial intention states, and the edges are allowed state transition relationships. For example, there is a directed edge connecting the idle state and the preparing state, but not connecting the meeting state with the meeting state; there is a directed edge connecting the meeting state and the end cleanup state, but not connecting the end cleanup state with the meeting state.

[0033] S202, Traverse the directed graph of state transitions and check whether there is an allowed state transition relationship between the spatial intention state pairs at each adjacent time step in the directed graph of state transitions.

[0034] S203, if it does not exist, then select the target successor state as the corrected successor state from all allowed successor states of the previous spatial intention state in the spatial intention state pair. The target successor state refers to the spatial intention state with the highest product of the inference confidence output by the corresponding preset large model and the matching degree between the corresponding preset feature center and the feature vector at the time of the next spatial intention state. This can be understood as: for each successor state, obtain the inference confidence output by the preset large model for that successor state, and the matching degree between the preset feature center corresponding to that successor state and the feature vector at the time when the next spatial intention state in the spatial intention state pair should be located, and take the successor state with the highest product of the two as the target successor state. Conversely, if an allowed state transition relationship exists, no correction is required.

[0035] Specifically, all allowed successor states of the previous spatial intention state refer to the set of spatial intention states obtained from the directed graph of state transition that have directed edges with the previous spatial intention state; where a directed edge with the previous spatial intention state means that the starting point of the direction of the edge is the previous spatial intention state.

[0036] As described above, the inferred state sequence is validated for compliance using a predefined directed graph of state transitions. This ensures that the state transitions conform to the business logic used in the conference room and avoids illogical state jumps. For illogical transitions, a successor state that satisfies both transition compliance and has high confidence and high feature matching is selected for correction, thereby improving the logical consistency and reliability of the state sequence.

[0037] Furthermore, the reasoning confidence of the preset large model for any spatial intention state meets the following condition: C t =sigmoid(α1×S t +α2×F t +α3×P t )×λ t , where C t Let S be the reasoning confidence of the spatial intention state at time t. t Let F be the cosine similarity between the feature vector of the monitoring dataset at time t and the preset feature center of the spatial intent state at time t. t P is the rationality score of the transition from the spatial intention state at time t-1 to the spatial intention state at time t, obtained based on the directed graph of the state transition. t λ is the prior probability of the spatial intention state at time t calculated based on the predetermined information. tLet F be the time decay factor at time t, and α1, α2, and α3 are all learnable weights. The prior probability is the probability obtained based on predetermined information, such as the prior probability of the state in a meeting being within a predetermined time interval being 0.8; when F... t For the first inference time step, the value is 0 or a preset constant. The initial values ​​of α1, α2, and α3 can be set to 0.5, 0.3, and 0.2 respectively. To adapt to real-time scene changes, they can be fine-tuned using linear gradient descent according to actual needs.

[0038] Among them, the preset feature center of the spatial intention state is set manually or obtained through historical unsupervised clustering.

[0039] Specifically, the transition rationality score refers to the logarithmic value of the transition probability from the spatial intention state at time t-1 to the spatial intention state at time t, obtained based on the statistical analysis of the historical state sequence output by the large model. The directed graph of state transition provides prior knowledge of whether a transition is possible between two spatial intention states. If the transition probability is 0, the transition rationality score is directly set to -10.

[0040] It should be noted that, in the specific implementation, each spatial intent state output is the spatial intent state corresponding to the highest inference confidence at that time. The definitions of each state in the directed graph of state transition are determined by the semantic understanding of the preset large model, and a mapping relationship between spatial intent states and features is established from the structured information text; it can be understood as follows: based on the semantic understanding of the preset large model, the initial confidence generated by the preset large model is corrected by the inference confidence calculation formula, thereby outputting an accurate sequence of spatial intent states.

[0041] Furthermore, λ t It conforms to the following formula: λ t =exp(-(|t-t0|) / τ t ), where t0 is the typical occurrence time of the spatial intent state at time t in history, for example, the state in a meeting is usually 5 minutes after the hour, extracted from the historical monitoring dataset; τ t The preset time constant is the spatial intent state at time t, such as 30 minutes for the idle state and 10 minutes for the meeting state, to ensure that the time decay logic of different spatial intent states conforms to the actual use scenario.

[0042] The confidence calculation formula described above does not directly use the confidence score output by a large model. Instead, it integrates three dimensions: feature similarity, state transition logic, and predetermined prior probability. It also introduces a time decay factor to correct for deviations caused by atypical moments, such as delayed start or early end of a meeting, effectively suppressing noise interference. This allows for a comprehensive assessment of the credibility of the inference results. Furthermore, it uses learnable weights to dynamically adjust the contribution of each dimension, enabling the confidence assessment to adapt to the changing patterns of different meeting scenarios and reduce misjudgments.

[0043] S3 employs a multi-objective constraint optimization approach to calculate the optimal control parameters for each device under each spatial intent state. This can be understood as follows: the optimal control parameters refer to the best setpoints obtained by solving a multi-objective optimization problem for each type of adjustable device, such as air conditioning, lighting brightness, and electric curtains, while meeting the requirements of the current spatial intent state. For example, in a meeting state, what temperature should the air conditioning be set to, and how should the lighting brightness be set? In a rest / intermission state, should the lighting brightness be lowered?

[0044] Specifically, the multiple objectives include at least energy consumption and comfort.

[0045] Furthermore, the calculation process of the multi-objective constrained optimization method is as follows: S301, calculate the initial control parameters of device j that minimize the objective function; the objective function is: , among which, E(u j ) represents device j under control parameter u j Energy consumption under E max The preset maximum energy consumption is defined by u0, which represents the recommended comfort parameter value under the current spatial intention state. r The difference corresponds to the parameter adjustment range of device j, where w1 and w2 are preset weighting coefficients that add up to 1. The energy consumption function is obtained by fitting the historical usage characteristics of device j; for example, air conditioner energy consumption has a quadratic relationship with the set temperature difference. Those skilled in the art can also use other methods according to actual needs, which will not be elaborated here. The control interval is set to 1 minute.

[0046] Preferably, the preset weighting coefficients are adaptively adjusted according to the current spatial intent state. For example, increasing w2 during a meeting prioritizes comfort, increasing w1 during idle periods prioritizes energy saving, and balancing the weights of both during rest / intermittent periods.

[0047] S302, based on the preset parameter constraint range of device j, the initial control parameters are boundary-trimmed to obtain the optimal control parameters of device j.

[0048] As described above, for each spatial intent state, the optimal control parameters of the equipment that balance energy consumption and comfort are solved through a multi-objective constraint optimization method. This avoids the problems of energy waste and insufficient comfort caused by fixed control parameters. By transforming the abstract state intent into specific executable equipment parameter values, a dynamic balance between energy saving and comfort is achieved.

[0049] S4. Collect real-time monitoring data of the target conference room and determine the current spatial intent state based on the matching degree between the real-time monitoring data and each typical feature center. Based on the current spatial intent state, real-time personnel information and equipment usage information extracted from the monitoring dataset, dynamically assign control priorities to each device in real time.

[0050] Specifically, the data dimensions of real-time monitoring data are consistent with the data dimensions of the corresponding monitoring dataset within the historical monitoring period. For example, the current meeting topic, real-time carbon dioxide concentration, dust concentration, real-time temperature, real-time humidity, current operating parameters and on / off status of each device, real-time number of people in the target meeting room, and entry / exit frequency in the last minute, etc.

[0051] Specifically, determining the current spatial intent state based on the matching degree between real-time monitoring data and each typical feature center includes the following steps: S401 continuously collects real-time monitoring data of the target conference room during the current time period at a preset sampling period, and obtains the real-time data vector at each collection moment within the current time period; this can be understood as: the sampling period here is consistent with the sampling period of the monitoring dataset corresponding to the historical monitoring period.

[0052] S402, For the current acquisition time, calculate the cosine similarity between the real-time data vector at the current acquisition time and the typical feature center of each spatial intention state.

[0053] S403, determine the spatial intention state with the highest corresponding cosine similarity as the current spatial intention state.

[0054] By matching real-time data vectors with typical feature centers, the current spatial intent state can be quickly determined without complex feature extraction and model inference. It has low computational cost and low latency, meeting the requirements of millisecond-level real-time control response. Furthermore, the typical feature centers are extracted offline based on historical data, ensuring the stability and accuracy of the matching benchmark.

[0055] Furthermore, the formula for calculating the real-time dynamic control priority Y of any device is as follows: Where w0 is the preset state weight coefficient corresponding to the current spatial intent state, such as 1.0, 0.8, and 0.6 for the meeting state, preparation state, and rest state, respectively; m is the historical number of times the current device has been used under the current spatial intent state or the current meeting topic, that is, when in the meeting state, m is the historical number of times the current device has been used under the current meeting topic; m max ψ is the maximum number of times all devices have been used in the current spatial intent state or the current meeting topic; D is the real-time personnel density at the current moment; A is a preset constant, which can be set to 10; φ, δ and θ are adjustable weights and add up to 1. In this embodiment, ψ is 0.5, δ is 0.3 and θ is 0.2.

[0056] The above-mentioned system dynamically identifies the current spatial intent state based on real-time monitoring data, and combines state weights, historical equipment usage patterns, and real-time personnel density to calculate dynamic control priorities for each device in real time. This ensures that highly important devices receive priority response in critical states, while also adapting to real-time changes brought about by personnel movement, thereby achieving a reasonable allocation of control resources.

[0057] S5 sorts the control commands of several devices according to the control priority of each device, generates the final control command by combining the optimal control parameters of each device corresponding to the current spatial intention state, and sends it to the corresponding device for coordinated execution.

[0058] Specifically, the step of generating the final control command by combining the optimal control parameters of each device corresponding to the current spatial intent state includes: S501, for switch-type devices, directly generate on or off execution commands based on the current spatial intent state. Examples include switches such as lamps, projection screens, and external speakers; lamps are also adjustment devices, and are treated as switch-type devices when an on or off operation is required, and as adjustment devices when a light brightness adjustment is required.

[0059] S502, for regulating equipment, the adjustment step size of the actual control parameters of the current equipment towards the optimal control parameters of the current equipment is dynamically adjusted according to the rate of change of personnel density. The actual control parameters are then gradually adjusted according to the adjusted step size to obtain the final control parameters. That is, when personnel enter quickly, the adjustment speed is accelerated to quickly bring the environment to a comfortable state, and when personnel leave quickly, the adjustment speed is also accelerated to quickly return the equipment to an energy-saving state, achieving adaptive matching between response speed and the rhythm of personnel flow.

[0060] Preferably, when the rate of change of population density in the current time window is greater than the preset rate of change threshold, the step size adjustment is the product of the rate of change of population density, the preset adjustment coefficient, and the basic adjustment step size; when the rate of change of population density in the current time window is not greater than the preset rate of change threshold, no adjustment is required.

[0061] S503, based on the preset parameter constraint range corresponding to the optimal control parameters of each device, performs boundary trimming on the final control parameters and generates the final control command.

[0062] As described above, different instruction generation strategies are adopted for switch-type and regulating-type equipment. Switch-type equipment directly responds to state changes, while regulating-type equipment introduces adjustment step size for dynamic adjustment, enabling equipment parameters to adapt to changes in personnel flow in real time, balancing response speed and control accuracy, and achieving coordinated linkage of multiple devices in complex scenarios under the premise of energy saving and consumption reduction.

[0063] Furthermore, step S5 also includes the following steps: S510 sorts the device control commands from high to low according to the control priority of each device, and generates a control queue for the current spatial intent state.

[0064] S520: If there are mutually exclusive control commands for at least two devices in the control queue, the control command of the device with higher control priority is retained, and the control command of the device with lower control priority is discarded. For example, if there are two commands in the current control queue: lowering the projection screen and opening the electric curtains, if the projection screen is lowered and the electric curtains are opened, external light will interfere with the projection display effect, and there is a coordination conflict between the two. The system compares the device priorities corresponding to the two commands and only retains the command with higher priority.

[0065] S530, if a control command in the control queue for the same device is mutually exclusive with a control command executed in the previous cycle, and the time interval is less than a preset time window, a gradual transition strategy is triggered for the regulating device. For example, if the previous cycle executed a command to set the air conditioner temperature to 22°C due to high personnel density, and this cycle requires the air conditioner to be set to 26°C due to a large number of people leaving, with only a 30-second interval between the two commands, which is less than the preset time window, a direct jump would cause drastic temperature fluctuations and a surge in energy consumption. Therefore, the system triggers a gradual transition strategy, gradually adjusting to 26°C at a rate of 1°C per minute. Those skilled in the art can set the preset time window according to actual needs, such as 1 minute.

[0066] The above-mentioned priority sorting and mutual exclusion arbitration mechanism solves the conflict problem when multiple device control commands are issued at the same time, ensuring that the key commands of critical devices are executed first, avoiding command conflicts or confusion. Furthermore, a gradual transition strategy is introduced for sudden changes in commands for the same device to avoid drastic changes in device status, extend device life and improve user comfort.

[0067] Embodiments of the present invention also provide a non-transitory computer-readable storage medium that can be disposed in an electronic device to store at least one instruction or at least one program related to implementing a method in the method embodiments, wherein the at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiments.

[0068] Embodiments of the present invention also provide an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0069] While specific embodiments of the invention have been described in detail by way of example, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of the invention. It should also be understood that various modifications can be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims

1. A method for coordinated control of conference room equipment based on spatial intent, characterized in that, The method includes the following steps: S1, preprocess the monitoring dataset of the target conference room within the acquired historical monitoring period and perform statistical analysis on indicators of multiple preset dimensions to construct structured information text of the target conference room; The monitoring dataset includes reservation information, environmental sensor information, equipment usage information, and personnel information for the target meeting room. S2, after embedding the structured information text into the preset prompt word template, input it into the preset large model, drive the preset large model to perform spatial intention state sequence reasoning of the target conference room, and extract the typical feature center of each spatial intention state based on the reasoning results; S3, using a multi-objective constraint optimization method to calculate the optimal control parameters of each device under each spatial intention state; the multi-objective includes at least energy consumption and comfort; S4. Collect real-time monitoring data of the target conference room and determine the current spatial intent state based on the matching degree between the real-time monitoring data and each typical feature center. Based on the current spatial intent state, real-time personnel information and equipment usage information extracted from the monitoring dataset, dynamically allocate control priorities to each device in real time. S5 sorts the control commands of several devices according to the control priority of each device, generates the final control command by combining the optimal control parameters of each device corresponding to the current spatial intention state, and sends it to the corresponding device for coordinated execution.

2. The conference room equipment linkage control method based on spatial intent according to claim 1, characterized in that, In step S1, the monitoring dataset is subjected to statistical analysis of multiple preset dimensions, including: For the aforementioned reservation information, extract the set of reservation time intervals for the target meeting room within the historical monitoring period, calculate the total reservation duration and reservation compactness based on the set of reservation time intervals, and record the meeting topic, number of participants, and required equipment list for each reservation time interval. For the environmental sensor information, calculate the comprehensive air quality index and the preset basic characteristic indicators associated with the comprehensive air quality index; For the device usage information, the number of times each device is used, the average duration of a single use, the usage preference weights under different meeting topics, and the usage efficiency indicators are statistically analyzed. For the personnel information, the actual arrival rate, entry and exit frequency, and real-time personnel density are statistically analyzed.

3. The conference room equipment linkage control method based on spatial intent according to claim 1, characterized in that, After generating the spatial intent state sequence in step S2, the following steps are also included: S201, Obtain a predefined directed graph of state transitions; the nodes of the directed graph of state transitions are a preset set of spatial intention states, and the edges are allowed state transition relationships; S202, Traverse the directed graph of state transitions and check whether there is an allowed state transition relationship in the directed graph of state transitions for each adjacent time point of the spatial intention state pair. S203, if it does not exist, then select the target successor state as the corrected next spatial intention state from all the allowed successor states of the previous spatial intention state in the spatial intention state pair; the target successor state refers to the spatial intention state with the highest product of the inference confidence of the corresponding preset large model output and the matching degree of the corresponding preset feature center and the feature vector at the time of the next spatial intention state.

4. The conference room equipment linkage control method based on spatial intent according to claim 3, characterized in that, The confidence level of the pre-defined large model for inference of any spatial intention state meets the following conditions: C t =sigmoid(α1×S t +α2×F t +α3×P t )×λ t , where C t Let S be the reasoning confidence of the spatial intention state at time t. t Let F be the cosine similarity between the feature vector of the monitoring dataset at time t and the preset feature center of the spatial intent state at time t. t P is the rationality score of the transition from the spatial intention state at time t-1 to the spatial intention state at time t, obtained based on the directed graph of the state transition. t λ is the prior probability of the spatial intention state at time t calculated based on the predetermined information. t Let α1, α2, and α3 be the time decay factor at time t, and α1, α2, and α3 are all learnable weights.

5. The conference room equipment linkage control method based on spatial intent according to claim 1, characterized in that, The calculation process of the multi-objective constrained optimization method described in step S3 is as follows: S301, calculate the initial control parameters of device j that minimize the objective function; the objective function is: , among which, E(u j ) represents device j under control parameter u j Energy consumption under E max The preset maximum energy consumption is defined by u0, which represents the recommended comfort parameter value under the current spatial intention state. r w1 and w2 are the difference corresponding to the parameter adjustment range of device j, and w1 and w2 are preset weight coefficients that add up to 1; S302, based on the preset parameter constraint range of device j, the initial control parameters are boundary-trimmed to obtain the optimal control parameters of device j.

6. The conference room equipment linkage control method based on spatial intent according to claim 4, characterized in that, In step S4, determining the current spatial intent state based on the matching degree between real-time monitoring data and each typical feature center includes the following steps: S401 continuously collects real-time monitoring data of the target conference room during the current time period at a preset sampling period, and obtains the real-time data vector at each collection moment within the current time period. S402, For the current acquisition time, calculate the cosine similarity between the real-time data vector at the current acquisition time and the typical feature center of each spatial intention state; S403, determine the spatial intention state with the highest corresponding cosine similarity as the current spatial intention state.

7. The conference room equipment linkage control method based on spatial intent according to claim 1, characterized in that, In step S5, generating the final control command by combining the optimal control parameters of each device corresponding to the current spatial intent state includes: S501, for switch-type devices, directly generate open or close execution commands based on the current spatial intent state; S502, For regulating equipment, the adjustment step size of the actual control parameters of the current equipment is dynamically adjusted to the optimal control parameters of the current equipment according to the rate of change of personnel density, and the actual control parameters are gradually adjusted according to the adjusted step size to obtain the final control parameters; S503, based on the preset parameter constraint range corresponding to the optimal control parameters of each device, performs boundary trimming on the final control parameters and generates the final control command.

8. The conference room equipment linkage control method based on spatial intent according to claim 1, characterized in that, Step S5 also includes the following steps: S510 sorts the device control commands from high to low according to the control priority of each device and generates a control queue for the current spatial intent state. S520, if there are at least two mutually exclusive control instructions in the control queue, then retain the control instruction of the device with higher control priority and discard the control instruction of the device with lower control priority. S530, if the control command of the same device in the control queue is mutually exclusive with the control command executed in the previous cycle, and the time interval is less than the preset time window, then a gradual transition strategy is triggered for the regulating device.

9. A non-transitory computer-readable storage medium, wherein the storage medium stores at least one instruction or at least one program segment, characterized in that, The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the spatial intent-based conference room equipment linkage control method as described in any one of claims 1-8.

10. An electronic device, characterized in that, Includes a processor and the non-transitory computer-readable storage medium as described in claim 9.

Images