An AI-based automated cleaning control system for livestock and veterinary operating tables
By dynamically identifying and optimizing contaminated areas on livestock and veterinary operating tables through an AI modular system, generating cleaning paths and executing automated cleaning, the system solves the problem of insufficient targeting of traditional cleaning systems and achieves efficient and safe cleaning results.
Patent Information
- Application Number
- CN202611094633.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-22
- Publication Date
- 2026-11-17
AI Technical Summary
Traditional livestock and veterinary operating table cleaning systems cannot dynamically identify contaminated areas based on animal species, type of surgery, etc., resulting in non-targeted cleaning, which may lead to waste of resources or omission of contaminated points, and pose risks of equipment damage or over-cleaning.
The system employs an AI-based pollution event prediction module, region construction module, scoring calculation module, cleanup path construction module, and cleanup strategy generation module. By combining the functional information of the polluted area, it generates a cleanup priority list and executes automated cleanup operations. Compensation cleanup is performed through time window segmentation management and image recognition.
It achieves targeted and resource-saving cleaning control, improves cleaning efficiency and safety, avoids ineffective cleaning and equipment damage, and ensures high-standard automatic cleaning results.
Smart Images

Figure CN123120224A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of automatic cleaning control technology, and in particular to an automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence. Background Technology
[0002] With the continuous improvement of animal husbandry and veterinary medical standards, veterinary surgical procedures are becoming increasingly refined and frequent. As the core equipment in the surgical process, the quality of cleaning and disinfection of the operating table directly affects the risk of postoperative infection and the control of disease transmission in animals.
[0003] In related technologies, traditional cleaning systems cannot dynamically identify contaminated areas based on contextual information such as different animal species and surgical procedures, resulting in non-targeted cleaning actions, resource waste, or missed contaminated points. Furthermore, traditional cleaning equipment typically uses fixed cleaning agents and operating intensities, ignoring the differences in materials and cleaning requirements across different areas, posing risks of equipment damage or over-cleaning, thus requiring improvement. Summary of the Invention
[0004] The purpose of this invention is to provide an automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence, so as to solve the problems mentioned in the background art.
[0005] This application provides an automatic cleaning and control system for livestock and veterinary operating tables based on artificial intelligence, which adopts the following technical solution:
[0006] The pollution event prediction module obtains information on the animal species, surgical type, and related operations of the operating table to be cleaned, inputs it into the pollution identification model, and obtains the predicted pollution event information.
[0007] The contaminated area construction module generates predicted contaminated area information based on the predicted contamination event information and the equipment space structure information of the operating table.
[0008] The pollution score calculation module calculates the pollution impact score for each predicted pollution area based on the predicted pollution area information, and obtains a comprehensive pollution area score.
[0009] The cleanup path construction module generates a pollution cleanup priority list based on the comprehensive score of the polluted area, and constructs the cleanup path of the polluted area using a path optimization algorithm in combination with the physical location data of the polluted area.
[0010] The cleaning strategy generation module obtains the functional information of the contaminated area, generates cleaning operation information based on the contaminated cleaning path, and executes automated cleaning operations.
[0011] The compensation cleanup trigger module executes the cleanup task in stages according to a preset time window. After each stage of cleanup is completed, it obtains the image information of the current area and determines whether to trigger the compensation cleanup operation.
[0012] Preferably, the contamination event prediction module acquires the animal species information, surgical type information, and related operational information corresponding to the operating table to be cleaned, inputs them into the contamination identification model, and obtains the predicted contamination event information. Specifically, the steps are as follows:
[0013] Obtain the animal species information corresponding to the operating table to be cleaned, wherein the animal species information includes animal species information and body size classification information;
[0014] Obtain the surgical type information corresponding to the operating table to be cleaned, including surgical type information and standard surgical procedure;
[0015] Obtain relevant operation information corresponding to the operating table to be cleaned, including the type of operating instrument, frequency of operating actions, and operation execution time;
[0016] The animal species information, surgery type information, and related operation information are subjected to feature encoding processing to generate structured feature vectors;
[0017] The structured feature vector is input into the contamination identification model, and trained based on historical surgical contamination samples to obtain predicted contamination event information.
[0018] Preferably, the contaminated area construction module, based on the predicted contamination event information and combined with the equipment space structure information of the operating table, generates predicted contamination area information in the following steps:
[0019] Based on the predicted pollution event information, the predicted pollution event information includes the spatial coordinates of the directly polluted area, the pollution type, and the pollution level;
[0020] Acquire the equipment spatial structure information of the operating table, which includes the operation path trajectory of instruments in each functional area of the operating table during the operation.
[0021] Based on the operation path trajectory, calculate the distance and proximity between the instrument contact point and each spatial area of the operating table, and establish the spatial proximity relationship between the operating table spatial area and the instrument operation path.
[0022] Based on the spatial proximity relationship, the spatial location coordinates, pollution type and pollution level of the directly polluted area are extracted, and the probability of pollutants spreading from the directly polluted area to the adjacent area is calculated to obtain the pollution spread probability.
[0023] Based on the pollution propagation probability, a pollution propagation probability map is constructed to generate information on predicted polluted areas.
[0024] Preferably, the step of constructing a pollution propagation probability map and generating predicted pollution area information based on the pollution propagation probability specifically includes:
[0025] Based on the pollution propagation probability, the operating table space area is used as a node, and the pollution propagation probability between areas is used as an edge to construct a pollution propagation probability graph.
[0026] Based on the pollution propagation probability map, and according to a preset probability threshold, a graph traversal algorithm is used to identify potential pollution propagation areas where the pollution propagation probability exceeds the preset probability threshold.
[0027] The directly polluted areas are merged with the potentially spreading polluted areas to form a complete polluted area combination, generating predicted polluted area information.
[0028] Preferably, the pollution score calculation module, based on the predicted pollution area information, calculates a pollution impact score for each predicted pollution area to obtain a comprehensive pollution area score, specifically as follows:
[0029] Based on the predicted pollution area information, the pollution type, pollution area, and pollution residual duration of each predicted pollution area are collected.
[0030] Based on the pollution type, pollution area, and pollution residual duration, the corresponding pollution type risk coefficient, pollution area risk coefficient, and pollution residual risk coefficient are obtained respectively.
[0031] Based on the pollution type risk coefficient, pollution area risk coefficient, and pollution residue risk coefficient, a weighted fusion model is used to calculate the pollution impact score, resulting in a comprehensive score for the polluted area.
[0032] Preferably, the cleanup path construction module generates a pollution cleanup priority list based on the comprehensive score of the polluted area, and constructs a cleanup path for the polluted area using a path optimization algorithm, specifically as follows:
[0033] Based on the comprehensive score of the polluted area, a pollution cleanup priority list is generated by sorting the comprehensive score values from high to low.
[0034] Based on the physical location data of the polluted areas, the geometric distance between the polluted areas is calculated, and a spatial distance matrix between the polluted areas is constructed according to the geometric distance.
[0035] Based on the pollution cleanup priority list and spatial distance matrix, a path optimization algorithm is used to generate a cleanup path for the polluted area.
[0036] Preferably, the cleaning strategy generation module acquires functional information of the contaminated area, generates cleaning operation information based on the contaminated cleaning path, and executes automated cleaning operations, specifically as follows:
[0037] Obtain functional information of the contaminated area, which includes the operational function type of the contaminated area and its corresponding surface material type and cleaning level requirements;
[0038] Based on the comprehensive score of the contaminated area, the surface material type, and the cleaning level requirements, cleaning operation information for the contaminated area is generated. The cleaning operation information includes the type of cleaning agent, the dosage of cleaning agent, the sequence of cleaning actions, and the cleaning duration.
[0039] Based on the pollution cleanup path, the cleanup operation information is combined with the pollution cleanup path to form a cleanup control sequence;
[0040] Based on the cleaning control sequence, corresponding automated cleaning operations are performed on the contaminated area to complete the automatic cleaning process of the contaminated area.
[0041] Preferably, the compensation cleaning triggering module executes the cleaning task in stages according to a preset time window. After each stage of cleaning is completed, it acquires the image information of the current area and determines whether to trigger the compensation cleaning operation. The specific steps are as follows:
[0042] The task of cleaning up contaminated areas is divided into multiple stages according to a preset time window, with each stage corresponding to a part of the contaminated area in the contaminated cleanup path.
[0043] After each stage of the cleaning operation is completed, the image information of the current area is obtained, the image of the current area is preprocessed, and the standard image of the current area is obtained.
[0044] Pollution residue features are extracted from the standard image of the current area, including brightness anomalies and color deviations;
[0045] The presence area and density of pollution residue characteristics are extracted, and the presence area and density are calculated to obtain the pollution residue value.
[0046] Based on the pollution residue value, determine whether to trigger a compensation cleanup operation.
[0047] Preferably, the step of determining whether to trigger a compensation cleanup operation based on the pollution residue value specifically includes:
[0048] Set a pollution residue threshold and compare the pollution residue value with the pollution residue threshold;
[0049] If the pollution residue value exceeds the pollution residue threshold, a pollution cleanup failure mark will be generated;
[0050] Based on the pollution cleanup failure indicator, a compensation cleanup signal for the corresponding target area is triggered, and re-cleanup operation information is generated according to the pollution residue characteristics to perform compensation cleanup operation on the target area.
[0051] Once the maximum number of compensation attempts is reached, the current phase of cleanup is confirmed to be complete, and the process proceeds to the next phase of cleanup.
[0052] In summary, this application includes at least one of the following beneficial technical effects:
[0053] 1. Contamination patterns vary significantly depending on animal species, surgical procedures, and corresponding operational methods. By extracting this information as input features and feeding it into a trained contamination identification model, the type, scope, and severity of potential contamination events can be predicted in advance, reducing misjudgments or omissions caused by relying on fixed rules. By combining contamination events with spatial structure data, potentially contaminated locations and area boundaries are accurately marked, enabling spatially targeted cleaning paths and intensity strategies. This avoids ineffective cleaning or omissions, improving overall cleaning efficiency and energy conservation. Contamination impact scoring quantifies the severity of contamination in each area, typically integrating multiple factors such as contamination type, contamination area, contamination residual time, and regional functional sensitivity. This allows the system to differentiate between different cleaning areas, achieving rational allocation of cleaning resources, prioritizing high-risk areas, and improving overall hygiene and safety levels. After establishing a cleaning priority order through comprehensive contamination area scoring, the system further generates optimal cleaning routes based on the spatial location of each area. Path optimization algorithms are typically used, which can shorten total cleaning time, reduce equipment energy consumption, and improve system operational efficiency while ensuring priority cleaning of high-risk areas. Different areas have varying functional sensitivities and cleaning requirements. By acquiring the functional attributes and cleaning level requirements of each area, the system can formulate targeted cleaning strategies, which are then executed by automated cleaning equipment. This achieves high-standard automated cleaning while avoiding resource waste or equipment wear caused by over-cleaning. The entire cleaning process is segmented and managed using a time window mechanism, improving real-time execution and process controllability. After each stage, the system acquires images of the cleaned area using visual acquisition devices and detects the presence of residual contaminants using image recognition algorithms. When the detected residue exceeds a set threshold, the system immediately triggers a compensatory cleaning command, executing enhanced cleaning operations. This feedback control loop effectively improves cleaning reliability and accuracy, preventing the risk of cross-contamination due to incomplete cleaning.
[0054] 2. By abstracting each spatial region as a node in a graph and establishing connections between regions using the pollution propagation probability as edge weights, the dynamic diffusion path of pollutants from directly polluted areas to surrounding areas can be simulated more realistically. By setting a threshold for the pollution propagation probability and using a graph traversal algorithm to analyze the nodes in the graph, regions with edge weights exceeding the threshold, i.e., those with a high probability of being polluted, can be efficiently identified. This improves the accuracy of pollution prediction and avoids the erroneous cleaning of invalid regions, making the system's cleaning tasks more targeted and resource-efficient. The directly polluted regions identified by the model are merged with the potential polluted regions inferred based on the propagation probability to obtain a comprehensive and complete set of polluted regions. This set covers the actual sources of pollutants and their potential spread range, providing decision support for achieving high-precision, low-redundancy intelligent cleaning control.
[0055] 3. The cleaning task is divided into multiple time window units to achieve phased management and dynamic monitoring of the task, improving the system's real-time performance and responsiveness during task execution. Segmented execution by time window not only facilitates resource allocation but also introduces an evaluation mechanism for each stage, improving cleaning accuracy and error correction capabilities. It also avoids problems such as heat accumulation and equipment fatigue caused by continuous long-term operation, making the system more stable and task scheduling more flexible. After cleaning, images of the cleaned area are acquired, and preprocessing algorithms such as image enhancement, edge preservation, and contrast normalization are applied to ensure that the images used for subsequent contamination residue identification have high clarity and feature recognition. Abnormal brightness reflects changes in reflectivity, while color deviation can identify visual color differences left by contaminants after cleaning. By setting a baseline image and comparing the pixel distribution patterns in the current image, potentially uncleaned contamination areas can be effectively located, providing an efficient, non-contact, and structured method for detecting cleaning residues. The severity of contamination is comprehensively assessed by combining the area (proportion of contamination in the image) and density (concentration of contaminated pixels per unit area) of residual features. This residual contamination value is a key indicator for determining whether compensation cleaning is necessary, ensuring that the system cleaning results meet the standard specifications, avoiding cross-contamination or equipment damage due to insufficient cleaning in one go, and also reflecting the system's adaptive and autonomous error correction capabilities in handling complex environments. Attached Figure Description
[0056] Figure 1 This is a schematic diagram of the module connections of an embodiment of an automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to the present invention.
[0057] Figure 2 This is a flowchart illustrating the specific steps of an embodiment of an artificial intelligence-based automatic cleaning control system for livestock and veterinary operating tables according to the present invention. Detailed Implementation
[0058] The following examples and... Figures 1-2 The present invention will be described in further detail, but the embodiments of the present invention are not limited thereto.
[0059] This invention discloses an automatic cleaning and control system for livestock and veterinary operating tables based on artificial intelligence, specifically including the following steps:
[0060] The pollution event prediction module 1 obtains information on the animal species, surgical type, and related operation information corresponding to the operating table to be cleaned, inputs it into the pollution identification model, and obtains the predicted pollution event information.
[0061] The contaminated area construction module 2 is signal-connected to the contaminated event prediction module 1. Based on the predicted contaminated event information and combined with the equipment space structure information of the operating table, it generates predicted contaminated area information.
[0062] The pollution score calculation module 3 is signal-connected to the pollution area construction module 2. Based on the predicted pollution area information, it calculates the pollution impact score for each predicted pollution area to obtain a comprehensive pollution area score.
[0063] The cleaning path construction module 4 is signal-connected to the pollution score calculation module 3. Based on the comprehensive score of the polluted area, it generates a pollution cleaning priority list and, combined with the physical location data of the polluted area, uses a path optimization algorithm to construct the cleaning path of the polluted area.
[0064] The cleaning strategy generation module 5 is signal-connected to the cleaning path construction module 4, obtains the functional information of the polluted area, generates cleaning operation information based on the pollution cleaning path, and executes automated cleaning operations.
[0065] The compensation cleaning trigger module 6 is connected to the cleaning strategy generation module 5 by signal. It executes the cleaning task in stages according to the preset time window. After the cleaning operation of each stage is completed, it obtains the image information of the current area and determines whether to trigger the compensation cleaning operation.
[0066] In practical applications, the contamination patterns resulting from different animal species (such as cattle, sheep, and pigs), surgical types (such as cesarean sections, limb surgeries, and open-chest surgeries), and their corresponding operational methods (such as cutting, suturing, and rinsing) exhibit significant differences. By extracting this information as input features and feeding it into a trained contamination identification model, the type, scope, and severity of potential contamination events can be predicted in advance, reducing misjudgments or omissions caused by relying on fixed rules. By combining contamination events with spatial structure data, the locations and boundaries of potentially contaminated areas can be accurately marked, making cleaning paths and cleaning intensity strategies spatially targeted, avoiding ineffective cleaning or omissions, and improving overall cleaning efficiency and energy saving. The contamination impact score is a quantitative evaluation of the severity of contamination in each contaminated area, typically integrating multiple factors such as contamination type, contaminated area, contamination residual time, and regional functional sensitivity. This allows the system to treat different cleaning areas differently, thereby achieving reasonable allocation of cleaning resources (cleaning agents, operation time, etc.), ensuring priority treatment of high-risk areas, and improving the overall hygiene and safety level. After establishing a cleaning priority order through comprehensive scoring of contaminated areas, the system further generates the optimal cleaning route by combining the spatial location of each area. Typically employing path optimization algorithms, this effectively reduces the movement distance and switching frequency of cleaning robotic arms or spraying devices while ensuring priority cleaning of high-risk areas. This shortens the total cleaning time, reduces equipment energy consumption, and improves system operational efficiency. Different areas have different functional sensitivities and cleaning requirements. For example, surgical operating areas require complete sterility, while edge areas can accept routine cleaning. By acquiring the functional attributes and cleaning level requirements of each area, the system can dynamically match the corresponding cleaning agent type and dosage, action sequence, and time, forming a targeted cleaning strategy. This is ultimately executed by automated cleaning equipment, achieving high-standard automated cleaning while avoiding over-cleaning that wastes resources or causes equipment wear. The entire cleaning process is segmented and managed through a time window mechanism, improving the real-time performance and process controllability. After each stage is completed, the system acquires images of the cleaned area through visual acquisition devices and uses image recognition algorithms to detect the presence of residual contaminants. When the detected residue exceeds the set threshold, the system can immediately trigger a compensation cleaning command to perform enhanced cleaning operations, forming a feedback control loop that effectively improves cleaning reliability and accuracy and prevents the risk of cross-infection due to incomplete cleaning.
[0067] The steps for obtaining animal species information, surgical type information, and relevant operational information corresponding to the operating table to be cleaned, inputting them into the contamination identification model, and obtaining predicted contamination event information are as follows:
[0068] Step S11: Obtain the animal species information corresponding to the operating table to be cleaned. The animal species information includes animal species information and body size classification information.
[0069] Step S12: Obtain the surgical type information corresponding to the operating table to be cleaned. The surgical type information includes surgical type information and standard surgical procedure.
[0070] Step S13: Obtain relevant operation information corresponding to the operating table to be cleaned. The relevant operation information includes the type of operating instrument, the frequency of operating actions, and the operation execution time.
[0071] Step S14: Perform feature encoding processing on the animal species information, surgery type information, and related operation information to generate a structured feature vector;
[0072] Step S15: Input the structured feature vector into the contamination identification model, train it based on historical surgical contamination samples, and obtain predicted contamination event information.
[0073] In practical applications, differences in physiological structure, biochemical characteristics, and intraoperative blood loss among different animal species often significantly influence the location, extent, and type of contaminants. Simultaneously, body size classification (e.g., small, medium, large) determines the actual space occupied and exposed area on the operating table, thus affecting the potential contamination area. Accurately extracting this information helps the model adjust regional risk factors during prediction, achieving more precise contamination area prediction. The type of surgery (e.g., open surgery, limb correction, cesarean section) determines the type of contamination (blood, body fluids, tissue fragments, etc.); standard surgical procedures reflect the rhythm and scope of the operation, helping to infer the temporal sequence and spatial path of contamination diffusion. Relevant operational information directly relates to the physical contact path and intensity of the operation during surgery, and is a key triggering factor for contamination formation and distribution. For example, different instruments can cause different contamination trajectories, while high-frequency repetitive operations may expand the contaminated area or create cross-contamination; the frequency and duration of the operation also affect the difficulty of cleaning, such as the greater difficulty in cleaning clotted blood. Therefore, obtaining these parameters helps identify high-risk contamination trigger points and improves the behavioral perception capability of the contamination prediction model, making contamination estimation closer to the actual operation process. Preprocessing techniques such as standardization, vector embedding, and normalization form feature vectors with uniform dimensions and stable distribution, which are then input into the contamination identification model, improving the model's generalization ability and processing efficiency. By inputting the structured feature vectors into the contamination identification model, which has been trained based on historical surgical contamination samples, the model intelligently identifies potential contamination events in surgical scenarios and outputs predicted contamination event information. This ensures that subsequent cleaning actions are proactive and targeted, significantly improving the automation level and overall efficiency of the cleaning system.
[0074] The steps for generating predicted contamination area information based on the predicted contamination event information and the equipment space structure information of the operating table are as follows:
[0075] Step S21: Based on the predicted pollution event information, the predicted pollution event information includes the spatial coordinates of the directly polluted area, the pollution type, and the pollution level;
[0076] Step S22: Obtain the equipment space structure information of the operating table, which includes the operation path trajectory of instruments in each functional area of the operating table during the operation.
[0077] Step S23: Based on the operation path trajectory, calculate the distance and proximity between the instrument contact point and each spatial area of the operating table, and establish the spatial proximity relationship between the operating table spatial area and the instrument operation path.
[0078] Step S24: Based on the spatial proximity relationship, extract the spatial location coordinates, pollution type and pollution level of the directly polluted area, calculate the probability of pollutants spreading from the directly polluted area to the adjacent area, and obtain the pollution spread probability.
[0079] Step S25: Based on the pollution propagation probability, construct a pollution propagation probability map and generate predicted pollution area information.
[0080] In practical applications, by incorporating the spatial structure information of the operating table, pollution prediction is expanded from point to area. This equipment structure information includes not only the physical layout, such as the central operating area, auxiliary instrument area, and waste drainage area, but also the operational path trajectory data of the instruments. By reconstructing the movement trajectory of the instruments during surgery, the spatiotemporal behavioral characteristics of each instrument's contact or proximity area can be clarified, enhancing the model's ability to judge the direction and scope of pollution diffusion. By calculating the spatial distance and proximity between the surgical instrument's operational path and each area, areas where contact, splashing, or close interaction may occur during the operation can be identified. After constructing spatial proximity relationships, it is possible to quantitatively analyze whether directly contaminated areas pose a pollution diffusion risk to surrounding areas, improving the system's ability to perceive spatial correlations and enabling pollution assessment to move beyond static scope and possess dynamic propagation modeling capabilities. Different types of pollution (such as liquids and solids) have different propagation capabilities in adjacent spaces; the higher the pollution level, the greater the propagation probability. Combining spatial proximity and pollution characteristics, a rule engine or probabilistic propagation model can be used to calculate the possibility and intensity of pollution diffusion, obtaining the probability value of pollution propagation from directly contaminated areas to adjacent areas, achieving probabilistic prediction of risk distribution. By integrating point pollution with nearby diffusion information into a pollution propagation probability map, this map uses the operating table space area as nodes and the pollution propagation probability as edge weights to form a structured graph model. This model can identify high-risk pollution areas, intermediate propagation links, and potential hidden pollution areas, thereby providing a precise basis for decision-making in cleanup path planning, resource allocation, and cleaning strategy matching.
[0081] The steps for constructing a pollution propagation probability map and generating predicted pollution area information based on the pollution propagation probability are as follows:
[0082] Step S251: Based on the pollution propagation probability, construct a pollution propagation probability graph by taking the operating table space area as nodes and the pollution propagation probability between areas as edges.
[0083] Step S252: Based on the pollution propagation probability map, and according to a preset probability threshold, identify potential pollution propagation areas where the pollution propagation probability exceeds the preset probability threshold using a graph traversal algorithm.
[0084] Step S253: The directly polluted area and the potentially spreading polluted area are merged to form a complete polluted area combination, and the predicted polluted area information is generated.
[0085] In practical applications, each spatial region is abstracted as a node in a graph, and connections between regions are established using the pollution propagation probability as edge weights. This allows for a more realistic simulation of the dynamic diffusion path of pollutants from directly polluted areas to their surroundings. By setting a threshold for the pollution propagation probability and using a graph traversal algorithm to analyze the nodes in the graph, regions with edge weights exceeding the threshold (i.e., those with a high probability of contamination) can be efficiently identified. This improves the accuracy of pollution prediction and avoids the erroneous cleaning of invalid regions, making the system's cleaning tasks more targeted and resource-efficient. Merging the directly polluted regions identified by the model with the potential polluted regions inferred based on the propagation probability yields a comprehensive and complete set of polluted regions. This set covers the actual origin of pollutants and their potential spread range, providing decision support for achieving high-precision, low-redundancy intelligent cleaning control.
[0086] Based on the predicted pollution area information, the steps for calculating the pollution impact score for each predicted pollution area to obtain the comprehensive pollution area score are as follows:
[0087] Step S31: Based on the predicted pollution area information, collect the pollution type, pollution area, and pollution residual duration of each predicted pollution area;
[0088] Step S32: Based on the pollution type, pollution area, and pollution residual duration, obtain the corresponding pollution type risk coefficient, pollution area risk coefficient, and pollution residual risk coefficient, respectively.
[0089] Step S33: Based on the pollution type risk coefficient, pollution area risk coefficient, and pollution residue risk coefficient, a weighted fusion model is used to calculate the pollution impact score to obtain a comprehensive score for the polluted area.
[0090] In practical applications, multi-dimensional pollution characteristics are extracted for each predicted contaminated area. The type of contamination determines the direct threat posed by the contaminant to the surgical environment; for example, blood contamination poses a high risk of infection, and tissue fragments easily cause residue. The contaminated area reflects the size of the contamination range, affecting the resources and time required for cleaning. The duration of contamination residue reflects the longer the contaminant remains on the work surface, the stronger its adhesion and the greater the difficulty of cleaning. A risk coefficient lookup model, trained through training, maps the contamination type, area, and duration of residue to their relative risk intensity in terms of contamination impact. For example, blood contamination may correspond to a higher type risk coefficient, large-area contamination to a high-area risk coefficient, and long-term residue to a high-residue risk coefficient. Multiple risk factors are integrated into a unified contamination impact score using a weighted fusion model. This model determines the weight of each factor based on experience, expert knowledge, or historical training samples to reflect its relative contribution to actual cleaning risk and treatment priority. This not only comprehensively assesses the cleaning urgency of each contaminated area but also uses the scoring results for key decisions such as prioritization and resource allocation, constructing a cleaning priority ranking system among contaminated areas, thereby achieving intelligent and strategic cleaning control execution.
[0091] Based on the comprehensive score of the polluted area, a pollution cleanup priority list is generated. Combining this with the physical location data of the polluted area, a path optimization algorithm is used to construct a cleanup path for the polluted area. The specific steps are as follows:
[0092] Step S41: Based on the comprehensive score of the polluted area, sort the areas according to the comprehensive score values from high to low to generate a pollution cleanup priority list;
[0093] Step S42: Based on the physical location data of the polluted areas, calculate the geometric distance between the polluted areas, and construct a spatial distance matrix between the polluted areas according to the geometric distance.
[0094] Step S43: Based on the pollution cleanup priority list and spatial distance matrix, a path optimization algorithm is used to generate a cleanup path for the polluted area.
[0095] In practical applications, the comprehensive score reflects the severity of contamination and its impact on functional areas, directly determining the urgency of cleaning. By sorting areas by score from high to low, the system can identify high-risk contaminated areas that should be prioritized, avoiding cross-contamination, pathogen transmission, or delays in cleaning key areas due to improper cleaning sequence. Physical location data, such as coordinates, area center points, and boundary ranges, accurately reflects the actual distribution of contaminated areas on the operating table. By calculating the geometric distances between areas and constructing a spatial distance matrix, this matrix can significantly affect the total path length, number of transfers, and robotic arm operating efficiency in multi-contaminated area cleaning tasks, providing key support for reducing system operating costs and improving cleaning continuity. By integrating contamination severity (priority list) and spatial location distribution (distance matrix), a path optimization algorithm generates a contaminated area cleaning path that balances cleaning priority and spatial movement efficiency. This path guides cleaning equipment to complete the cleaning task with minimal actions and optimal sequence, saving energy and time, reducing equipment wear, improving cleaning accuracy, and enhancing the system's adaptive optimization capability for complex cleaning processes.
[0096] The steps for obtaining functional information about the contaminated area, generating cleaning operation information based on the contamination cleanup path, and executing automated cleanup operations are as follows:
[0097] Step S51: Obtain functional information of the contaminated area, which includes the operational function type of the contaminated area and its corresponding surface material type and cleaning level requirements;
[0098] Step S52: Based on the comprehensive score of the contaminated area, the surface material type, and the cleaning level requirements, generate cleaning operation information for the contaminated area. The cleaning operation information includes the type of cleaning agent, the dosage of cleaning agent, the sequence of cleaning actions, and the cleaning duration.
[0099] Step S53: Based on the pollution cleanup path, combine the cleanup operation information with the pollution cleanup path to form a cleanup control sequence;
[0100] Step S54: Based on the cleaning control sequence, perform corresponding automated cleaning operations on the contaminated area to complete the automated cleaning process of the contaminated area.
[0101] In practical applications, different areas of the operating table may have different functions, such as instrument handling areas, waste fluid collection areas, and biological sample placement areas. The functional type determines the potential risk level of biocontamination in that area. Furthermore, different surface materials, such as stainless steel, polyurethane, and glass, have varying compatibility with cleaning agents and adaptability to cleaning methods; excessive or improper handling may damage the surface. Therefore, identifying the functional and material types in advance and obtaining the corresponding cleaning level requirements helps to ensure safer and more effective subsequent cleaning actions, achieving regionalized and refined contamination treatment. By comprehensively considering the contamination score, the surface material's tolerance, and the cleaning level standard, the most suitable cleaning strategy for the current area's condition is generated. For example, high-contamination areas require high-concentration cleaning agents and vigorous rinsing; while easily corroded materials require mild cleaning agents and low-intensity treatment methods, maximizing cleaning efficiency and safety while ensuring stable equipment operation. The cleaning control sequence not only specifies the movement order of the cleaning equipment but also clarifies the specific operational parameters to be performed at each step in different areas, helping to improve the overall smoothness of the operation, avoid frequent switching of cleaning modes or repetitive paths, and improve the continuity and efficiency of system operation. By inputting the control sequence generated by the preceding module into the control unit of the cleaning robot or robotic arm, the system can complete the fine treatment of high-risk contaminated areas without human intervention, ensuring that the cleaning intensity meets the requirements of different functional areas, significantly improving work efficiency and reducing the risk of manual cleaning. It is particularly suitable for high-frequency surgical scenarios, easily infected environments and large-scale animal husbandry and veterinary institutions.
[0102] The cleaning task is executed in stages according to a preset time window. After each stage of cleaning is completed, the image information of the current area is obtained, and it is determined whether to trigger the compensation cleaning operation. Specifically:
[0103] Step S61: Divide the contaminated area cleanup task into multiple stages according to a preset time window, with each stage corresponding to a portion of the contaminated area in the cleanup path.
[0104] Step S62: After each stage of cleaning operation is completed, obtain the image information of the current area, preprocess the image of the current area, and obtain the standard image of the current area.
[0105] Step 63: Extract residual pollution features from the standard image of the current area, including brightness anomalies and color deviations.
[0106] Step S64: Extract the presence area and density of the pollution residue characteristics, and calculate the presence area and density to obtain the pollution residue value.
[0107] Step S65: Based on the pollution residue value, determine whether to trigger a compensation cleaning operation.
[0108] In practical applications, the cleaning task is divided into multiple time window units to achieve phased management and dynamic monitoring of the task. This improves the system's real-time performance and responsiveness during task execution. Segmented execution by time window not only facilitates resource allocation, such as controlling the amount of cleaning fluid and the robotic arm's running time, but also introduces an evaluation mechanism for each stage, improving cleaning accuracy and error correction capabilities. Simultaneously, it avoids problems such as heat accumulation and equipment fatigue caused by continuous long-term operation, making the system more stable and task scheduling more flexible. After cleaning, images of the cleaned area are acquired, and preprocessing algorithms such as image enhancement, edge preservation, and contrast normalization are applied to ensure that the images used for subsequent contamination residue identification have high clarity and feature recognition. Abnormal brightness reflects changes in reflectivity, such as moisture, grease, or blood residue. Color deviations can identify visual color differences left by contaminants after cleaning, such as abnormal color spots like light red or dark brown. By setting a baseline image and comparing the pixel distribution patterns in the current image, potentially uncleaned contamination areas can be effectively located, providing an efficient, non-contact, and structured method for detecting cleaning residues. The severity of contamination is comprehensively assessed by combining the area (proportion of contamination in the image) and density (concentration of contaminated pixels per unit area) of residual features. This residual contamination value is a key indicator for determining whether compensation cleaning is necessary, ensuring that the system cleaning results meet the standard specifications, avoiding cross-contamination or equipment damage due to insufficient cleaning in one go, and also reflecting the system's adaptive and autonomous error correction capabilities in handling complex environments.
[0109] The step of determining whether to trigger a compensation cleanup operation based on the aforementioned residual pollution value is as follows:
[0110] Step S651: Set a pollution residue threshold and compare the pollution residue value with the pollution residue threshold;
[0111] Step S652: If the pollution residue value exceeds the pollution residue threshold, a pollution cleanup failure mark is generated.
[0112] Step S653: Based on the pollution cleaning failure mark, trigger the corresponding compensation cleaning signal for the target area, and generate re-cleaning operation information according to the pollution residue characteristics, and perform compensation cleaning operation on the target area.
[0113] Step S654 continues until the maximum number of compensation attempts is reached. Once the current stage of cleanup is confirmed to be complete, proceed to the next stage of the cleanup task process.
[0114] In practical applications, by setting a pollution residue threshold, when the pollution residue level reaches or exceeds the threshold, a cleaning failure flag is automatically generated, marking the current cleaning stage as requiring treatment. This replaces traditional manual review and re-inspection methods, offering advantages such as strong real-time performance and automated decision-making. It can also serve as a scheduling signal source for subsequent cleaning compensation tasks, ensuring that cleaning tasks are not missed or misjudged. The failure flag triggers a second cleaning task for the designated area. Simultaneously, based on the pollution residue characteristics left after the initial cleaning (such as color difference type, residue boundary location, and stubborn stain morphology), the compensation cleaning parameters are optimized, such as increasing detergent concentration, extending cleaning time, or strengthening physical cleaning intensity. This gives the system the ability to adaptively handle complex cleaning scenarios such as corner contamination and localized heavy contamination, greatly improving the reliability of the final cleaning effect. Setting a maximum number of compensation attempts and automatically confirming cleaning completion after exceeding the limit helps ensure the controllability and efficiency of system execution. While ensuring cleaning effectiveness, it also balances the task rhythm and operational safety of the automated system, avoiding unlimited compensation cleaning in a highly contaminated area, which could lead to waste of cleaning resources or equipment wear risks.
[0115] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.
Claims
1. An automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence, characterized in that, Includes the following steps: The pollution event prediction module (1) obtains the animal species information, surgical type information and related operation information corresponding to the operating table to be cleaned, inputs them into the pollution identification model, and obtains the predicted pollution event information. The contaminated area construction module (2) is signal-connected to the contaminated event prediction module (1). Based on the predicted contaminated event information and combined with the equipment space structure information of the operating table, it generates the predicted contaminated area information. The pollution score calculation module (3) is connected to the pollution area construction module (2) by signal. Based on the predicted pollution area information, it calculates the pollution impact score for each predicted pollution area to obtain the comprehensive pollution area score. The cleaning path construction module (4) is connected to the pollution score calculation module (3) by signal. Based on the comprehensive score of the polluted area, a pollution cleaning priority list is generated. Combined with the physical location data of the polluted area, a path optimization algorithm is used to construct the cleaning path of the polluted area. The cleaning strategy generation module (5) is signal-connected to the cleaning path construction module (4), obtains the functional information of the polluted area, generates cleaning operation information based on the pollution cleaning path, and performs automated cleaning operation; The compensation cleaning trigger module (6) is connected to the cleaning strategy generation module (5) by signal. The cleaning task is executed in stages according to the preset time window. After each stage of cleaning operation is completed, the image information of the current area is obtained, and it is determined whether to trigger the compensation cleaning operation.
2. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 1, characterized in that, The step of obtaining the animal species information, surgical type information, and related operation information corresponding to the operating table to be cleaned, inputting them into the contamination identification model, and obtaining the predicted contamination event information is as follows: Obtain the animal species information corresponding to the operating table to be cleaned, wherein the animal species information includes animal species information and body size classification information; Obtain the surgical type information corresponding to the operating table to be cleaned, including surgical type information and standard surgical procedure; Obtain relevant operation information corresponding to the operating table to be cleaned, including the type of operating instrument, frequency of operating actions, and operation execution time; The animal species information, surgery type information, and related operation information are subjected to feature encoding processing to generate structured feature vectors; The structured feature vector is input into the contamination identification model, and trained based on historical surgical contamination samples to obtain predicted contamination event information.
3. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 1, characterized in that, The step of generating predicted contamination area information based on the predicted contamination event information and the equipment spatial structure information of the operating table is as follows: Based on the predicted pollution event information, the predicted pollution event information includes the spatial coordinates of the directly polluted area, the pollution type, and the pollution level; Acquire the equipment spatial structure information of the operating table, which includes the operation path trajectory of instruments in each functional area of the operating table during the operation. Based on the operation path trajectory, calculate the distance and proximity between the instrument contact point and each spatial area of the operating table, and establish the spatial proximity relationship between the operating table spatial area and the instrument operation path. Based on the spatial proximity relationship, the spatial location coordinates, pollution type and pollution level of the directly polluted area are extracted, and the probability of pollutants spreading from the directly polluted area to the adjacent area is calculated to obtain the pollution spread probability. Based on the pollution propagation probability, a pollution propagation probability map is constructed to generate information on predicted polluted areas.
4. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 3, characterized in that, The step of constructing a pollution propagation probability map and generating predicted pollution area information based on the pollution propagation probability is as follows: Based on the pollution propagation probability, the operating table space area is used as a node, and the pollution propagation probability between areas is used as an edge to construct a pollution propagation probability graph. Based on the pollution propagation probability map, and according to a preset probability threshold, a graph traversal algorithm is used to identify potential pollution propagation areas where the pollution propagation probability exceeds the preset probability threshold. The directly polluted areas are merged with the potentially spreading polluted areas to form a complete polluted area combination, generating predicted polluted area information.
5. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 1, characterized in that, The step of calculating a pollution impact score for each predicted pollution area based on the predicted pollution area information to obtain a comprehensive pollution area score is as follows: Based on the predicted pollution area information, the pollution type, pollution area, and pollution residual duration of each predicted pollution area are collected. Based on the pollution type, pollution area, and pollution residual duration, the corresponding pollution type risk coefficient, pollution area risk coefficient, and pollution residual risk coefficient are obtained respectively. Based on the pollution type risk coefficient, pollution area risk coefficient, and pollution residue risk coefficient, a weighted fusion model is used to calculate the pollution impact score, resulting in a comprehensive score for the polluted area.
6. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 1, characterized in that, The steps of generating a pollution cleanup priority list based on the comprehensive score of the polluted area, and constructing a cleanup path for the polluted area using a path optimization algorithm in conjunction with the physical location data of the polluted area, are as follows: Based on the comprehensive score of the polluted area, a pollution cleanup priority list is generated by sorting the comprehensive score values from high to low. Based on the physical location data of the polluted areas, the geometric distance between the polluted areas is calculated, and a spatial distance matrix between the polluted areas is constructed according to the geometric distance. Based on the pollution cleanup priority list and spatial distance matrix, a path optimization algorithm is used to generate a cleanup path for the polluted area.
7. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 1, characterized in that, The steps of obtaining functional information about the contaminated area, generating cleaning operation information based on the contaminated cleaning path, and executing automated cleaning operations are as follows: Obtain functional information of the contaminated area, which includes the operational function type of the contaminated area and its corresponding surface material type and cleaning level requirements; Based on the comprehensive score of the contaminated area, the surface material type, and the cleaning level requirements, cleaning operation information for the contaminated area is generated. The cleaning operation information includes the type of cleaning agent, the dosage of cleaning agent, the sequence of cleaning actions, and the cleaning duration. Based on the pollution cleanup path, the cleanup operation information is combined with the pollution cleanup path to form a cleanup control sequence; Based on the cleaning control sequence, corresponding automated cleaning operations are performed on the contaminated area to complete the automatic cleaning process of the contaminated area.
8. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 1, characterized in that, The step of executing the cleaning task in stages according to a preset time window, and obtaining the current area image information and determining whether to trigger a compensation cleaning operation after each stage of cleaning is completed, specifically includes: The task of cleaning up contaminated areas is divided into multiple stages according to a preset time window, with each stage corresponding to a part of the contaminated area in the contaminated cleanup path. After each stage of the cleaning operation is completed, the image information of the current area is obtained, the image of the current area is preprocessed, and the standard image of the current area is obtained. Pollution residue features are extracted from the standard image of the current area, including brightness anomalies and color deviations; The presence area and density of pollution residue characteristics are extracted, and the presence area and density are calculated to obtain the pollution residue value. Based on the pollution residue value, determine whether to trigger a compensation cleanup operation.
9. The automatic cleaning control system for livestock and veterinary operating tables based on artificial intelligence according to claim 8, characterized in that, The step of determining whether to trigger a compensation cleanup operation based on the residual pollution value specifically includes: Set a pollution residue threshold and compare the pollution residue value with the pollution residue threshold; If the pollution residue value exceeds the pollution residue threshold, a pollution cleanup failure mark will be generated; Based on the pollution cleanup failure indicator, a compensation cleanup signal for the corresponding target area is triggered, and re-cleanup operation information is generated according to the pollution residue characteristics to perform compensation cleanup operation on the target area. Once the maximum number of compensation attempts is reached, the current phase of cleanup is confirmed to be complete, and the process proceeds to the next phase of cleanup.
