Intelligent control method and system for laminar air supply unit dedicated to operating room

Abstract

The present invention relates to the field of intelligent control technology, and discloses an intelligent control method and system for a laminar air supply unit dedicated to an operating room. The method comprises: constructing a three-dimensional spatial topological model of an operating room, collecting spatial position data of medical staff and equipment during an operation in real time, and forming a spatiotemporal trajectory data set; analyzing the spatiotemporal trajectory data set, and generating a pollution source distribution thermodynamic map; acquiring an environmental state data stream and a personnel state data stream of the operating room, and predicting a dynamic heat and moisture load demand curve of the operating room in a future period through a pre-trained heat and moisture load demand prediction model; generating a dynamic air supply control strategy based on the dynamic heat and moisture load demand curve, the three-dimensional spatial topological model of the operating room, and the pollution source distribution thermodynamic map, and controlling the laminar air supply unit to perform intelligent air supply; the present invention can accurately predict the heat and moisture load demand of the operating room, dynamically optimize the air supply strategy, and significantly reduce the air supply energy consumption while ensuring the cleanliness of the operating room.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and more specifically, to an intelligent control method and system for a laminar flow air supply unit dedicated to an operating room. Background Art

[0002] The operating room environment control system can ensure the safety of surgery and improve the quality of surgery. As a key component of the operating room environment control system, the operating room dedicated laminar flow air supply unit has a direct impact on the safety and comfort of the surgical process due to its intelligent control level.

[0003] At present, there are some technical solutions for intelligent air supply control in operating rooms. For example, the Chinese patent application with publication number CN111578429A discloses an intelligent air supply system for operating rooms and a control method thereof. The system collects total fresh air data, branch fresh air data, operating room air volume data, and pressure difference data inside and outside the operating room to control the rotation speed of the fresh air unit and the circulation unit, as well as the opening of the fresh air control valve and the return air control valve to maintain the air supply volume, ventilation frequency, wind speed and pressure difference in the operating room. The system can flexibly and accurately control the air supply quality of a single operating room and reduce energy consumption. However, the prior art does not consider the impact of the spatiotemporal distribution of medical staff and equipment on the intensity of pollution sources during surgery, lacks real-time monitoring and dynamic analysis of the distribution of pollution sources inside the operating room, and is difficult to achieve real-time optimization and adjustment of the air supply strategy according to the dynamic changes of the surgical process. It is not adaptable enough in complex surgical environments.

[0004] On the other hand, the Chinese patent application with publication number CN115574403A proposes a precise humidity control air treatment method for clean operating rooms, which precools and dehumidifies the fresh air, mixes the treated fresh air with the return air, and dynamically adjusts the temperature and humidity of the mixed air to offset the wet load of the operating room, thereby achieving precise control of the temperature and humidity of the operating room. This method focuses on achieving dynamic balance of the wet load of the operating room through fresh air pretreatment, but does not involve the prediction and balance control of the heat load, nor does it consider the dynamic changes in the distribution of personnel and equipment during the operation, making it difficult to fully optimize the thermal and humid environment of the operating room.

[0005] In general, the existing operating room air supply control methods lack real-time monitoring of the temporal and spatial distribution of medical staff and equipment, making it difficult to accurately characterize the dynamic changes of pollution sources during surgery. It is difficult to balance the distribution of pollution sources and personnel comfort in the operating room while meeting the dynamic needs of the surgical process. The comprehensive optimization level of air supply control needs to be further improved. Summary of the invention

[0006] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent control method and system for a laminar flow air supply unit dedicated to an operating room, which monitors the temporal and spatial distribution of medical staff and equipment in real time, and accurately depicts the dynamic changes of pollution sources during the operation; dynamically predicts the heat and humidity load requirements of the operating room in future time periods according to the real-time progress of the operation, and improves the foresight and timeliness of the air supply strategy; while meeting the dynamic requirements of the operation process, takes into account the distribution of pollution sources and the comfort of personnel in the operating room, and further improves the comprehensive optimization level of air supply control.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] The intelligent control method of the laminar flow air supply unit for operating room includes:

[0009] Construct a three-dimensional spatial topological model of the operating room. Based on the three-dimensional spatial topological model of the operating room, collect the spatial position data of medical staff and equipment during the operation in real time to form a spatiotemporal trajectory data set of medical staff and equipment; analyze the spatiotemporal trajectory data set to generate a heat map of pollution source distribution;

[0010] The environmental status data stream and personnel status data stream of the operating room are obtained, and the environmental status data stream and the personnel status data stream are input into the pre-trained heat and humidity load demand prediction model to obtain the dynamic heat and humidity load demand curve of the operating room in the future period; based on the dynamic heat and humidity load demand curve of the operating room in the future period, the three-dimensional spatial topological model of the operating room and the thermal map of the pollution source distribution, a dynamic air supply control strategy is generated; the dynamic air supply control strategy is transmitted to the laminar air supply unit for intelligent air supply control.

[0011] Furthermore, constructing the three-dimensional spatial topological model of the operating room includes:

[0012] Perform a full-scale scan of the interior space of the operating room to obtain high-precision point cloud data;

[0013] Extract geometric information of key components inside the operating room from high-precision point cloud data and reconstruct the 3D solid model of the operating room;

[0014] Demarcate the clean partition boundary in the three-dimensional solid model of the operating room and generate clean partition parameter information;

[0015] The 3D entity model of the operating room is integrated with the parameter information of the clean partition to construct a 3D spatial topological model of the operating room.

[0016] Furthermore, the analyzing the spatiotemporal trajectory data set to generate a pollution source distribution heat map includes:

[0017] The activity trajectories of medical staff and equipment are extracted from the spatiotemporal trajectory dataset, and residence events are identified from the activity trajectories; a pollution source intensity model is constructed, and based on the pollution source intensity model, the pollution source intensity of each residence event is calculated to generate a residence event dataset; based on the residence event dataset, a pollution source distribution heat map is generated.

[0018] Further, identifying the dwell event from the activity track includes:

[0019] Discretize the activity trajectory into a series of space-time coordinate points to form a trajectory point sequence;

[0020] Taking the spatial distribution density of trajectory point sequence as the criterion, the density-based spatial clustering algorithm is applied to extract candidate residence points;

[0021] A spatiotemporal neighborhood of the candidate dwelling point is constructed, and a dwelling event is generated based on the candidate dwelling point and the spatiotemporal neighborhood of the candidate dwelling point.

[0022] Furthermore, the step of constructing the spatiotemporal neighborhood of the candidate residence point includes:

[0023] For the candidate residence point p0, a neighborhood N(p0) with a time span of Δt1 and a space span of d is constructed with p0 as the center; for any candidate residence point p1, if p1∈N(p0), p0 is determined to be adjacent to p1 in time and space.

[0024] Further, generating a resident event includes:

[0025] Medical staff and equipment are defined as resident subjects; any two candidate resident points of any resident subject are defined as resident segments;

[0026] If two candidate dwelling points in any dwelling segment of a dwelling subject are adjacent in time and space, then the dwelling segment is a continuous dwelling of the dwelling subject;

[0027] All consecutive stays of the stay subject are merged into a stay event, and each stay event is represented by a four-tuple (s, t0, Δt, l ) is used to represent the residence; where s is the residence subject, t0 is the residence start time, Δt is the residence duration, l are the resident location coordinates; the resident subjects include medical personnel and equipment.

[0028] Furthermore, the pollution source intensity model includes a human pollution source intensity model P A and equipment pollution source intensity model P B ;

[0029] If the main residents of the residence event are medical staff, the personnel pollution source intensity model P is used. ACalculate the pollution source intensity of the residence event; if the residence subject of the residence event is equipment, the equipment pollution source intensity model P is used B Calculate the pollution source intensity of the dwelling event.

[0030] Furthermore, generating a pollution source distribution heat map according to the resident event data set includes:

[0031] Performing spatiotemporal aggregation on the resident event data set to form a spatiotemporal cube representation of the pollution source distribution; the pollution source distribution represents the distribution state of the resident events in the three-dimensional spatial topological model of the operating room, which is jointly determined by the resident position coordinates of each resident event and the pollution source intensity;

[0032] Slice the space-time cube in the spatial dimension, discretize the continuous spatial coordinates of the space-time cube into three-dimensional spatial grids, map the pollution source distribution into three-dimensional spatial grid data, and the attribute value of each three-dimensional spatial grid represents the intensity of the pollution source;

[0033] The three-dimensional spatial raster data is smoothed using the pollution source intensity as the weight to generate a heat map of the pollution source distribution.

[0034] Furthermore, the personnel status data stream includes the number and body surface area of ​​medical personnel; the body surface area of ​​the medical personnel is calculated from the height and weight of the medical personnel;

[0035] The generating of the dynamic air supply control strategy comprises:

[0036] Obtain the layout information of the air supply equipment, and construct a multi-source and multi-sink air supply path directed graph based on the three-dimensional spatial topological model of the operating room and the layout information of the air supply equipment;

[0037] Based on the dynamic heat and humidity load demand curve of the operating room in the future period and the thermal map of pollution source distribution, the optimal air supply tree is solved in the multi-source and multi-sink air supply path directed graph;

[0038] The optimal air supply tree is mapped to a specific combination of air supply equipment to form a dynamic air supply control strategy.

[0039] An intelligent control system for a laminar flow air supply unit for an operating room, which is used to implement the intelligent control method for the laminar flow air supply unit for an operating room, and the system comprises:

[0040] Heat map generation module: used to build a three-dimensional spatial topological model of the operating room. Based on the three-dimensional spatial topological model of the operating room, the spatial position data of medical staff and equipment during the operation are collected in real time to form a spatiotemporal trajectory data set of medical staff and equipment; the spatiotemporal trajectory data set is analyzed to generate a heat map of the pollution source distribution;

[0041] Load prediction module: used to obtain the environmental status data stream and personnel status data stream of the operating room, input the environmental status data stream and the personnel status data stream into the pre-trained heat and humidity load demand prediction model, and obtain the dynamic heat and humidity load demand curve of the operating room in the future period;

[0042] Air supply strategy generation module: Generates dynamic air supply control strategy based on the dynamic heat and humidity load demand curve of the operating room in the future period, the three-dimensional spatial topology model of the operating room and the thermal map of pollution source distribution;

[0043] Air supply control module: used to transmit the dynamic air supply control strategy to the laminar air supply unit for intelligent air supply control.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention constructs a three-dimensional spatial topological model of the operating room and a heat map of the pollution source distribution, accurately depicts the temporal and spatial distribution characteristics of the pollution sources during the operation, combines environmental status and personnel status data, dynamically predicts the heat and humidity load requirements of the operating room, and optimizes the air supply control strategy on this basis. Compared with the traditional constant air volume or constant wind speed control method, the present invention can dynamically adjust the air supply parameters according to the real-time status of the surgical process, and minimize the air supply energy consumption while meeting the cleanliness requirements of the operating room.

[0046] In addition, the present invention can also identify high-risk pollution areas based on the spatiotemporal trajectories of medical staff and equipment, increase local air supply in a targeted manner, and further improve the quality of the operating room environment. At the same time, with the help of the three-dimensional spatial topological model of the operating room, the present invention can optimize the air supply path, reduce airflow short circuits and stagnation, and improve air supply uniformity.

[0047] The intelligent control method provided by the present invention can significantly improve the intelligence level of the operating room air-conditioning system, reduce energy consumption while ensuring a clean environment in the operating room, and is of great significance for improving the air quality in the operating room, ensuring medical safety, and saving hospital operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 creative work.

[0049] Figure 1 It is a principle flow chart of the intelligent control method of the laminar air supply unit dedicated to the operating room in the present invention;

[0050] Figure 2A flow chart of a method for constructing a three-dimensional spatial topological model of an operating room in the intelligent control method of a laminar air supply unit dedicated to an operating room of the present invention;

[0051] Figure 3 A flow chart of a method for analyzing a spatiotemporal trajectory data set and generating a pollution source distribution thermodynamic map in an intelligent control method for a laminar air supply unit dedicated to an operating room of the present invention;

[0052] Figure 4 A flow chart of a method for identifying a resident event from an activity track in the intelligent control method of a laminar air supply unit dedicated to an operating room of the present invention;

[0053] Figure 5 A flow chart of a method for generating a dwell event in the intelligent control method of the laminar air supply unit dedicated to an operating room of the present invention;

[0054] Figure 6 A flow chart of a method for generating a resident event data set in an intelligent control method for a laminar air supply unit dedicated to an operating room of the present invention;

[0055] Figure 7 A flow chart of a method for generating a pollution source distribution thermodynamic map in an intelligent control method for a laminar air supply unit dedicated to an operating room of the present invention;

[0056] Figure 8 A flow chart of a method for generating a dynamic air supply control strategy in the intelligent control method of the laminar air supply unit dedicated to an operating room of the present invention;

[0057] Fig. 9 This is a functional module diagram of the intelligent control system of the laminar air supply unit dedicated to the operating room in the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Example 1

[0060] See also Figure 1 As shown, this embodiment provides an intelligent control method for a laminar air supply unit dedicated to an operating room, including:

[0061] Step S1000, constructing a three-dimensional spatial topological model of the operating room, based on which the spatial position data of medical staff and equipment during the operation are collected in real time to form a spatiotemporal trajectory data set of medical staff and equipment; analyzing the spatiotemporal trajectory data set to generate a heat map of pollution source distribution;

[0062] Furthermore, step S1000 includes:

[0063] Step S1100, constructing a three-dimensional spatial topological model of an operating room;

[0064] Furthermore, if Figure 2 As shown, step S1100 includes:

[0065] Step S1110, performing an all-round scan of the interior space of the operating room to obtain high-precision point cloud data;

[0066] Step S1120, extracting geometric information of key components inside the operating room from the high-precision point cloud data, and reconstructing a three-dimensional solid model of the operating room;

[0067] Step S1130, calibrating the clean partition boundary in the three-dimensional solid model of the operating room and generating clean partition parameter information;

[0068] Step S1140: The three-dimensional entity model of the operating room is integrated with the parameter information of the clean partition to construct a three-dimensional spatial topological model of the operating room.

[0069] Specifically, a 3D laser scanner is used to perform an all-round, blind-angle scan of the interior space of the operating room to obtain high-precision point cloud data that can reflect the structural characteristics of the internal space. The principle of 3D laser scanning is to determine the 3D coordinates of discrete points in space by the distance measurement method. The scanner emits a laser beam to the target object and receives its reflected echo. By recording the emission direction and flight time of the laser beam, the spatial rectangular coordinates of the reflection point relative to the scanner can be calculated. By repeatedly scanning the same area from multiple angles, a large amount of point cloud data covering the surface of the operating room can be obtained, and then a high-precision 3D digital model of the interior of the operating room can be reconstructed. Compared with ordinary point clouds, high-precision point cloud data has the characteristics of high point density, low noise, and rich details. It is an ideal data source for building high-fidelity solid models. Obtaining high-precision point cloud data provides a geometric basis for subsequent 3D reconstruction.

[0070] From the high-precision point cloud data, the point cloud data processing algorithm is used to extract the geometric information of key components such as the walls, floors, ceilings, and large equipment inside the operating room, and based on this, a three-dimensional solid model of the operating room that is highly consistent with the physical space is reconstructed. Commonly used point cloud processing algorithms include point cloud filtering, segmentation, and registration. Among them, the filtering algorithm can remove outliers and noise points in the point cloud data to improve the data quality; the segmentation algorithm can divide the point cloud into multiple sub-point clouds according to certain rules to facilitate the extraction of independent components; the registration algorithm can spatially splice multiple point clouds obtained from different perspectives to obtain a complete scene point cloud. Through a series of processing of the original point cloud data, a realistic three-dimensional solid model of the operating room is finally reconstructed in the digital space. This model can not only accurately reflect the spatial geometric relationship of the main components such as walls and floors, but also express the detailed features of auxiliary components such as ceilings, columns, and equipment, providing a model basis for the visual management and analysis of the surgical environment.

[0071] On the basis of the constructed three-dimensional solid model of the operating room, the specific boundary ranges of different clean areas such as the sterile area, semi-contaminated area, and contaminated area are defined according to the plan layout of the operating room, and the spatial mapping relationship between each clean partition and the three-dimensional solid model is established. At the same time, according to the cleanliness level requirements of each clean partition, the corresponding parameterized description information of the clean partition is generated. These description information covers multiple environmental parameters such as air cleanliness, ventilation frequency, temperature and humidity, and pressure difference of each partition, forming a standardized parameter set that meets standards such as ISO 14644. Through this step, a one-to-one correspondence between the physical space division and the digital space division of the operating room is established, which realizes the refined expression and management of the clean partition, and provides a basis for independent control of the partition.

[0072] The three-dimensional entity model of the operating room constructed in step S1120 is integrated with the parameter information of the clean partition generated in step S1130 to establish a unified three-dimensional spatial topological model of the operating room including environmental semantics. In terms of technical implementation, the data structure and storage mechanism of the topological model can be constructed through technologies such as graph databases to achieve full-element parametric expression of the operating environment. The so-called topological model refers to a data model that describes the logical association between entities through topological elements such as nodes, edges, and faces on the basis of maintaining the spatial adjacency relationship between entities. It emphasizes the spatial topological relationship between objects more than the geometric model. In the three-dimensional spatial topological model of the operating room, the topological relationship between the clean partition and the room boundary can be modeled as a "containment" relationship, the relationship between medical staff / equipment and the clean partition can be modeled as a "located" relationship, and the relationship between personnel / equipment can be modeled as a "neighboring" relationship, and so on. By constructing a semantically rich three-dimensional spatial topological model, the spatiotemporal characteristics of the operating environment can be more realistically reflected, thereby providing a digital twin foundation for intelligent environmental management decisions.

[0073] The beneficial effects of constructing a three-dimensional spatial topological model of the operating room are mainly reflected in the following aspects:

[0074] Realize the digital description of the surgical environment. Through the three-dimensional parametric modeling of the operating room layout and clean partitions, a digital mirror of the physical environment is formed, laying a data foundation for applications such as visual supervision and data analysis.

[0075] Supports independent control of different clean areas. The boundaries of each clean area and its environmental parameters are clearly defined in the topology model, providing a basis for independent control of each area, thereby meeting the customized environmental requirements of different areas.

[0076] Facilitates the semantic expression of spatial data. The topological model contains environmental semantic information, which can be mapped with the ontology knowledge base to give environmental data clear semantics, making it easier to achieve data-driven intelligent management and control.

[0077] Provide an environmental carrier for digital twins. The three-dimensional spatial topological model of the operating room is an important environmental carrier for the development of digital twins, providing realistic environmental boundary conditions for simulation, prediction, and optimization in the digital space.

[0078] In summary, constructing a three-dimensional spatial topological model of the operating room plays a key role in supporting refined management and intelligent control through the digital and semantic expression of the surgical environment. It is an important basis for realizing the "intelligent air supply control method for operating rooms based on pollution distribution heat map".

[0079] Step S1200, based on the three-dimensional spatial topological model of the operating room, the spatial position data of the medical staff and equipment during the operation are collected in real time to form a spatiotemporal trajectory data set of the medical staff and equipment;

[0080] Specifically, first of all, it is necessary to select a suitable indoor positioning sensor to collect dynamic position data of medical staff and equipment. In this embodiment, a micro-optical sensor with a wavelength of λ nanometers is selected as a positioning sensor. When selecting a model, it is necessary to comprehensively consider multiple performance parameters of the sensor, such as detection accuracy, sampling frequency, signal penetration, etc., weigh the sensor performance and cost, and finally determine the model specifications that meet the requirements. Generally speaking, the higher the detection accuracy, the faster the sampling frequency, and the better the signal penetration, the more accurate the collected position data, the more real-time the response, and the less interference from occlusion, but the cost of the sensor will also increase accordingly. It is necessary to combine the actual needs of the surgical environment to select an economical sensor that meets the positioning performance requirements.

[0081] After determining the positioning sensor model, it is necessary to further optimize the spatial layout of sensor nodes. Due to the complex internal space of the operating room, there are differences in the degree of openness of vision and signal transmission channels at different locations, so the spatial layout of the sensor will significantly affect the positioning effect. In order to achieve full coverage dynamic positioning of the entire operating room, it is necessary to reasonably plan the spatial distribution of the sensor array. Here, the sensor layout problem can be modeled as a key point visual domain coverage problem in a three-dimensional spatial topological model. By solving this problem, a sensor layout scheme that meets the positioning requirements of the operating room and has better comprehensive performance is obtained. The so-called visual domain coverage problem refers to the layout of fewer sensors in three-dimensional space so that each target point can be "seen" by at least one sensor (that is, within the detection range of the sensor). By introducing a three-dimensional spatial topological model, the geometric environment inside the operating room can be accurately described. On this basis, the spatial topological relationship between the sensor and the target point (such as "visible" and "occluded" etc.) is established, and the optimization model and solution algorithm of the sensor layout are constructed accordingly. By solving the model, a better sensor array layout scheme can be obtained, which minimizes the number of required sensors while meeting the global detection accuracy requirements and saves the system construction cost.

[0082] The location data collected in real time by each sensor node is aggregated and fused, and finally a unified format of personnel / equipment spatiotemporal trajectory data set can be generated. This spatiotemporal trajectory data set records the dynamic position changes of each participant during the operation in the form of a standardized spatiotemporal sequence, and the spatial coordinates of each target at the current moment can be obtained at each timestamp. The high-quality trajectory data set clearly depicts the spatiotemporal dynamic characteristics of the surgical process, and is the basis for further analysis such as behavior semantic extraction, residence identification, and pollution source tracing.

[0083] Step S1300: Analyze the spatiotemporal trajectory data set to generate a pollution source distribution heat map.

[0084] Furthermore, if Figure 3 As shown, step S1300 includes:

[0085] Step S1310, extracting the activity trajectories of medical staff and equipment from the spatiotemporal trajectory dataset;

[0086] Specifically, after obtaining a complete spatiotemporal trajectory dataset, it is necessary to further extract the independent activity trajectories of medical staff and equipment. Since the original spatiotemporal trajectory data is usually obtained by discrete sampling, its temporal continuity and spatial integrity cannot be guaranteed. The movement states of medical staff and equipment at different times and in different areas may not be sampled. Therefore, it is necessary to design a suitable trajectory splicing algorithm to splice discontinuous discrete trajectory fragments into relatively complete and smooth trajectories according to certain rules.

[0087] The commonly used trajectory splicing algorithm is a method based on graph optimization. The basic idea is to regard the discrete trajectory sampling points as nodes in the graph, and the spatiotemporal relationship between nodes as the edge of the graph. The trajectory splicing problem is transformed into the problem of finding the node sequence with the smallest global error in the graph. By introducing various prior constraints (such as the smoothness assumption of the trajectory, the speed / acceleration assumption, etc.), an optimization objective function that reflects the spatiotemporal dependency between nodes can be constructed. Common objective functions include geodesic distance, Mahalanobis distance, etc. They characterize the spatiotemporal similarity and consistency between nodes from different aspects. On the basis of the objective function, by introducing penalty terms, regularization terms, etc., prior constraints can be further imposed on the optimization process. Finally, by solving the optimization problem, the optimal node arrangement that minimizes the global objective function can be obtained, and the splicing and correction of discrete trajectory points can be realized.

[0088] Trajectory stitching can not only restore continuous and complete activity trajectories, but also eliminate noise caused by sensor errors, packet loss, etc., and smooth the trajectory data. Stitching and smoothing the original spatiotemporal trajectory data of medical staff and equipment can effectively restore the spatiotemporal activity patterns of personnel and materials during surgery, facilitating subsequent more in-depth semantic analysis such as resident identification. At the same time, trajectory stitching fills the gap of discrete sampling, reducing the density of sensor deployment and the amount of data while still fully recording the movement process, saving hardware and storage costs.

[0089] Step S1320, identifying a dwell event from the activity track;

[0090] Furthermore, if Figure 4 As shown, step S1320 includes:

[0091] Step S1321, discretizing the activity trajectory into a series of spatiotemporal coordinate points to form a trajectory point sequence;

[0092] Step S1322, taking the spatial distribution density of the trajectory point sequence as a criterion, applying a density-based spatial clustering algorithm to extract candidate residence points;

[0093] Step S1323: construct a spatiotemporal neighborhood of the candidate stay point, and generate a stay event based on the candidate stay point and the spatiotemporal neighborhood of the candidate stay point.

[0094] The construction of the spatiotemporal neighborhood of the candidate residence point includes:

[0095] For the candidate residence point p0, a neighborhood N(p0) with a time span of Δt1 and a space span of d is constructed with p0 as the center; for any candidate residence point p1, if p1∈N(p0), p0 is determined to be adjacent to p1 in time and space.

[0096] Furthermore, if Figure 5 As shown, step S1323 includes:

[0097] Step S13231, defining medical staff and equipment as resident subjects, and defining any two candidate resident points of any resident subject as resident segments;

[0098] Step S13232: if two candidate residence points in any residence segment of the resident subject are adjacent in time and space, then the residence segment is a continuous residence of the resident subject;

[0099] Step S13233: merge all the consecutive stays of the stay subject into a stay event. Each stay event is represented by a four-tuple (s, t0, Δt, l ) is used to represent the residence; where s is the residence subject, t0 is the residence start time, Δt is the residence duration, l are the resident location coordinates; the resident subjects include medical personnel and equipment.

[0100] Specifically, after obtaining a smooth and complete activity trajectory, it is necessary to further detect residence events with specific semantics. The so-called residence event refers to the long-term stay of personnel or equipment in a certain area. Accurate identification of residence events is of great significance for analyzing the surgical process and optimizing the diagnosis and treatment process. The spatiotemporal characteristics of residence events can directly reflect the diagnosis and treatment behaviors of medical staff and the placement status of items, and then reflect the spatiotemporal process of the surgical process. By exploring the residence patterns of different subjects in typical areas, the spatiotemporal elements of surgical operations can be characterized, and the layout and scheduling can be optimized.

[0101] However, since there is no strict definition of residence at the semantic level, it is difficult to directly characterize residence events from a mathematical perspective. The determination of residence requires comprehensive consideration of two core elements: spatial clustering and temporal continuity. Therefore, identifying residence events from activity trajectories can be divided into two steps: spatial clustering and temporal association.

[0102] In step S1321, the continuous trajectory curve is first uniformly sampled to obtain a discrete trajectory point sequence, including the spatial coordinates and timestamp of each trajectory point. The trajectory point sequence is the basic data structure for subsequent cluster analysis. Next, in step S1322, the spatial distribution density characteristics of the trajectory point sequence are examined. Intuitively, if the spatial density of the trajectory points in the local area is very high, it can be preliminarily judged that there is potential residence behavior in the area. Commonly used spatial clustering algorithms include DBSCAN, OPTICS, etc., which classify sample points based on spatial density and can effectively detect densely distributed point clusters. Taking DBSCAN as an example, it has two key parameters: scanning radius ε and minimum number of included points minPts. If the number of sample points contained in the area with any core object as the center and ε as the radius is not less than minPts, then these sample points are connected in density and belong to the same cluster; otherwise, they are noise points or boundary points. Through cluster analysis, several point clusters with prominent density can be found, and their centroids are used as candidate residence points. Each candidate stay point actually represents a potential stay location of a person or equipment at a certain moment.

[0103] Finally, in step S1323, the continuity of the candidate dwelling points in the time dimension is examined. For adjacent candidate dwelling points of the same subject, if they are in the spatiotemporal neighborhood, they can be further aggregated into a macro dwelling event. By iteratively expanding the neighborhood of the candidate dwelling points, the spatiotemporal adjacent dwelling segments are merged into continuous dwelling events, thereby achieving aggregation from micro-dwelling to macro-dwelling. After the dwelling event is formed, the semantic extraction of the activity trajectory is completed.

[0104] Dwelling events are high-frequency events during surgery, and their patterns and efficiency have a significant impact on the quality of surgery. By identifying dwellings from continuous trajectories, on the one hand, we can characterize the temporal and spatial distribution patterns of personnel in various links and areas of surgery and evaluate the diagnosis and treatment process; on the other hand, we can examine the utilization and placement strategies of equipment and items and optimize instrument management. Therefore, the temporal and spatial characteristic analysis of dwelling events is of great value for improving operating room management and improving surgical efficiency. At the same time, identifying dwellings is also crucial for subsequent pollution tracing. Dwellings represent the long-term stay of personnel / equipment in a specific area, during which the corresponding pollution release is much higher than that of instantaneous passage. Therefore, accurately characterizing dwelling behavior is a prerequisite for quantitatively analyzing the diffusion pathways and concentration distribution of pollution sources.

[0105] In short, extracting the resident segments from continuous trajectories and giving them clear spatiotemporal semantics is the key to achieving behavior recognition and event detection. The multi-scale spatiotemporal association method based on density clustering can make up for the limitations of single feature analysis, integrate the two perspectives of time and space, and accurately mine valuable resident events from complex trajectories, thereby providing data support for surgical process optimization and contamination control. As the intermediate semantics between low-level trajectories and high-level events, resident events play an important role in the "behavior-event" multi-layer semantic system.

[0106] Step S1330, constructing a pollution source intensity model, and based on the pollution source intensity model, calculating the pollution source intensity of each resident event, and generating a resident event data set;

[0107] Furthermore, if Figure 6 As shown, step S1330 includes:

[0108] Step S1331, define the pollution source intensity index, and build a pollution source intensity model based on the pollution source intensity index; the pollution source intensity model includes a personnel pollution source intensity model P A and equipment pollution source intensity model P B ;

[0109] Step S1332: Based on the pollution source intensity model, the pollution source intensity of each resident event is calculated to generate a resident event data set carrying pollution attribute labels.

[0110] If the main residents of the residence event are medical staff, the personnel pollution source intensity model P is used. A Calculate the pollution source intensity of the residence event; if the residence subject of the residence event is equipment, the equipment pollution source intensity model P is used B Calculate the pollution source intensity of the dwelling event.

[0111] The personnel pollution source intensity model P A include:

[0112]

[0113] in:

[0114] : The respiratory rate of the person (times / minute), obtained through respiratory monitoring equipment;

[0115] : The length of time a person stays at a stay location, determined by the stay event extract;

[0116] : The number of conversations that the personnel have during their stay, obtained through voice recognition technology or manual recording;

[0117] : Weight coefficient of respiratory rate;

[0118] : Weight coefficient of residence time;

[0119] : The weight coefficient of the number of conversations, Need to be determined based on medical research or statistical analysis;

[0120] : Real-time temperature of the station location ( ), obtained through the temperature sensor;

[0121] : The temperature of the station meets the requirements of the surgical environment , determined by environmental standards or experiments;

[0122] : Real-time humidity of the station location (%), obtained through the humidity sensor;

[0123] : Humidity (%) at the resident position that meets the requirements of the surgical environment, determined by environmental standards or experiments;

[0124] : The influence coefficient of temperature deviation from the reference value that meets the requirements of the surgical environment in the personnel pollution source intensity model;

[0125] : The influence coefficient of humidity deviation from the reference value that meets the requirements of the surgical environment in the personnel pollution source intensity model, and Obtained through experimental measurement or simulation analysis.

[0126] The intensity of personnel pollution sources is proportional to the personnel's breathing rate, length of stay, and number of conversations;

[0127] The greater the deviation of temperature and humidity from the reference values ​​that meet the requirements of the surgical environment, the faster the contamination spreads and the intensity of the pollution source increases accordingly.

[0128] Equipment pollution source intensity model P B :

[0129]

[0130] in:

[0131] W: Equipment operation intensity (such as power, speed, etc.), which needs to be obtained from the equipment monitoring system;

[0132] : The proportion of the equipment's operating time during the residence time, calculated through the equipment's power on and off records;

[0133] : Weight coefficient of equipment operation intensity;

[0134] : Weight coefficient of equipment operation time ratio, Need to be estimated based on equipment characteristics and experience;

[0135] : The influence coefficient of the temperature deviation from the reference value that meets the requirements of the surgical environment in the equipment pollution source intensity model;

[0136] : The influence coefficient of humidity deviation from the reference value that meets the requirements of the surgical environment in the equipment pollution source intensity model, and Obtained through experimental measurement or simulation analysis.

[0137] The intensity of equipment pollution sources is proportional to the operating intensity and operating time of the equipment; it is also affected by the nonlinear effects of temperature and humidity, and the pollution intensity increases when it deviates from the reference value that meets the requirements of the surgical environment.

[0138] With other conditions unchanged, the intensity of pollution sources increases monotonically with the increase of residence time, respiratory rate, number of conversations, equipment operation intensity and operation time, indicating that the cumulative effect of various types of pollution is significant; at the same time, the intensity of pollution sources increases sharply with the increase in the degree of deviation of temperature and humidity from the reference values ​​that meet the requirements of the surgical environment, showing nonlinear characteristics, reflecting the amplifying effect of environmental factors on the spread of pollution. Therefore, in the management of clean surgical environments, it is necessary to focus on high-pollution risk scenarios such as long-term residence of personnel, frequent conversations, and high-intensity equipment operation, and strictly control the temperature and humidity in a constant numerical range that meets the requirements of surgical environment air quality assurance, thereby effectively reducing the intensity of pollution sources and ensuring the air quality of the surgical environment.

[0139] The pollution source intensity model quantitatively evaluates the pollution degree of different residence events, providing a basis for identifying key pollution sources; reveals the intrinsic connection between personnel behavior, equipment operation and pollution diffusion, provides decision-making support for the intelligent control and dynamic scheduling of air purification devices, and improves purification effect and energy efficiency; supports the prediction and early warning of air quality under different scenarios, and provides a reference for the refined management of clean surgical environments.

[0140] After identifying the residence events during the operation, it is necessary to further characterize the pollution release intensity of different residence events and construct a pollution source intensity model. The so-called pollution source intensity model refers to a mathematical model that reflects the mass of pollutants released into the environment by different residence events per unit time. Through this model, the pollution emission level of various residence events can be quantitatively evaluated, and then the pollution source label can be introduced into the residence event attributes, and finally a residence event data set with pollution attributes can be obtained.

[0141] For each identified residence event (such as a doctor's residence in front of the operating table, a nurse's residence in the instrument area, etc.), first extract the residence subject (personnel, equipment), residence duration, number of conversations of personnel, breathing frequency, equipment operation intensity, operation time ratio, environmental parameters during residence and other attribute information, and input them into the constructed pollution source intensity model, and then calculate the theoretical total amount of pollutants released by the residence event during the entire residence period. The pollutant release amount is used as the pollution source intensity attribute value of the event, and the residence event is assigned a pollution label. After calculations one by one, the original residence event data set is expanded into a residence event data set with pollution attributes, and each event record is added with a pollution source intensity field. In layman's terms, it is to "portrait" different residence events, characterize their pollution emission characteristics in different regions and stages, and use attribute labels to intuitively characterize the degree of impact of the event on local air quality.

[0142] The introduction of the pollution source intensity model to estimate the pollution attributes of the resident events is a further deepening of the semantic level of the activity trajectory. It associates the resident events with the surgical environment parameters to build a data bridge between people, things and the environment. On this basis, the spatiotemporal tracing, process deconstruction, and trend prediction of surgical pollution can be achieved. For example, by comparing and sorting the pollution intensity of resident events at different stages and in different areas, high-risk personnel and high-risk areas of pollution release can be found, and targeted control measures can be formulated; for example, by correlating the changes in pollution intensity before and after different events, the air impact of different operating procedures can be portrayed and the surgical process can be optimized. The introduction of pollution attribute data enables environmental intelligent analysis to move from macro to fine, providing a data basis for applications such as pollution tracing and efficiency evaluation.

[0143] Step S1340: Generate a pollution source distribution heat map based on the resident event data set.

[0144] Furthermore, if Figure 7 As shown, step S1340 includes:

[0145] Step S1341, performing spatiotemporal aggregation on the resident event data set to form a spatiotemporal cube representation of the pollution source distribution; the pollution source distribution represents the distribution state of the resident events in the three-dimensional spatial topological model of the operating room, which is jointly determined by the resident position coordinates of each resident event and the pollution source intensity;

[0146] Step S1342, slicing the space-time cube in the spatial dimension, discretizing the continuous spatial coordinates of the space-time cube into three-dimensional spatial grids, mapping the pollution source distribution into three-dimensional spatial grid data, and the attribute value of each three-dimensional spatial grid represents the intensity of the pollution source;

[0147] Step S1343, smoothing the three-dimensional space grid data using the pollution source intensity as a weight to generate a pollution source distribution heat map.

[0148] Specifically, identifying resident events and quantifying pollution intensity are only intermediate steps. The ultimate goal is to intuitively depict the dynamic evolution of pollution throughout the entire surgical process. To this end, it is necessary to generate a heat map visualization of the distribution of pollution sources based on the resident event data set carrying pollution attribute labels. The heat map can vividly reflect the differences in pollution levels in different time and space regions and provide intuitive awareness of the pollution situation.

[0149] First, it is necessary to perform spatiotemporal aggregation on the resident event data to obtain the evolution characteristics of the pollution source distribution along the dual dimensions of time and space. The core of aggregation is to map discrete event data into a continuous spatiotemporal framework to build a unified representation form. In specific implementation, the classic spatiotemporal data representation model, the space-time cube, can be used. The so-called space-time cube refers to a three-dimensional Cartesian coordinate system, in which two coordinate axes represent the spatial dimension and the third coordinate axis represents the time dimension. Any point inside the cube can be represented by its spatiotemporal coordinates in the form of (x, y, t), reflecting the state of the position at time t. The surgical process is discretized into finite time slices, each time slice corresponds to a time section, forming a two-dimensional plane sequence stacked along the time axis. In each section, the spatial grid is divided according to the layout of the operating room, and each grid is mapped to a pixel on the plane. In this way, the distribution of pollution sources throughout the surgical process is represented as a stack of a series of two-dimensional space-time sections, and the intensity of the pollution source becomes the attribute value of each pixel. Based on the unified space-time cube representation, the originally discrete residence event data is converted into regular space-time series data, so that the distribution pattern of pollution sources can be analyzed under a unified space-time benchmark.

[0150] Next, the space-time cube data needs to be further discretized in the spatial dimension to match the subsequent generation of heat maps. Through slicing perpendicular to the time axis, the continuous plane coordinates on each time section are discretized into a limited space grid, and each grid corresponds to a tiny area in reality. The projection of a series of grids at the same position on the time axis forms a virtual "pixel column", reflecting the dynamic changes in the pollution level in the area. The originally continuously distributed pollution source intensity is discretized into attribute values ​​on the grid, and the gradient characteristics in space are represented by the difference between adjacent grids. After discretization, the space-time cube is represented as a regular three-dimensional grid array, and the attribute value of each grid represents the average pollution intensity in the corresponding space-time area. This rasterized representation helps to reduce data complexity and improve analysis efficiency on the one hand; on the other hand, it also provides a natural mapping unit for the generation of heat maps. The pollution sources distributed in the three-dimensional space are anchored in a fixed space-time coordinate system, so that the dynamic changes in the pollution level can be uniformly characterized.

[0151] Finally, based on the above three-dimensional spatial grid data, a continuous and smooth pollution distribution estimate is obtained through spatial interpolation to generate a pollution source distribution heat map. Commonly used spatial interpolation algorithms include inverse distance weighted interpolation, Kriging interpolation, spline function interpolation, etc. Among them, kernel density estimation (KDE) has become the mainstream method for generating heat maps due to its unique non-parametric characteristics and strong micro-sensitivity. KDE estimates the attribute value of the location by weighted accumulation of known data points in a certain neighborhood around the target location. Starting from the discrete pollution source intensity data, KDE makes full use of local spatial autocorrelation, and well describes the local spatial heterogeneity while maintaining the overall trend of pollution distribution. After KDE reconstruction, the original discrete grid pollution distribution is "fuzzy" to generate a smooth pollution intensity distribution map. By selecting a suitable color scheme and generating a color scale, a pollution source heat map can be obtained. The heat map vividly and intuitively reflects the pollution level in different time and space areas throughout the operation. The high-risk area forms a "hot spot" in the map, which has a strong visual impact and is convenient for pollution tracing, situation awareness and risk warning.

[0152] In short, the process of generating a heat map of pollution source distribution is a process of extracting pollution information from discrete to continuous, from qualitative to quantitative. A unified space-time cube representation is obtained through spatiotemporal aggregation, data rasterization is achieved through spatial slicing, and a continuous and smooth distribution estimate is generated through KDE interpolation, ultimately forming a heat map of pollution sources. This heat map is more intuitive and easier to interpret than the original resident event data, and can provide visual auxiliary decision-making for operating room environmental management. On this basis, we can further carry out high-risk period identification, key pollution source location, cleaning path planning, etc., to more finely guide the air quality control of the operating room.

[0153] Step S2000, obtain the environmental status data stream and personnel status data stream of the operating room, input the environmental status data stream and the personnel status data stream into the pre-trained heat and moisture load demand prediction model, and obtain the dynamic heat and moisture load demand curve of the operating room in the future time period; based on the dynamic heat and moisture load demand curve of the operating room in the future time period, the three-dimensional spatial topological model of the operating room and the pollution source distribution thermal map, generate a dynamic air supply control strategy; transmit the dynamic air supply control strategy to the laminar air supply unit to perform intelligent air supply control.

[0154] Furthermore, step S2000 includes:

[0155] Step S2100, obtaining the environment status data stream and the personnel status data stream of the operating room, inputting the environment status data stream and the personnel status data stream into the pre-trained heat and humidity load demand prediction model, and obtaining the dynamic heat and humidity load demand curve of the operating room in the future period;

[0156] The personnel status data stream includes the number and body surface area of ​​medical staff; the body surface area of ​​the medical staff is calculated based on the height and weight of the medical staff.

[0157] Specifically, real-time environmental parameters such as temperature, humidity, and pressure difference in each area of ​​the operating room are collected through distributed sensors to form an environmental status data stream; these parameters directly reflect the thermal and humid characteristics of the surgical environment and are important inputs for thermal and humid load prediction. To ensure the timeliness of the prediction, sensors need to use low-power, self-organizing Internet of Things technology to achieve millisecond-level data sampling and feedback. At the same time, the data gateway needs to perform pre-processing such as cleaning, calibration, and synchronization on multi-source heterogeneous sensor data to improve data quality and simplify subsequent analysis processes. As time series data, environmental parameters generally require time series feature extraction, including operations such as moving average, difference, and wavelet decomposition, to highlight the time-frequency domain characteristics contained therein and provide richer feature inputs for machine learning models.

[0158] The number of personnel and body surface area during the operation recorded by the medical staff management system are obtained to form a personnel status data stream; the personnel status data stream is designed to take into account the heat and humidity load generated by human metabolism. Based on the Dubois & Dubois formula in the ASHRAE55-2013 standard, the body surface area of ​​the medical staff is estimated by their height and weight information. During the operation, the dynamic changes in the number of medical staff will cause fluctuations in sensible heat and latent heat, and this part of the load is often ignored. By obtaining the number of medical staff registered in the medical staff management system, combined with human physiological parameters such as body surface area and metabolic rate, the heat production of the personnel can be roughly estimated. Taking the personnel factor as a supplement can improve the comprehensiveness of the heat and humidity load prediction.

[0159] The environmental status data stream and the personnel status data stream are input into the pre-trained heat and moisture load demand prediction model. Heat and moisture load is a physical quantity that describes the heat and moisture balance of the air conditioning system. Its dynamic changes usually have a certain time series correlation. Therefore, the long short-term memory network can effectively extract the spatiotemporal law of load changes. The model is trained and optimized in an offline manner, with a large amount of historical load data collected as samples, and the model parameters are continuously updated iteratively through mechanisms such as error back propagation, and finally a load prediction engine with excellent generalization ability and prediction accuracy is formed. In the real-time stage, the model takes the current environment and personnel status as initial input, and continuously rolls to predict the load change curve in the future period, providing constraint boundaries for subsequent air supply optimization. The use of machine learning algorithms can overcome the problem of insufficient accuracy of traditional physical modeling in complex environments, learn the inherent laws of the environment through massive data, reduce the errors caused by manual fitting, and improve the dynamic adaptability of load prediction. At the same time, compared with the pure data-driven method, the heat and moisture load demand prediction model integrates environmental perception data, which not only improves the timeliness of prediction, but also makes the prediction process more transparent and controllable.

[0160] Step S2100 fully considers the sensible heat and latent heat loads by integrating environmental monitoring data and personnel physiological parameters, improves the comprehensiveness of heat and moisture load estimation, and avoids the accumulation of errors in sub-item estimation. The machine learning algorithm is introduced to establish a heat and moisture load demand forecasting model. Through massive historical data training, the accuracy of load forecasting is improved, and the problem of insufficient adaptability of traditional empirical formulas in complex environments is overcome. The use of time series models such as long short-term memory networks can fully explore the spatiotemporal characteristics of load changes, output continuous dynamic load forecasting curves, and ensure the dynamic real-time nature of heat and moisture load boundary conditions.

[0161] Step S2200, generating a dynamic air supply control strategy based on the dynamic heat and humidity load demand curve of the operating room in the future period, the three-dimensional spatial topology model of the operating room, and the pollution source distribution heat map;

[0162] Furthermore, if Figure 8 As shown, step S2200 includes:

[0163] Step S2210, obtaining the layout information of the air supply equipment, and constructing a multi-source and multi-sink air supply path directed graph based on the three-dimensional spatial topological model of the operating room and the layout information of the air supply equipment;

[0164] Step S2220, based on the dynamic heat and humidity load demand curve of the operating room in the future period and the pollution source distribution thermodynamic map, solve the optimal air supply tree in the multi-source and multi-sink air supply path directed graph;

[0165] Step S2230, mapping the optimal air supply tree to a specific air supply equipment combination to form a dynamic air supply control strategy.

[0166] Specifically, in step S2210, it is first necessary to integrate the layout information of the air supply equipment on the basis of the three-dimensional space topological model of the operating room to form a directed graph of multi-source and multi-sink air supply paths that can reflect the air supply process. The so-called multi-source and multi-sink refers to the existence of multiple source points (such as air supply outlets) and multiple sinks (such as return air outlets) in the same network. By mapping the physical space inside the operating room into the mathematical space, a topological structure consisting of nodes and directed edges can be obtained. Among them, key components such as room boundaries, clean partitions, and obstacles correspond to nodes in the graph, representing the spatial constraints of the air supply process; the connectivity between different components and between components and air supply equipment corresponds to directed edges, representing the potential transmission channels of airflow. The attribute information of the node can be directly inherited from the three-dimensional space topological model, including environmental parameters such as position, volume, and pressure. The weight information of the directed edge is closely related to the flow characteristics of the airflow, reflecting the dynamic characteristic parameters such as resistance loss and flow distribution. By embedding the layout information of the air supply equipment, the graph theory expression of the entire process of airflow transportation is realized. The multi-source and multi-sink air supply path optimization problem can be further modeled as a network flow problem and solved using mature graph theory algorithms.

[0167] The beneficial effects of step S2210 are: constructing a unified mathematical model, mapping discrete and heterogeneous environmental information into a standardized graph space, establishing the relationship between environmental entities, air flow fields, and control objects through the topological relationship of nodes and edges, laying the model foundation for subsequent path optimization; at the same time, introducing a graph data structure to express the environmental topology, which is simpler and more efficient than the geometric model and facilitates algorithm implementation.

[0168] Step S2220 further considers the dynamic change characteristics of the thermal and humid environment of the operating room. The dynamic thermal and humid load demand curve of the operating room in the future period obtained by step S2100 depicts the fluctuation law of thermal and humid load caused by factors such as personnel, equipment, and environment during the operation. The dynamic change of load will directly affect the time-varying distribution of environmental parameters, and then put forward the demand for dynamic optimization of air supply control. The pollution source distribution heat map obtained in step S1300 reveals the dynamic evolution characteristics of particulate pollution from another perspective, providing a basis for targeted local decontamination. The dynamic thermal and humid load demand curve is embedded in the path optimization model as a time-varying constraint condition, and the pollution source distribution heat map is used as the weight distribution of the optimization target. The high-value area of ​​the heat map is taken as the center, and the air supply volume is appropriately increased in its neighborhood. In this way, the fixed-point removal of pollutants can be achieved as efficiently as possible while meeting the dynamic supply and demand balance. Solving this time-varying, multi-objective optimization model can obtain the optimal air supply tree that is dynamically updated over time. Compared with the static centralized air supply solution, the optimal air supply tree can adapt to the dynamic changes of load, achieve precise temporal and spatial matching of air volume, and ensure air quality while improving comfort.

[0169] Here we further explain the concept of the optimal air supply tree. In the graph theory model of the multi-source and multi-sink air supply path optimization problem, the source point is the air supply outlet, corresponding to the fresh air unit; the sink point is the return air outlet, corresponding to the exhaust equipment. With the source point as the root node and the sink point as the leaf node, a breadth-first spanning tree covering all nodes and edges is formed, which is the "air supply tree". Each air supply tree uniquely corresponds to one air supply path combination scheme. By iteratively optimizing the topological structure and node weight distribution of the air supply tree, the optimal air supply path under specific constraints can be obtained. This optimal solution not only takes into account the effective delivery of fresh air at the macro level, but also realizes the precise control of local wind speed and wind direction at the micro level, and has significant improvements in airflow organization effect, ventilation efficiency and other aspects.

[0170] Step S2230 further maps the optimal solution in the mathematical space back to the control instructions of the physical space. Since the optimal air supply tree is essentially a logical topological structure, it needs to be further converted into a specific hardware device combination. This process can be achieved through a table lookup method. A mapping table of logical topology and hardware configuration is generated offline in advance for the model parameters and installation locations of various types of air supply equipment. A typical mapping item is as follows: (equipment number, model, installation location, rated air volume, diversion direction, ...). During online matching, the attribute information of the nodes and edges of the optimal air supply tree is matched with the key values ​​of the mapping table, and the corresponding hardware configuration parameters can be quickly retrieved to generate a device-level dynamic control sequence, such as (fresh air unit 1, fan frequency = 50Hz, air guide door opening = 60%, operating time = 15min, ...). Since the mapping relationship is generated offline in advance, it basically does not introduce additional online computing overhead, which can meet the needs of real-time control. By sending the control sequence to various actuators on site, the fresh air unit, electric air valve and other equipment can be manipulated for dynamic coordination, and the air supply volume, air supply direction, and operating time can be adjusted in real time, ultimately forming a three-dimensional and precise dynamic air supply control strategy.

[0171] In summary, the innovations of step S2200 are mainly reflected in the following three aspects: first, the graph data structure and network flow model are introduced to build a unified topology optimization framework; second, the dynamics of heat and humidity load and the thermal map of pollution distribution are introduced into the optimization target to achieve precise air supply; third, through topology-hardware mapping, the engineering implementation problem of the model solution is solved. This strategy can fully explore the spatiotemporal characteristics of environmental data, achieve precise control based on multi-source information fusion, and thus comprehensively improve the comfort and cleanliness of the surgical environment.

[0172] Step S2300, transmitting the dynamic air supply control strategy to the laminar air supply unit to perform intelligent air supply control.

[0173] Specifically, first, the generated dynamic air supply control strategy needs to be converted into specific air supply unit operation instructions. The control strategy here is usually a set of abstract, high-level air supply adjustment schemes, such as "increase the air supply volume in the sterile area by 20%, and reduce the air supply volume in the contaminated area by 30%", etc., which cannot be directly used to guide the actual operation of the air supply unit. Therefore, it is necessary to design a strategy interpreter to interpret the abstract strategy into the underlying control instructions that can be directly executed by the air supply unit, such as "set the wind speed of the No. 1 air supply outlet to 0.8m / s, and the wind speed of the No. 3 air supply outlet to 0.5m / s". The process of strategy interpretation needs to comprehensively consider factors such as the installation layout and rated parameters of the air supply unit, and design predefined interpretation rules for different strategy modes. Through strategy interpretation, a semantic mapping is established between the control strategy and the control instruction, so that the high-level regulation logic and the underlying execution mechanism can work together.

[0174] After completing the strategy interpretation, the control instructions need to be further issued to each laminar flow air supply unit. The air supply system in a modern operating room is usually a multi-component, heterogeneous complex system. The air supply units equipped in different areas differ in specifications, models, manufacturers, etc. Their communication protocols, control interfaces, etc. are not the same, and it is difficult to achieve unified scheduling. Therefore, it is necessary to introduce a device adaptation layer to standardize the device interfaces of different manufacturers, provide a unified device driver interface, and shield device differences. Based on a unified device driver interface, consistent communication protocols and data formats can be used to achieve transparent access and control of heterogeneous air supply units. Common device communication protocols can be Modbus, BACnet, etc., which achieve cross-platform and cross-manufacturer interconnection by defining standardized data frame structures and interaction processes.

[0175] After the equipment access and protocol adaptation are completed, the control instructions can be reliably and in real time transmitted to each laminar air supply unit. After receiving the instruction, the air supply unit immediately adjusts its own air supply parameters according to the instruction content, such as air supply volume, air supply speed, air supply direction, temperature, humidity, etc. Modern laminar air supply units are usually equipped with intelligent servo motors and precision sensors, which can achieve high-precision and fast-response air supply parameter adjustment. Its basic working principle is: the frequency conversion control system adjusts the motor speed in real time according to the air supply instruction, drives the fan impeller to rotate at high speed, and dynamically adjusts the guide plate elevation angle, and finally realizes the precise control of air supply volume and air supply flow field. This process can be completed within the millisecond time scale, meeting the needs of rapid dynamic adjustment of surgical environment conditions. During the adjustment process, the multi-parameter sensor of the unit collects parameters such as wind speed, air volume, wind direction, temperature and humidity in real time, and transmits them back to the monitoring system to realize real-time parameter monitoring and closed-loop control. This organic combination of intelligent hardware and refined control mode is an important basis for realizing precise flow field regulation in medical clean space.

[0176] While the air supply unit is executing the adjustment instructions, it is also necessary to continuously collect the environmental parameters in the operating room and dynamically evaluate the air supply effect. By deploying an environmental sensor network in various areas of the operating room, environmental quality parameters such as temperature field and concentration field can be obtained in real time. By comparing the measured parameters with the set values ​​of the air supply control strategy, the current air supply effect can be judged, and air supply fine-tuning instructions can be generated accordingly to further optimize the air supply control accuracy. Through the introduction of sensor feedback and dynamic optimization mechanisms, the air supply unit can adaptively adjust the control strategy according to the actual effect and continuously approach the optimal air supply control mode. This dynamic closed-loop control paradigm of equipment control, environmental perception, and strategy optimization is an effective means to ensure a constant and controllable clean environment in the operating room.

[0177] Through intelligent air supply, the air supply volume in different areas can be adjusted in a targeted manner, local pollution can be accurately traced and removed, and the dynamic stability of the surgical environment can be maintained. The implementation of this method is conducive to improving the environmental quality of the operating room, reducing the risk of infection during surgery, improving clinical effects, and ensuring the safety of doctors and patients.

[0178] Example 2

[0179] This embodiment provides an intelligent control system for a laminar air supply unit dedicated to an operating room on the basis of the first embodiment. Fig. 9 As shown, including:

[0180] Heat map generation module: used to build a three-dimensional spatial topological model of the operating room. Based on the three-dimensional spatial topological model of the operating room, the spatial position data of medical staff and equipment during the operation are collected in real time to form a spatiotemporal trajectory data set of medical staff and equipment; the spatiotemporal trajectory data set is analyzed to generate a heat map of the pollution source distribution;

[0181] Load prediction module: used to obtain the environmental status data stream and personnel status data stream of the operating room, input the environmental status data stream and the personnel status data stream into the pre-trained heat and humidity load demand prediction model, and obtain the dynamic heat and humidity load demand curve of the operating room in the future period;

[0182] Air supply strategy generation module: Generates dynamic air supply control strategy based on the dynamic heat and humidity load demand curve of the operating room in the future period, the three-dimensional spatial topology model of the operating room and the thermal map of pollution source distribution;

[0183] Air supply control module: used to transmit the dynamic air supply control strategy to the laminar air supply unit for intelligent air supply control.

[0184] In the heat map generation module, the construction of the three-dimensional spatial topological model of the operating room includes:

[0185] Step S1110, performing an all-round scan of the interior space of the operating room to obtain high-precision point cloud data;

[0186] Step S1120, extracting geometric information of key components inside the operating room from the high-precision point cloud data, and reconstructing a three-dimensional solid model of the operating room;

[0187] Step S1130, calibrating the clean partition boundary in the three-dimensional solid model of the operating room and generating clean partition parameter information;

[0188] Step S1140: The three-dimensional entity model of the operating room is integrated with the parameter information of the clean partition to construct a three-dimensional spatial topological model of the operating room.

[0189] In the heat map generation module, the analysis of the spatiotemporal trajectory data set to generate a pollution source distribution heat map includes:

[0190] Step S1310, extracting the activity trajectories of medical staff and equipment from the spatiotemporal trajectory dataset;

[0191] Step S1320, identifying a dwell event from the activity track;

[0192] Step S1330, constructing a pollution source intensity model, and based on the pollution source intensity model, calculating the pollution source intensity of each resident event, and generating a resident event data set;

[0193] Step S1340: Generate a pollution source distribution heat map based on the resident event data set.

[0194] The step S1320 includes:

[0195] Step S1321, discretizing the activity trajectory into a series of spatiotemporal coordinate points to form a trajectory point sequence;

[0196] Step S1322, taking the spatial distribution density of the trajectory point sequence as a criterion, applying a density-based spatial clustering algorithm to extract candidate residence points;

[0197] Step S1323: construct a spatiotemporal neighborhood of the candidate stay point, and generate a stay event based on the candidate stay point and the spatiotemporal neighborhood of the candidate stay point.

[0198] The step S1323 includes:

[0199] Step S13231, defining medical staff and equipment as resident subjects, and defining any two candidate resident points of any resident subject as resident segments;

[0200] Step S13232: if two candidate residence points in any residence segment of the resident subject are adjacent in time and space, then the residence segment is a continuous residence of the resident subject;

[0201] Step S13233: merge all the consecutive stays of the stay subject into a stay event. Each stay event is represented by a four-tuple (s, t0, Δt, l ) is used to represent the residence; where s is the residence subject, t0 is the residence start time, Δt is the residence duration, l are the resident location coordinates; the resident subjects include medical personnel and equipment.

[0202] The step S1330 includes:

[0203] Step S1331, define the pollution source intensity index, and build a pollution source intensity model based on the pollution source intensity index; the pollution source intensity model includes a personnel pollution source intensity model P A and equipment pollution source intensity model P B ;

[0204] Step S1332: Based on the pollution source intensity model, the pollution source intensity of each resident event is calculated to generate a resident event data set carrying pollution attribute labels.

[0205] The step S1340 includes:

[0206] Step S1341, performing spatiotemporal aggregation on the resident event data set to form a spatiotemporal cube representation of the pollution source distribution; the pollution source distribution represents the distribution state of the resident events in the three-dimensional spatial topological model of the operating room, which is jointly determined by the resident position coordinates of each resident event and the pollution source intensity;

[0207] Step S1342, slicing the space-time cube in the spatial dimension, discretizing the continuous spatial coordinates of the space-time cube into three-dimensional spatial grids, mapping the pollution source distribution into three-dimensional spatial grid data, and the attribute value of each three-dimensional spatial grid represents the intensity of the pollution source;

[0208] Step S1343, smoothing the three-dimensional space grid data using the pollution source intensity as a weight to generate a pollution source distribution heat map.

[0209] In the air supply strategy generation module, the generation of a dynamic air supply control strategy includes:

[0210] Step S2210, obtaining the layout information of the air supply equipment, and constructing a multi-source and multi-sink air supply path directed graph based on the three-dimensional spatial topological model of the operating room and the layout information of the air supply equipment;

[0211] Step S2220, based on the dynamic heat and humidity load demand curve of the operating room in the future period and the pollution source distribution thermodynamic map, solve the optimal air supply tree in the multi-source and multi-sink air supply path directed graph;

[0212] Step S2230, mapping the optimal air supply tree to a specific air supply equipment combination to form a dynamic air supply control strategy.

[0213] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0214] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent control method for a laminar air supply unit for an operating room, characterized in that: The method comprises: Construct a three-dimensional spatial topological model of the operating room. Based on the three-dimensional spatial topological model of the operating room, collect the spatial position data of medical staff and equipment during the operation in real time to form a spatiotemporal trajectory data set of medical staff and equipment; analyze the spatiotemporal trajectory data set to generate a heat map of pollution source distribution; Obtain the environmental status data stream and personnel status data stream of the operating room, input the environmental status data stream and the personnel status data stream into the pre-trained heat and humidity load demand prediction model, and obtain the dynamic heat and humidity load demand curve of the operating room in the future period; generate a dynamic air supply control strategy based on the dynamic heat and humidity load demand curve of the operating room in the future period, the three-dimensional spatial topological model of the operating room, and the pollution source distribution thermal map; transmit the dynamic air supply control strategy to the laminar air supply unit to perform intelligent air supply control; The constructing of the three-dimensional spatial topological model of the operating room comprises: Perform a full-scale scan of the interior space of the operating room to obtain high-precision point cloud data; Extract geometric information of key components inside the operating room from high-precision point cloud data and reconstruct the 3D solid model of the operating room; Demarcate the clean partition boundary in the 3D solid model of the operating room and generate clean partition parameter information; The 3D entity model of the operating room is integrated with the parameter information of the clean partition to construct a 3D spatial topological model of the operating room; The analyzing of the spatiotemporal trajectory data set to generate a pollution source distribution heat map includes: The activity trajectories of medical staff and equipment are extracted from the spatiotemporal trajectory dataset, and residence events are identified from the activity trajectories; a pollution source intensity model is constructed, and based on the pollution source intensity model, the pollution source intensity of each residence event is calculated to generate a residence event dataset; based on the residence event dataset, a pollution source distribution heat map is generated.

2. The intelligent control method for the laminar air supply unit for operating room according to claim 1 is characterized in that: The identifying of the dwelling event from the activity track comprises: Discretize the activity trajectory into a series of space-time coordinate points to form a trajectory point sequence; Taking the spatial distribution density of trajectory point sequence as the criterion, the density-based spatial clustering algorithm is applied to extract candidate residence points; A spatiotemporal neighborhood of the candidate dwelling point is constructed, and a dwelling event is generated based on the candidate dwelling point and the spatiotemporal neighborhood of the candidate dwelling point.

3. The intelligent control method for the laminar air supply unit for operating room according to claim 2 is characterized in that: The construction of the spatiotemporal neighborhood of the candidate residence point includes: For the candidate residence point p0, a neighborhood N(p0) with a time span of Δt1 and a space span of d is constructed with p0 as the center; for any candidate residence point p1, if p1∈N(p0), p0 is determined to be adjacent to p1 in time and space.

4. The intelligent control method for the laminar air supply unit for operating room according to claim 2 is characterized in that: The generating of the resident event comprises: Medical staff and equipment are defined as resident subjects; any two candidate resident points of any resident subject are defined as resident segments; If two candidate dwelling points in any dwelling segment of a dwelling subject are adjacent in time and space, then the dwelling segment is a continuous dwelling of the dwelling subject; All consecutive stays of the stay subject are merged into a stay event, and each stay event is represented by a four-tuple (s, t0, Δt, l ) is used to represent the residence; where s is the residence subject, t0 is the residence start time, Δt is the residence duration, l are the resident location coordinates; the resident subjects include medical personnel and equipment.

5. The intelligent control method for the laminar air supply unit for operating room according to claim 4 is characterized in that: The pollution source intensity model includes a human pollution source intensity model P A and equipment pollution source intensity model P B ; If the main residents of the residence event are medical staff, the personnel pollution source intensity model P is used. A Calculate the pollution source intensity of the residence event; if the residence subject of the residence event is equipment, the equipment pollution source intensity model P is used B Calculate the pollution source intensity of the dwelling event.

6. The intelligent control method for the laminar air supply unit for operating room according to claim 5, characterized in that: Generating a pollution source distribution heat map according to the resident event data set includes: Performing spatiotemporal aggregation on the resident event data set to form a spatiotemporal cube representation of the pollution source distribution; the pollution source distribution represents the distribution state of the resident events in the three-dimensional spatial topological model of the operating room, which is jointly determined by the resident position coordinates of each resident event and the pollution source intensity; Slice the space-time cube in the spatial dimension, discretize the continuous spatial coordinates of the space-time cube into three-dimensional spatial grids, map the pollution source distribution into three-dimensional spatial grid data, and the attribute value of each three-dimensional spatial grid represents the intensity of the pollution source; The three-dimensional spatial raster data is smoothed using the pollution source intensity as the weight to generate a heat map of the pollution source distribution.

7. The intelligent control method for the laminar air supply unit for operating room according to claim 1, characterized in that: The personnel status data stream includes the number and body surface area of ​​medical personnel; the body surface area of ​​the medical personnel is calculated based on the height and weight of the medical personnel; The generating of the dynamic air supply control strategy comprises: Obtain the layout information of the air supply equipment, and construct a multi-source and multi-sink air supply path directed graph based on the three-dimensional spatial topological model of the operating room and the layout information of the air supply equipment; Based on the dynamic heat and humidity load demand curve of the operating room in the future period and the thermal map of pollution source distribution, the optimal air supply tree is solved in the multi-source and multi-sink air supply path directed graph; The optimal air supply tree is mapped to a specific combination of air supply equipment to form a dynamic air supply control strategy.

8. An intelligent control system for a laminar air supply unit for an operating room, which is used to implement the intelligent control method for a laminar air supply unit for an operating room according to any one of claims 1 to 7, characterized in that: The system comprises: Heat map generation module: used to build a three-dimensional spatial topological model of the operating room. Based on the three-dimensional spatial topological model of the operating room, the spatial position data of medical staff and equipment during the operation are collected in real time to form a spatiotemporal trajectory data set of medical staff and equipment; the spatiotemporal trajectory data set is analyzed to generate a heat map of the pollution source distribution; Load prediction module: used to obtain the environmental status data stream and personnel status data stream of the operating room, input the environmental status data stream and the personnel status data stream into the pre-trained heat and humidity load demand prediction model, and obtain the dynamic heat and humidity load demand curve of the operating room in the future period; Air supply strategy generation module: Generates dynamic air supply control strategy based on the dynamic heat and humidity load demand curve of the operating room in the future period, the three-dimensional spatial topology model of the operating room and the thermal map of pollution source distribution; Air supply control module: used to transmit the dynamic air supply control strategy to the laminar air supply unit for intelligent air supply control.

Images