A hotel pest control device capable of autonomously planning a disinfection path

Abstract

The present application relates to the technical field of intelligent disinfection and killing equipment, and more particularly to a hotel pest control device capable of autonomously planning a disinfection and killing path, which comprises environment perception, initial path planning, dynamic interference calibration, pesticide efficacy feedback optimization and scheme pushing modules. An initial disinfection and killing path is generated by collecting pest distribution situation data, dynamic correction and avoidance compensation are performed based on real-time human flow density coefficient and guest room privacy sensitivity coefficient, and energy efficiency calibration and adhesion correction are performed in combination with pesticide diffusion attenuation coefficient and surface material adsorption rate. The present application solves the problems of rigid disinfection and killing path, pesticide waste and high personnel exposure risk in the prior art, realizes the synergistic optimization of disinfection and killing effect, privacy protection and energy efficiency, and is suitable for pest control in complex hotel scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent pest control equipment technology, and in particular to a hotel pest control device that can autonomously plan its pest control path. Background Technology

[0002] In the field of pest control in hotels, although ultraviolet trapping technology, pheromone trapping devices, and thermal fogging machines have been widely used in recent years, existing solutions mostly rely on fixed-point deployment or manual inspection, making it difficult to respond in real time to dynamic occupancy status in hotel rooms, fluctuations in pedestrian density in public areas, and the spatiotemporal decay characteristics of pesticides. Especially given the high time sensitivity, high privacy, and high mobility characteristics of hotels, traditional pest control route planning often leads to noise disturbances during guest room occupancy, exposure risks to people in public areas, and excessive chemical residues caused by repeated spraying.

[0003] For example, Chinese patent CN117281095A discloses an automatic robot for pest capture, which includes a base plate, a solar cell system on the base plate, a pest capture system, and a driving system on both sides of the base plate. Pests are attracted by pest trapping lights on the pest capture system. After the pests approach, they are selected by a fan negative pressure system and a pest directional selection net, and then eliminated by a high-voltage electric grid and stored in a pest storage box. However, this solution does not involve dynamic planning of the pest control path, nor does it consider the coupling effect of personnel flow and pesticide diffusion in a hotel setting, and cannot achieve coordinated optimization of pest control operations and hotel operation status. Summary of the Invention

[0004] Therefore, the present invention provides a hotel pest control device that can autonomously plan the pest control path to overcome the problems of rigid pest control paths, waste of pesticides, and high risk of personnel exposure caused by the failure to consider dynamic human flow interference, real-time occupancy status of guest rooms, and spatiotemporal decay of pesticide efficacy in the prior art.

[0005] To achieve the above objectives, the present invention provides a hotel pest control device capable of autonomously planning its pest control path, comprising:

[0006] The environmental sensing module is used to collect data on hotel building topology, pest distribution patterns, and guest room occupancy status.

[0007] The initial path planning module is used to generate an initial pest control path plan based on the pest distribution data.

[0008] The dynamic interference calibration module is used to calculate the real-time crowd density coefficient and to perform real-time calibration of the generation process of the initial disinfection path scheme based on the real-time crowd density coefficient. It is also used to obtain the guest room privacy sensitivity coefficient and to perform avoidance compensation in the real-time calibration process based on the guest room privacy sensitivity coefficient.

[0009] The drug efficacy feedback optimization module is used to calculate the drug diffusion attenuation coefficient, perform energy efficiency calibration on the dynamic interference calibration process based on the drug diffusion attenuation coefficient, obtain the surface material adsorption rate, perform adhesion correction on the energy efficiency calibration process based on the surface material adsorption rate, obtain the real-time electricity price peak-valley coefficient, and perform economic optimization on the adhesion correction process based on the real-time electricity price peak-valley coefficient.

[0010] The solution push module is used to push the final disinfection path solution.

[0011] Furthermore, when the initial path planning module generates the initial pest control path plan based on the pest distribution data, it uploads the pest density field ρ, pest activity frequency ν, and building equivalent barrier coefficient R from the pest distribution data collected at each edge node to the pre-trained path optimization calculation model in the cloud to obtain the theoretical path length L0 and the theoretical pesticide spraying amount Q0, where:

[0012] The theoretical path length L0 = ∫(ρ×ν) / Rds, where s is the path integration variable;

[0013] The theoretical pesticide spraying amount Q0 = μ × ∫ρ × dA, where μ is the standard pesticide amount per unit area and dA is the area of ​​the micro-element;

[0014] The theoretical spray volume Q0 is compared with the corrosion resistance limit value Qmax of the floor material. Based on the comparison results, the design spray volume is judged, and the basic target spray volume Qbase is calculated based on the judgment results.

[0015] When Q0≤Qmax, the initial path planning module determines that the design spray volume is not overloaded, sets Qbase=Q0, and outputs the basic target spray volume Qbase, theoretical path length L0 and theoretical moving speed vbase as the initial disinfection path scheme.

[0016] When Q0 > Qmax, the initial path planning module determines that the design spray volume is overloaded, sets Qbase = Qmax × 0.85, vbase = L0 / tmax, where tmax is the maximum allowable operation time, and outputs the basic target spray volume Qbase, theoretical path length L0 and theoretical moving speed vbase as the initial disinfection path scheme.

[0017] Furthermore, the initial path planning module constructs the path optimization calculation model through the following steps:

[0018] Step S1: Collect path planning sample data under different hotel star ratings, different building layouts, and different pest density levels. Remove outliers using the Grubbs criterion, fill in missing values ​​using the Kriging interpolation algorithm, and map data of different dimensions to the [0,1] interval using the Min-Max standardization method to construct a path optimization training dataset containing 5000 samples.

[0019] Step S2: A deep neural network is used as the basic architecture, and the number of input layer nodes is set to 10. The 10 nodes of the input layer receive pest density features, building geometric features and timeliness features respectively.

[0020] Step S3: Set the hidden layer to 3 fully connected layers, where each fully connected layer contains 128 neurons and uses ReLU activation function and Batch Normalization. The number of output layer nodes is 2. The output layer nodes are used to output the theoretical path length and theoretical drug spray amount.

[0021] Step S4: The Adam optimizer is used with an initial learning rate of 0.001 and a batch size of 32. The training iterations are 300 rounds, and the mean squared error is used as the loss function. Training is stopped when the validation set loss does not decrease for 10 consecutive rounds, thus obtaining the path optimization calculation model.

[0022] Furthermore, the dynamic interference calibration module calculates the real-time pedestrian density coefficient and performs real-time calibration of the initial disinfection path generation process based on the real-time pedestrian density coefficient. It calculates the real-time pedestrian density coefficient D based on the transient value ΔP of the corridor pedestrian flow, the elevator hall occupancy rate η, and the equivalent width W of the building passageway in the environmental perception data, setting D=(α×|ΔP| / W+β×η), where α is the pedestrian flow change influence coefficient and β is the space occupancy influence coefficient. The real-time pedestrian density coefficient D is compared with a preset density threshold Dth, and the pedestrian interference situation is judged based on the comparison result. The initial disinfection path generation process is then calibrated in real-time based on the judgment result.

[0023] When D≤Dth, the situation of pedestrian interference is determined to be in a steady state, and the generation process of the initial disinfection path plan is not checked in real time.

[0024] When D > Dth, the pedestrian flow interference is determined to be a transient interval. The initial disinfection path generation process is checked in real time. The theoretical path length L0 is extended by the path detour coefficient γ, and γ is set to 1 + 0.35 × (D - Dth) / Dmax to obtain the checked path length Lc1. Lc1 is set to L0 × γ. The theoretical moving speed vbase is reduced by the speed attenuation coefficient φ, and φ is set to 1 - 0.22 × (D - Dth) / Dth to obtain the checked moving speed vc1. vc1 is set to vbase × φ. The theoretical path length L0 and theoretical moving speed vbase are replaced with the checked path length Lc1 and checked moving speed vc1 to obtain the checked disinfection path plan.

[0025] Furthermore, the dynamic interference calibration module acquires the guest room privacy sensitivity coefficient and, when performing avoidance compensation during the real-time calibration process based on the guest room privacy sensitivity coefficient, calculates the guest room privacy sensitivity coefficient P based on the room door closure status Sc, indoor personnel activity frequency Fa, and time-period privacy weight coefficient ω in the guest room occupancy status data. P is set as P = Sc × (Fa × ω + (1 - Fa) × 0.3), where Sc = 1 indicates the room door is closed and Sc = 0 indicates the room door is open. The guest room privacy sensitivity coefficient P is compared with a preset sensitivity threshold P0. Based on the comparison result, a judgment is made regarding the privacy risk, and based on the judgment result, avoidance compensation is performed during the real-time calibration process.

[0026] When P≤P0, the privacy risk situation is determined to be in the low-risk range, and no compensation is given for the real-time verification process.

[0027] When P > P0, the privacy risk situation is determined to be in the high-risk range. The real-time calibration process is compensated by avoiding obstacles. The calibration path length Lc1 is extended again by the avoidance compensation coefficient δ. δ is set to 1 + 0.45 × (P - P0) / P0 to obtain the compensated path length Lc2. Lc2 is set to Lc1 × δ. The calibration path length Lc1 in the calibration elimination path scheme is replaced with the compensated path length Lc2 to obtain the compensated elimination path scheme.

[0028] Further, the step of comparing the drug diffusion attenuation coefficient E with the first preset diffusion threshold E1 and the second preset diffusion threshold E2, judging the drug energy efficiency attenuation based on the comparison result, and performing energy efficiency calibration on the dynamic interference calibration process based on the judgment result includes;

[0029] When E≤E1, the energy efficiency decay of the agent is judged to be normal diffusion, and energy efficiency calibration is not performed during the dynamic interference calibration process.

[0030] When E1<E≤E2, the agent energy efficiency decay is determined to be slight decay. Energy efficiency calibration is performed during the dynamic interference calibration process. The base target spray volume Qbase is increased by the first spray volume compensation coefficient ε1. ε1=1.08 is set to obtain the first calibrated target spray volume Qcal1. Qcal1=Qbase×ε1 is set to replace the value of the base target spray volume Qbase with the value of the first calibrated target spray volume Qcal1.

[0031] When E > E2, the agent's energy efficiency is determined to be severely degraded. Energy efficiency calibration is performed during the dynamic interference calibration process. The base target spray volume Qbase is increased by the second spray volume compensation coefficient ε. ε2 is set to 1.18 to obtain the second calibrated target spray volume Qcal2. Qcal2 is set to Qbase × ε2. The theoretical moving speed vbase is replaced with the compensation speed vcomp. vcomp is set to 0.85 × vbase. The value of the base target spray volume Qbase is replaced with the value of the second calibrated target spray volume Qcal2, and the value of the theoretical moving speed vbase is replaced with the value of the compensation speed vcomp.

[0032] Furthermore, when the drug efficacy feedback optimization module acquires the surface material adsorption rate and performs adhesion correction on the energy efficiency calibration process based on the surface material adsorption rate, it calculates the surface material adsorption rate ηs based on the carpet material coefficient λc, wood flooring coefficient λw, and tile smoothness coefficient λt in the environmental perception data. ηs is set as ηs = κc × λc + κw × λw + κt × λt, where κc, κw, and κt are the weighting coefficients for different materials. The surface material adsorption rate ηs is compared with the preset adsorption threshold ηs0. Based on the comparison result, the adhesion risk is judged, and the adhesion correction is performed on the energy efficiency calibration process based on the judgment result.

[0033] Furthermore, the step of comparing the surface material adsorption rate ηs with the preset adsorption threshold ηs0, judging the adhesion risk based on the comparison result, and correcting the adhesion in the energy efficiency calibration process based on the judgment result includes:

[0034] When ηs≥ηs0, the adhesion risk is determined to be in the high adsorption range, and no adhesion correction is performed during the energy efficiency calibration process.

[0035] When ηs < ηs0, the adhesion risk is determined to be in the low adsorption range. Adhesion correction is applied to the energy efficiency calibration process, the increase in spray volume in the energy efficiency calibration is canceled, and the target spray volume Qcal after calibration is reduced by the adhesion correction coefficient ζ. ζ = 0.92 is set to obtain the corrected target spray volume Qrev. Qrev = Qcal × ζ is set, and the value of the target spray volume Qcal after calibration is replaced with the value of the corrected target spray volume Qrev.

[0036] Furthermore, when the solution push module pushes the final disinfection path solution, it pushes the final disinfection path solution to the distributed mobile disinfection terminal through a wireless communication network.

[0037] Compared with existing technologies, the beneficial effects of this invention are as follows: The device uses an environmental perception module to collect high-frequency data on hotel building topology, pest distribution patterns, and guest room occupancy status, providing a data foundation for subsequent dual-objective optimization control; the device uses an initial path planning module to generate an initial pest control path, solving the static setting deviation problem in existing technologies where pest control paths rely on fixed values ​​based on human experience and do not consider the actual distribution characteristics of pests, thus making the initial control parameters physically interpretable and adaptable to operating conditions; the device uses a dynamic interference calibration module to calculate the real-time pedestrian density coefficient and perform real-time calibration of the initial pest control path, solving the response lag problem in existing technologies where pest control systems cannot cope with sudden changes in passenger flow leading to personnel interference, and simultaneously using a guest room privacy sensitivity coefficient to avoid and compensate for the real-time calibration process, solving the problem of... The existing technology addresses privacy violations and customer complaints caused by differences in guest room occupancy status. The device uses a drug efficacy feedback optimization module to calculate the drug diffusion attenuation coefficient and perform energy efficiency calibration on the dynamic interference calibration process, solving the problem of drug efficacy attenuation caused by environmental airflow changes and temperature and humidity fluctuations. Simultaneously, it uses surface material adsorption rate to correct adhesion during the energy efficiency calibration process, addressing waste or insufficient coverage caused by differences in drug adhesion on different ground materials. Furthermore, it uses real-time electricity price peak-valley coefficients to optimize the economic efficiency of the adhesion correction process, solving the problem of high operating costs caused by the failure to avoid high electricity prices during peak hours in existing technologies. The device also uses a scheme push module to push the final disinfection path scheme, enabling mobile disinfection units to coordinate actions based on multi-objective optimization results, achieving Pareto optimality between disinfection effect and energy efficiency. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the hotel pest control device that can autonomously plan its pest control path, as described in this embodiment.

[0039] Figure 2 This is a schematic diagram of the dynamic interference calibration module in this embodiment;

[0040] Figure 3 This is a schematic diagram of the drug efficacy feedback optimization module in this embodiment. Detailed Implementation

[0041] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0042] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0043] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0044] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0045] Please see Figure 1 As shown, this is a schematic diagram of the hotel pest control device that can autonomously plan its pest control path according to this embodiment. The device includes:

[0046] The environmental sensing module is used to collect data on hotel building topology, pest distribution patterns, and guest room occupancy status.

[0047] An initial path planning module is used to generate an initial pest control path plan based on the pest distribution data. The initial path planning module is connected to the environmental perception module.

[0048] The dynamic interference calibration module is used to calculate the real-time pedestrian density coefficient and to perform real-time calibration of the generation process of the initial disinfection path plan based on the real-time pedestrian density coefficient. It is also used to obtain the guest room privacy sensitivity coefficient and to perform avoidance compensation in the real-time calibration process based on the guest room privacy sensitivity coefficient. The dynamic interference calibration module is connected to the initial path planning module.

[0049] The drug efficacy feedback optimization module is used to calculate the drug diffusion attenuation coefficient, perform energy efficiency calibration on the dynamic interference calibration process based on the drug diffusion attenuation coefficient, obtain the surface material adsorption rate, and perform adhesion correction on the energy efficiency calibration process based on the surface material adsorption rate. The drug efficacy feedback optimization module is connected to the dynamic interference calibration module.

[0050] The scheme push module is used to push the final disinfection path scheme, and the scheme push module is connected to the drug efficacy feedback optimization module.

[0051] Specifically, the hotel pest control device capable of autonomously planning its pest control path is applied to hotel pest control scenarios. The device uses an environmental sensing module to collect high-frequency data on hotel building topology, pest distribution patterns, and guest room occupancy status, providing a data foundation for subsequent dual-objective optimization control. The device generates an initial pest control path plan through an initial path planning module, addressing the static setting deviation problems in existing technologies where pest control paths rely on fixed values ​​based on human experience and do not consider the actual distribution characteristics of pests, thus ensuring the initial control parameters have physical interpretability and adaptability to operating conditions. The device uses a dynamic interference calibration module to calculate the real-time pedestrian density coefficient and perform real-time calibration of the initial pest control path plan, solving the response lag problem in existing pest control systems that cannot cope with sudden changes in passenger flow causing interference. Simultaneously, it utilizes guest room privacy settings... The privacy sensitivity coefficient compensates for the real-time calibration process, addressing privacy violations and customer complaints caused by differences in guest room occupancy status in existing technologies. The device calculates the agent diffusion attenuation coefficient and performs energy efficiency calibration on the dynamic interference calibration process through an efficacy feedback optimization module, resolving efficacy attenuation issues caused by environmental airflow changes and temperature / humidity fluctuations in existing technologies. Simultaneously, it corrects adhesion during the energy efficiency calibration process through surface material adsorption rate, addressing waste or insufficient coverage caused by differences in agent adhesion on different ground materials in existing technologies, and resolving high operating costs due to the failure to avoid high electricity prices during peak grid periods in existing technologies. The device also pushes the final disinfection path plan through a scheme push module, enabling mobile disinfection units to coordinate actions based on multi-objective optimization results, achieving Pareto optimality between disinfection effect and energy efficiency.

[0052] Specifically, the hotel building topology data includes building floor plans, corridor widths, room layouts, and floor heights. The building floor plans are digital vector maps showing the functional zones and corridor connections on each floor of the hotel, acquired by the environmental perception module through the hotel's BIM system or CAD drawings. Corridor widths refer to the net width of corridors and passageways, measured in meters, collected by the environmental perception module through LiDAR scanning or architectural drawing annotations. Room layouts refer to the location distribution of functional spaces such as guest rooms, elevator lobbies, and stairwells, identified by the environmental perception module through semantic segmentation algorithms on the building floor plans. Do not extract; the floor height refers to the floor height of each floor and the vertical traffic connection relationship, in meters; the pest distribution data includes pest density field, pest activity frequency, and building equivalent barrier coefficient; the pest density field refers to the spatial distribution function of the number of pests per unit area, in insects / m², obtained by the environmental sensing module through spatial interpolation of the sticky trap monitoring data using an infrared trap counter or image recognition algorithm; the pest activity frequency refers to the activity level of pests within a specific time period, dimensionless, with a value range of 0-1, obtained by the environmental sensing module through historical monitoring data statistics or real-time video analysis. The building equivalent barrier coefficient is a comprehensive quantitative index, dimensionless, representing the degree of obstruction to robot passage and agent diffusion caused by narrow passages, threshold height differences, and fire doors. The environmental perception module determines this through geometric topology analysis combined with manual inspection records. The guest room occupancy status data includes the door's closed status, the frequency of indoor occupancy activities, and the privacy weight coefficient for the time period. The door's closed status refers to the opening and closing detection signal of the guest room entrance door; Sc=1 indicates the door is closed, and Sc=0 indicates the door is open. The environmental perception module acquires this data in real time through a door magnetic sensor or a smart lock communication interface. The frequency of indoor occupancy activities refers to the single... The ratio of the number of times the infrared human body sensor in the guest room is triggered to the maximum number of triggers within a given time period is dimensionless and ranges from 0 to 1. The environmental perception module obtains this value through the data interface of the guest room intelligent guest control system. The time period privacy weight coefficient is a weighted coefficient for the degree of privacy sensitivity in different time periods based on the hotel's operational rules. It is dimensionless and is determined by the environmental perception module through occupancy rate data from the hotel's PMS system combined with time period characteristics. Specifically, it takes 0.4-0.5 for nighttime from 22:00 to 8:00 the next day, 0.2-0.3 for daytime from 8:00 to 18:00, and 0.3-0.4 for evening from 18:00 to 22:00.

[0053] Specifically, when the initial path planning module generates the initial pest control path plan based on pest distribution data, it uploads the pest density field ρ, pest activity frequency ν, and building equivalent barrier coefficient R from the pest distribution data collected at each edge node to a pre-trained path optimization calculation model in the cloud to obtain the theoretical path length L0 and the theoretical pesticide spraying amount Q0, where:

[0054] The theoretical path length L0 = ∫(ρ×ν) / Rds, where s is the path integration variable;

[0055] The theoretical pesticide spraying amount Q0 = μ × ∫ρ × dA, where μ is the standard pesticide amount per unit area and dA is the area of ​​the micro-element;

[0056] The theoretical spray volume Q0 is compared with the corrosion resistance limit value Qmax of the floor material. Based on the comparison results, the design spray volume is judged, and the basic target spray volume Qbase is calculated based on the judgment results.

[0057] When Q0≤Qmax, the initial path planning module determines that the design spray volume is not overloaded, sets Qbase=Q0, and outputs the basic target spray volume Qbase, theoretical path length L0 and theoretical moving speed vbase as the initial disinfection path scheme.

[0058] When Q0 > Qmax, the initial path planning module determines that the design spray volume is overloaded, sets Qbase = Qmax × 0.85, vbase = L0 / tmax, where tmax is the maximum allowable operation time, and outputs the basic target spray volume Qbase, theoretical path length L0 and theoretical moving speed vbase as the initial disinfection path scheme.

[0059] Specifically, the pest density field ρ refers to the spatial distribution function of the number of pests per unit area; the pest activity frequency ν refers to the activity level of pests within a specific time period; the building equivalent barrier coefficient R refers to the comprehensive quantitative index of the degree of obstruction to robot passage and pesticide diffusion by narrow passages, threshold height differences, and fire doors; the path integral variable s refers to the arc length parameter along the disinfection path; the unit area baseline pesticide dosage μ refers to the pesticide mass required for effective disinfection per unit area, determined according to the pesticide instructions and hygiene and epidemic prevention standards, in g / m², typically 0.5-1.5 g / m²; the micro-element area dA refers to the infinitesimal area element within the path coverage area; and the floor material corrosion resistance limit value Qmax refers to the critical value at which the hotel floor decoration material can withstand the maximum amount of pesticide deposition per unit area without corrosion, discoloration, or permanent stains, in g / m², determined according to the "Technical Specifications for Hygiene and Disinfection of Public Places" and material corrosion resistance testing, with carpets taking 3.0 g / m², wood flooring taking 2.5 g / m², and ceramic taking 1.5 g / m². The brick dosage is 4.0 g / m². The designed spray volume refers to the classification and judgment result of the matching state between the required amount of pesticide and the tolerance of the ground material based on the comparison between the theoretical pesticide spray volume and the corrosion resistance limit of the floor material. The designed spray volume includes two types: no overload and overload. The basic target spray volume Qbase refers to the actual pesticide spray volume determined after the overload judgment, in g / m². The theoretical path length L0 refers to the total length of the optimal pesticide path calculated based on the distribution of pests, in meters. The theoretical moving speed vbase refers to the baseline running speed of the robot in an ideal barrier-free state, in meters / s, usually 0.3-0.8 m / s. The maximum allowable operation time tmax refers to the maximum time limit for a single pesticide operation stipulated by the hotel operator, in minutes, usually 120 minutes for night operations and 60 minutes for daytime operations. The 0.85 in Qbase=Qmax×0.85 is a 15% safety margin reserved to prevent local accumulation of pesticide and damage to the floor.

[0060] Specifically, the initial path planning module constructs the path optimization calculation model through the following steps:

[0061] Step S1: Collect path planning sample data under different hotel star ratings, different building layouts, and different pest density levels. Remove outliers using the Grubbs criterion, fill in missing values ​​using the Kriging interpolation algorithm, and map data of different dimensions to the [0,1] interval using the Min-Max standardization method to construct a path optimization training dataset containing 5000 samples.

[0062] Step S2: A deep neural network is used as the basic architecture, and the number of input layer nodes is set to 10. The 10 nodes of the input layer receive pest density features, building geometric features and timeliness features respectively.

[0063] Step S3: Set the hidden layer to 3 fully connected layers, where each fully connected layer contains 128 neurons and uses ReLU activation function and Batch Normalization. The number of output layer nodes is 2. The output layer nodes are used to output the theoretical path length and theoretical drug spray amount.

[0064] Step S4: The Adam optimizer is used with an initial learning rate of 0.001 and a batch size of 32. The training iterations are 300 rounds, and the mean squared error is used as the loss function. Training is stopped when the validation set loss does not decrease for 10 consecutive rounds, thus obtaining the path optimization calculation model.

[0065] Specifically, the path planning sample data under different hotel star ratings, building layouts, and pest density levels refers to a combination of path planning parameters and pest control effect data collected in three- to five-star hotels, and in different building forms such as corridor-style, courtyard-style, and tower-style hotels, under different pest density conditions (low, medium, and high density). The path planning sample data covers a complete spectrum of operating conditions, including a pest density field range of 0.1-10 insects / m², a building passage width range of 1.2-3.0m, and a pest activity frequency range of 0.1-0.9. The pest density characteristics refer to a feature vector composed of the density field mean, density field variance, high-density area proportion, and pest aggregation index extracted from pest distribution data. The building geometric characteristics refer to a feature vector composed of the total passage length, average passage width, room density, and floor area extracted from hotel building topology data. The timeliness characteristics refer to a feature vector composed of time period codes, operation duration limits, and cleaning window periods extracted from time data. The Batch... Normalization refers to batch normalization, which involves standardizing the neuron inputs in batches within the fully connected layers of the hidden layers. By calculating the mean and variance of the current batch of data, the input distribution is normalized to a standard normal distribution with a mean of 0 and a variance of 1. This is then transformed using learnable scaling and translation parameters to accelerate network convergence and prevent gradient vanishing. The theoretical path length refers to the total length of the optimal pest control path calculated based on the pest distribution pattern, expressed in meters (m). The theoretical pesticide application rate refers to the theoretical pesticide requirement calculated based on the pest density field integral, expressed in g / m².

[0066] Specifically, the dynamic interference calibration module calculates the real-time pedestrian density coefficient and performs real-time calibration of the initial disinfection path generation process based on the real-time pedestrian density coefficient. It calculates the real-time pedestrian density coefficient D based on the transient value ΔP of corridor pedestrian flow, the elevator hall occupancy rate η, and the equivalent width W of the building passageway from the environmental perception data. The value is set as D = (α × |ΔP| / W + β × η), where α is the pedestrian flow change influence coefficient and β is the space occupancy influence coefficient. The real-time pedestrian density coefficient D is compared with a preset density threshold Dth. Based on the comparison result, the pedestrian flow interference situation is judged, and based on the judgment result, the initial disinfection path generation process is calibrated in real-time.

[0067] When D≤Dth, the situation of pedestrian interference is determined to be in a steady state, and the generation process of the initial disinfection path plan is not checked in real time.

[0068] When D > Dth, the pedestrian flow interference is determined to be a transient interval. The initial disinfection path generation process is checked in real time. The theoretical path length L0 is extended by the path detour coefficient γ, and γ is set to 1 + 0.35 × (D - Dth) / Dmax to obtain the checked path length Lc1. Lc1 is set to L0 × γ. The theoretical moving speed vbase is reduced by the speed attenuation coefficient φ, and φ is set to 1 - 0.22 × (D - Dth) / Dth to obtain the checked moving speed vc1. vc1 is set to vbase × φ. The theoretical path length L0 and theoretical moving speed vbase are replaced with the checked path length Lc1 and checked moving speed vc1 to obtain the checked disinfection path plan.

[0069] Specifically, the transient value ΔP of the corridor pedestrian flow refers to the ratio of the absolute value of the difference between the number of people in the corridor at the current sampling time and the previous sampling time to the sampling time interval, with units of people / min. It characterizes the drastic change in passenger flow; the larger the value, the more concentrated the sudden flow of people, such as when a meeting ends or a group checks in. The elevator lobby occupancy rate η refers to the ratio of the current number of people in the elevator lobby area to the maximum capacity of the elevator lobby. It is dimensionless and ranges from 0 to 1. The environmental perception module obtains the value through the elevator car weighing sensor or the elevator lobby camera's people counting algorithm. The equivalent width W of the building passageway refers to the width considering obstacles on both sides of the passageway and... The effective passage width after temporary storage, in meters, is obtained by the environmental perception module through real-time scanning with lidar or manual annotation and updating. The real-time pedestrian density coefficient D is a dimensionless quantitative index of dynamic interference, combining the rate of change in pedestrian flow in the corridor with the degree of space occupancy in the elevator lobby. The pedestrian flow change influence coefficient α is an empirical coefficient used in the calculation of the real-time pedestrian density coefficient to weight the contribution of transient values ​​of pedestrian flow in the corridor to dynamic interference. This coefficient is dimensionless and is determined based on the hotel's passenger flow characteristics, typically ranging from 0.6 to 0.7, representing the risk of sudden pedestrian flow interrupting disinfection operations. The space occupancy influence coefficient β is a quantitative index of the dynamic interference caused by sudden pedestrian flow changes in the corridor. In the calculation of the real-time pedestrian density coefficient, an empirical coefficient, dimensionless, is used to weight the contribution of elevator lobby occupancy rate to dynamic interference. This coefficient is determined based on the spatial dimensions of the elevator lobby, typically ranging from 0.3 to 0.4, representing the degree to which spatial congestion hinders robot passage, and satisfies the condition that the sum of the pedestrian flow change influence coefficient and the space occupancy influence coefficient is 1. The preset density threshold Dth is determined based on hotel service standards and safety regulations, typically set at 0.5 people / m². When the real-time pedestrian density coefficient D > 0.5, it indicates that the personnel density in the passage exceeds the robot's safe operating threshold, requiring real-time calibration. The pedestrian interference situation refers to the situation based on real-time pedestrian flow density. The comparison results between the flow density coefficient and the preset density threshold, and the classification and judgment results of the matching status between the degree of passenger flow fluctuation and the robot operation safety, are used to determine the passenger flow interference situation, which includes two types: steady-state interval and transient interval; the path detour coefficient γ = 1 + 0.35 × (D - Dth) / Dmax, where 0.35 is used. According to the "Safety Requirements for Service Robots" and hotel operation experience, under high passenger flow density conditions, the path needs to be extended by 30%-40% to avoid pedestrians. In this embodiment, the middle value of 35% is taken, so it is set to 0.35; the speed attenuation coefficient φ = 1 - 0.22 × (D - Dth) / Dth, where 0 is used.22. Based on the kinematic characteristics of robots and the principles of ergonomics, when the pedestrian density increases, the operating speed needs to be reduced by 20%-25% to ensure safety. Taking the median value of 22%, this meets the safety speed limit requirements for service robots in densely populated areas. The calibrated path length Lc1 refers to the actual operating path length after calibration for pedestrian interference, in meters (m). The calibrated moving speed vc1 refers to the actual operating speed after calibration for pedestrian interference, in m / s. The calibrated disinfection path scheme refers to the intermediate path planning result after calibration using the real-time pedestrian density coefficient.

[0070] Specifically, the dynamic interference calibration module acquires the guest room privacy sensitivity coefficient and, when performing avoidance compensation during the real-time calibration process based on the guest room privacy sensitivity coefficient, calculates the guest room privacy sensitivity coefficient P based on the room door closure status Sc, indoor occupancy activity frequency Fa, and time-period privacy weight coefficient ω in the guest room occupancy status data. P is set as P = Sc × (Fa × ω + (1 - Fa) × 0.3), where Sc = 1 indicates the room door is closed and Sc = 0 indicates the room door is open. The guest room privacy sensitivity coefficient P is compared with a preset sensitivity threshold P0. Based on the comparison result, the privacy risk is judged, and based on the judgment result, avoidance compensation is performed during the real-time calibration process.

[0071] When P≤P0, the privacy risk situation is determined to be in the low-risk range, and no compensation is given for the real-time verification process.

[0072] When P > P0, the privacy risk situation is determined to be in the high-risk range. The real-time calibration process is compensated by avoiding obstacles. The calibration path length Lc1 is extended again by the avoidance compensation coefficient δ. δ is set to 1 + 0.45 × (P - P0) / P0 to obtain the compensated path length Lc2. Lc2 is set to Lc1 × δ. The calibration path length Lc1 in the calibration elimination path scheme is replaced with the compensated path length Lc2 to obtain the compensated elimination path scheme.

[0073] Specifically, the door closing status Sc refers to the opening and closing detection signal of the guest room entrance door. Sc=1 indicates the door is closed, and Sc=0 indicates the door is open. The environmental sensing module acquires this information in real time through a door magnetic sensor or a smart door lock communication interface. The indoor occupant activity frequency Fa refers to the ratio of the number of times the infrared sensor for human presence in the guest room is triggered per unit time to the maximum number of triggers. It is dimensionless and ranges from 0 to 1. The environmental sensing module acquires this information through the guest room intelligent guest control system data interface. Fa=0 indicates that the guest room is occupied but the occupants are out, and Fa=1 indicates that the occupants are continuously active in the guest room. The time period hidden... The privacy weighting coefficient ω is a dimensionless, dimensionless coefficient that weights the degree of privacy sensitivity at different times based on the hotel's operational patterns. It is determined by the environmental perception module using occupancy data from the hotel's PMS system combined with time-specific characteristics. Specifically, it is 0.4-0.5 for nighttime (22:00-8:00 the next day), 0.2-0.3 for daytime (8:00-18:00), and 0.3-0.4 for evening (18:00-22:00), representing the differences in guests' sensitivity to noise and visual intrusion from robots operating outside their doors at different times. The guest room privacy sensitivity coefficient P is a coefficient that comprehensively considers the room door status, indoor activities, and time-specific characteristics to influence guest room privacy. The quantitative indicator of protection needs is dimensionless and ranges from 0 to 1. The preset sensitivity threshold P0 is determined based on hotel service quality standards and guest complaint data analysis, and is typically set to 0.35. When the room privacy sensitivity coefficient P > 0.35, it indicates that the room is currently in a state of high privacy protection needs, and avoidance compensation needs to be activated. The privacy risk situation refers to the classification and judgment result of the degree of privacy leakage risk of the room based on the comparison result of the room privacy sensitivity coefficient and the preset sensitivity threshold. The privacy risk situation includes two types: low risk range and high risk range. The avoidance compensation coefficient δ = 1 + 0.4 The 0.45 in 5×(P-P0) / P0, according to hotel service psychology and privacy protection standards, requires that when a guest room is in a highly privacy-sensitive state, the path detour distance should be increased by 40%-50% to avoid the area directly opposite the room door. The median value of 45% is taken. This value has been verified by on-site tests of multiple high-star hotels. It can effectively reduce the perceived intrusion of guests without excessively sacrificing the disinfection coverage. The compensated path length Lc2 refers to the final operation path length after privacy avoidance compensation, in meters. The compensated disinfection path scheme refers to the final path planning result after compensation by the guest room privacy sensitivity coefficient.

[0074] Specifically, the drug efficacy feedback optimization module calculates the drug diffusion attenuation coefficient. When performing energy efficiency calibration for the dynamic interference calibration process based on the drug diffusion attenuation coefficient, it calculates the drug diffusion attenuation coefficient E based on the air flow rate v, temperature and humidity attenuation factor η, and drug particle size distribution coefficient dp in the environmental perception data. The module sets E = ω1 × (v / v0) + ω2 × (η / η0) + ω3 × (dp / dp0), where ω1 is the first diffusion weight coefficient, ω2 is the second diffusion weight coefficient, ω3 is the third diffusion weight coefficient, v0 is the reference air flow rate, η0 is the reference temperature and humidity factor, and dp0 is the reference particle size. The drug diffusion attenuation coefficient E is compared with the first preset diffusion threshold E1 and the second preset diffusion threshold E2. Based on the comparison result, the drug energy efficiency attenuation is judged, and based on the judgment result, energy efficiency calibration is performed for the dynamic interference calibration process, including:

[0075] When E≤E1, the energy efficiency decay of the agent is judged to be normal diffusion, and energy efficiency calibration is not performed during the dynamic interference calibration process.

[0076] When E1<E≤E2, the agent energy efficiency decay is determined to be slight decay. Energy efficiency calibration is performed during the dynamic interference calibration process. The base target spray volume Qbase is increased by the first spray volume compensation coefficient ε1. ε1=1.08 is set to obtain the first calibrated target spray volume Qcal1. Qcal1=Qbase×ε1 is set to replace the value of the base target spray volume Qbase with the value of the first calibrated target spray volume Qcal1.

[0077] When E > E2, the agent's energy efficiency is determined to be severely degraded. Energy efficiency calibration is performed during the dynamic interference calibration process. The base target spray volume Qbase is increased by the second spray volume compensation coefficient ε. ε2 is set to 1.18 to obtain the second calibrated target spray volume Qcal2. Qcal2 is set to Qbase × ε2. The theoretical moving speed vbase is replaced with the compensation speed vcomp. vcomp is set to 0.85 × vbase. The value of the base target spray volume Qbase is replaced with the value of the second calibrated target spray volume Qcal2, and the value of the theoretical moving speed vbase is replaced with the value of the compensation speed vcomp.

[0078] Specifically, the air circulation rate v refers to the airflow speed in the hotel's public areas caused by the air conditioning system or the opening of doors and windows, measured in m / s. The environmental sensing module collects this data using a hot-wire anemometer or ultrasonic anemometer. The temperature and humidity attenuation factor η is a dimensionless coefficient representing the combined influence of current ambient temperature and relative humidity on the evaporation rate of the pesticide. The environmental sensing module calculates this value using temperature and humidity sensor data combined with the pesticide evaporation characteristic curve; higher temperatures and lower humidity result in a larger η value and faster pesticide attenuation. The pesticide particle size distribution coefficient dp is the ratio of the pesticide droplet size generated by the spray device to the ideal deposition particle size, also dimensionless. The environmental sensing module obtains this value using a laser particle size analyzer or a spray image analysis algorithm. The diffusion weighting coefficient ω1 is the weighting coefficient used to weight the airflow rate when calculating the drug diffusion attenuation coefficient. It is obtained by pairwise comparison matrix calculation based on the relative importance of airflow carrying, temperature and humidity volatilization, and particle size drift on drug diffusion using the analytic hierarchy process. It is typically taken as 0.40-0.50, and in this embodiment, it is taken as 0.45, representing the dominant role of airflow in drug diffusion. The second diffusion weighting coefficient ω2 is the weighting coefficient used to weight the temperature and humidity attenuation factor when calculating the drug diffusion attenuation coefficient. It is typically taken as 0.30-0.40, and in this embodiment, it is taken as 0.35, representing the degree of influence of environmental conditions on drug stability. The third diffusion weighting coefficient ω3 is the weighting coefficient used to weight the drug diffusion attenuation factor when calculating the drug diffusion attenuation coefficient. The weighting coefficient of the agent particle size distribution coefficient is usually taken as 0.15-0.25, and in this embodiment, it is taken as 0.20. It characterizes the influence of droplet size on deposition efficiency and satisfies the condition that the sum of the first, second, and third diffusion weighting coefficients is 1. The reference air velocity v0 refers to the typical wind speed when the hotel air conditioning system is operating normally, and is usually taken as 0.2 m / s. The reference temperature and humidity factor η0 refers to the attenuation reference value under standard operating conditions, and is taken as 1.0. The reference particle size dp0 refers to the ideal deposition particle size, and is usually taken as 50 μm. The first preset diffusion threshold E1 refers to the first critical threshold for determining the degree of agent diffusion attenuation, and is usually taken as 1.10-1.20, and in this embodiment, it is taken as 1.15. The second preset diffusion threshold E2 refers to the threshold for determining the degree of agent diffusion attenuation. The second critical threshold for the degree is usually set at 1.35-1.45, and in this embodiment, it is set at 1.40. The agent energy efficiency attenuation refers to the classification and judgment result of the degree of attenuation of the agent's effective deposition ability in the environment based on the comparison result of the agent diffusion attenuation coefficient and the preset diffusion threshold. The agent energy efficiency attenuation includes three types: normal diffusion, slight attenuation, and severe attenuation. The reason for ε1=1.08 is that, according to the agent deposition kinetics and the "Technical Specifications for Hygiene and Disinfection in Public Places", when the agent diffusion attenuation coefficient E is in the range of 1.15-1.40, in order to maintain the effective disinfection concentration, the spray volume needs to be increased by about 8%. This value has been verified by aerosol deposition experiments, which can compensate for diffusion loss without causing the ground to become too wet. ε2=1.The rationale for 18% is as follows: When the agent diffusion attenuation coefficient E > 1.40, it indicates that the environmental conditions are extremely unfavorable for agent deposition, requiring an increase in spray volume of approximately 18% to ensure the disinfection effect; the compensation velocity vcomp = 0.85 × vbase, with a coefficient of 0.85, according to spray deposition theory, reduces the moving speed, which can extend the agent's residence time in the target area by approximately 17%, compensating for the agent drift loss caused by high-speed airflow, consistent with the physical laws of aerosol gravity settling and inertial deposition; the first calibrated target spray volume Qcal1 refers to the agent spray volume after slight attenuation energy efficiency calibration, in g / m², and the second calibrated target spray volume Qcal2 refers to the agent spray volume after severe attenuation energy efficiency calibration, in g / m².

[0079] Specifically, the drug efficacy feedback optimization module acquires the surface material adsorption rate and, when correcting adhesion during the energy efficiency calibration process based on the surface material adsorption rate, calculates the surface material adsorption rate ηs based on the carpet material coefficient λc, wood flooring coefficient λw, and tile smoothness coefficient λt from the environmental perception data. ηs is set as ηs = κc × λc + κw × λw + κt × λt, where κc, κw, and κt are weighting coefficients for different materials. The surface material adsorption rate ηs is compared with a preset adsorption threshold ηs0. Based on the comparison result, the adhesion risk is judged, and the adhesion correction is applied to the energy efficiency calibration process based on the judgment result.

[0080] When ηs≥ηs0, the adhesion risk is determined to be in the high adsorption range, and no adhesion correction is performed during the energy efficiency calibration process.

[0081] When ηs < ηs0, the adhesion risk is determined to be in the low adsorption range. Adhesion correction is applied to the energy efficiency calibration process, the increase in spray volume in the energy efficiency calibration is canceled, and the target spray volume Qcal after calibration is reduced by the adhesion correction coefficient ζ. ζ = 0.92 is set to obtain the corrected target spray volume Qrev. Qrev = Qcal × ζ is set, and the value of the target spray volume Qcal after calibration is replaced with the value of the corrected target spray volume Qrev.

[0082] Specifically, the carpet material coefficient λc refers to a dimensionless quantitative index of the adsorption capacity of carpet fibers for pharmaceutical droplets, determined according to the carpet material type: 0.8-0.9 for nylon carpets, 0.9-1.0 for wool carpets, and 0.7-0.8 for synthetic fiber carpets; the wood flooring coefficient λw refers to a dimensionless quantitative index of the adsorption capacity of the wood flooring surface coating for pharmaceutical droplets, determined according to the flooring finish type: 0.6-0.7 for UV finishes, 0.7-0.8 for water-based finishes, and 0.8-0.9 for wax-based finishes; the tile smoothness coefficient λt refers to the tile smoothness... The quantitative index of the surface adsorption capacity for pharmaceutical droplets is dimensionless and determined based on the gloss of the tile: 0.3-0.4 for polished tiles, 0.4-0.5 for glazed tiles, and 0.5-0.6 for matte tiles. The weighting coefficients κc, κw, and κt for different materials are dimensionless weighting coefficients determined based on the proportion of each material on the hotel floor. These coefficients are obtained by normalizing the area proportion of each material from hotel cleaning and maintenance records or visual recognition algorithms, and satisfy κc + κw + κt = 1. The surface material adsorption rate ηs is the weighted average adsorption rate after considering the adsorption characteristics and area proportion of each material. The attached capability index is dimensionless, ranging from 0 to 1. A higher ηs value indicates that the surface more easily adsorbs the agent, and the greater the risk of residue. The preset adsorption threshold ηs0 is determined based on the agent residue safety standard and hotel cleaning cost analysis, and is usually set to 0.65. When the surface material adsorption rate ηs < 0.65, it indicates that the surface is mainly smooth, and the agent is easily lost and volatilized, requiring the activation of adhesion correction. The adhesion risk status refers to the classification and judgment result of the agent deposition stability and residue risk on the ground based on the comparison result of the surface material adsorption rate and the preset adsorption threshold. The adhesion risk status includes... It includes two types: high adsorption range and low adsorption range; the adhesion correction coefficient ζ is set to 0.92, which means reducing the spray volume by 8%. According to surface chemistry and wetting theory, droplets on smooth surfaces have a large contact angle and poor spreadability. Reducing the spray volume can avoid the agent from agglomerating and flowing. At the same time, by reducing the spray volume, the droplets form a uniform thin layer on the smooth surface instead of agglomerating droplets. This value has been verified by spraying experiments on ground materials, which can ensure the disinfection effect and reduce the agent residue by more than 50%; the corrected target spray volume Qrev refers to the final agent spray volume after surface material adhesion correction, and the unit is g / m².

[0083] Specifically, when the solution push module pushes the final disinfection path solution, it pushes the final disinfection path solution to the distributed mobile disinfection terminal through a wireless communication network.

[0084] Specifically, the wireless communication network refers to a wireless data transmission network built based on IEEE 802.11, Zigbee, LoRa, and 4G / 5G mobile communication standards. This wireless communication network is used to transmit the final disinfection path plan from the efficacy feedback optimization module to the distributed mobile disinfection terminal. It is suitable for complex partitions and mobile operation scenarios within hotels. The network uses the AES-128 encryption algorithm to ensure data transmission security, with a packet loss rate of less than 1%, meeting the real-time requirements of the disinfection control system. The distributed mobile disinfection terminal refers to an autonomous mobile disinfection robot deployed on each floor of the hotel, including a mobile chassis, spray actuators, variable frequency fans, and navigation sensor groups.

[0085] Please see Figure 2 As shown, this is a schematic diagram of the dynamic interference calibration module in this embodiment. The dynamic interference calibration module includes:

[0086] The crowd interference calibration unit is used to calculate the real-time crowd density coefficient and to perform real-time calibration of the initial disinfection path generation process based on the real-time crowd density coefficient.

[0087] A privacy avoidance compensation unit is used to obtain the privacy sensitivity coefficient of the guest room and to perform avoidance compensation on the real-time calibration process based on the privacy sensitivity coefficient of the guest room. The privacy avoidance compensation unit is connected to the crowd interference calibration unit.

[0088] Please see Figure 3 As shown, this is a schematic diagram of the structure of the drug efficacy feedback optimization module in this embodiment. The drug efficacy feedback optimization module includes:

[0089] The diffusion attenuation calibration unit is used to calculate the drug diffusion attenuation coefficient and perform energy efficiency calibration on the dynamic interference calibration process based on the drug diffusion attenuation coefficient.

[0090] The material adhesion correction unit is used to obtain the surface material adsorption rate and to perform adhesion correction on the energy efficiency calibration process based on the surface material adsorption rate. The material adhesion correction unit is connected to the diffusion attenuation calibration unit.

[0091] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0092] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A hotel pest control device capable of autonomously planning its elimination path, characterized in that, include: The environmental sensing module is used to collect data on hotel building topology, pest distribution patterns, and guest room occupancy status. The initial path planning module is used to generate an initial pest control path plan based on the pest distribution data. The dynamic interference calibration module is used to calculate the real-time crowd density coefficient and to perform real-time calibration of the generation process of the initial disinfection path scheme based on the real-time crowd density coefficient. It is also used to obtain the guest room privacy sensitivity coefficient and to perform avoidance compensation in the real-time calibration process based on the guest room privacy sensitivity coefficient. The drug efficacy feedback optimization module is used to calculate the drug diffusion attenuation coefficient, perform energy efficiency calibration on the dynamic interference calibration process based on the drug diffusion attenuation coefficient, obtain the surface material adsorption rate, perform adhesion correction on the energy efficiency calibration process based on the surface material adsorption rate, obtain the real-time electricity price peak-valley coefficient, and perform economic optimization on the adhesion correction process based on the real-time electricity price peak-valley coefficient. The solution push module is used to push the final disinfection path solution.

2. The hotel pest control device capable of autonomously planning its elimination path according to claim 1, characterized in that, When the initial path planning module generates the initial pest control path plan based on pest distribution data, it uploads the pest density field ρ, pest activity frequency ν, and building equivalent barrier coefficient R from the pest distribution data collected at each edge node to a pre-trained path optimization calculation model in the cloud to obtain the theoretical path length L0 and the theoretical pesticide spraying amount Q0, where: The theoretical path length L0 = ∫(ρ×ν) / Rds, where s is the path integration variable; The theoretical pesticide spraying amount Q0 = μ × ∫ρ × dA, where μ is the standard pesticide amount per unit area and dA is the area of ​​the micro-element; The theoretical spray volume Q0 is compared with the corrosion resistance limit value Qmax of the floor material. Based on the comparison results, the design spray volume is judged, and the basic target spray volume Qbase is calculated based on the judgment results. When Q0≤Qmax, the initial path planning module determines that the design spray volume is not overloaded, sets Qbase=Q0, and outputs the basic target spray volume Qbase, theoretical path length L0 and theoretical moving speed vbase as the initial disinfection path scheme. When Q0 > Qmax, the initial path planning module determines that the design spray volume is overloaded, sets Qbase = Qmax × 0.85, vbase = L0 / tmax, where tmax is the maximum allowable operation time, and outputs the basic target spray volume Qbase, theoretical path length L0 and theoretical moving speed vbase as the initial disinfection path scheme.

3. The hotel pest control device capable of autonomously planning its elimination path according to claim 2, characterized in that, The initial path planning module constructs the path optimization calculation model through the following steps: Step S1: Collect path planning sample data under different hotel star ratings, different building layouts, and different pest density levels. Remove outliers using the Grubbs criterion, fill in missing values ​​using the Kriging interpolation algorithm, and map data of different dimensions to the [0,1] interval using the Min-Max standardization method to construct a path optimization training dataset containing 5000 samples. Step S2: A deep neural network is used as the basic architecture, and the number of input layer nodes is set to 10. The 10 nodes of the input layer receive pest density features, building geometric features and timeliness features respectively. Step S3: Set the hidden layer to 3 fully connected layers, where each fully connected layer contains 128 neurons and uses ReLU activation function and Batch Normalization. The number of output layer nodes is 2. The output layer nodes are used to output the theoretical path length and theoretical drug spray amount. Step S4: The Adam optimizer is used with an initial learning rate of 0.001 and a batch size of 32. The training iterations are 300 rounds, and the mean squared error is used as the loss function. Training is stopped when the validation set loss does not decrease for 10 consecutive rounds, thus obtaining the path optimization calculation model.

4. The hotel pest control device capable of autonomously planning its elimination path according to claim 3, characterized in that, The dynamic interference calibration module calculates the real-time pedestrian density coefficient and performs real-time calibration of the initial disinfection path generation process based on the real-time pedestrian density coefficient. It calculates the real-time pedestrian density coefficient D based on the transient value ΔP of corridor pedestrian flow, elevator hall occupancy rate η, and equivalent width W of building passageways from the environmental perception data. The value is set as D = (α × |ΔP| / W + β × η), where α is the pedestrian flow change influence coefficient and β is the space occupancy influence coefficient. The real-time pedestrian density coefficient D is compared with a preset density threshold Dth. Based on the comparison result, the pedestrian flow interference situation is judged, and the initial disinfection path generation process is calibrated in real-time based on the judgment result. When D≤Dth, the situation of pedestrian interference is determined to be in a steady state, and the generation process of the initial disinfection path plan is not checked in real time. When D > Dth, the pedestrian flow interference is determined to be a transient interval. The initial disinfection path generation process is checked in real time. The theoretical path length L0 is extended by the path detour coefficient γ, and γ is set to 1 + 0.35 × (D - Dth) / Dmax to obtain the checked path length Lc1. Lc1 is set to L0 × γ. The theoretical moving speed vbase is reduced by the speed attenuation coefficient φ, and φ is set to 1 - 0.22 × (D - Dth) / Dth to obtain the checked moving speed vc1. vc1 is set to vbase × φ. The theoretical path length L0 and theoretical moving speed vbase are replaced with the checked path length Lc1 and checked moving speed vc1 to obtain the checked disinfection path plan.

5. The hotel pest control device capable of autonomously planning its elimination path according to claim 4, characterized in that, The dynamic interference calibration module acquires the guest room privacy sensitivity coefficient and, when performing avoidance compensation during the real-time calibration process based on the guest room privacy sensitivity coefficient, calculates the guest room privacy sensitivity coefficient P based on the room door closure status Sc, indoor occupancy activity frequency Fa, and time-period privacy weight coefficient ω in the guest room occupancy status data. P is set as P = Sc × (Fa × ω + (1 - Fa) × 0.3), where Sc = 1 indicates the room door is closed and Sc = 0 indicates the room door is open. The guest room privacy sensitivity coefficient P is compared with a preset sensitivity threshold P0. Based on the comparison result, the privacy risk is judged, and based on the judgment result, avoidance compensation is performed during the real-time calibration process. When P≤P0, the privacy risk situation is determined to be in the low-risk range, and no compensation is given for the real-time verification process. When P > P0, the privacy risk situation is determined to be in the high-risk range. The real-time calibration process is compensated by avoiding obstacles. The calibration path length Lc1 is extended again by the avoidance compensation coefficient δ. δ is set to 1 + 0.45 × (P - P0) / P0 to obtain the compensated path length Lc2. Lc2 is set to Lc1 × δ. The calibration path length Lc1 in the calibration elimination path scheme is replaced with the compensated path length Lc2 to obtain the compensated elimination path scheme.

6. The hotel pest control device capable of autonomously planning its elimination path according to claim 5, characterized in that, The drug efficacy feedback optimization module calculates the drug diffusion attenuation coefficient. When performing energy efficiency calibration for the dynamic interference calibration process based on the drug diffusion attenuation coefficient, it calculates the drug diffusion attenuation coefficient E based on the air flow rate v, temperature and humidity attenuation factor η, and drug particle size distribution coefficient dp in the environmental sensing data. The formula is set as E=ω1×(v / v0)+ω2×(η / η0)+ω3×(dp / dp0), where ω1 is the first diffusion weight coefficient, ω2 is the second diffusion weight coefficient, ω3 is the third diffusion weight coefficient, v0 is the reference air flow rate, η0 is the reference temperature and humidity factor, and dp0 is the reference particle size. The drug diffusion attenuation coefficient E is compared with the first preset diffusion threshold E1 and the second preset diffusion threshold E2. The energy efficiency attenuation of the drug is judged based on the comparison result, and the energy efficiency calibration process for the dynamic interference calibration is performed based on the judgment result.

7. The hotel pest control device capable of autonomously planning its elimination path according to claim 6, characterized in that, The process of comparing the drug diffusion attenuation coefficient E with the first preset diffusion threshold E1 and the second preset diffusion threshold E2, judging the drug energy efficiency attenuation based on the comparison result, and performing energy efficiency calibration on the dynamic interference calibration process based on the judgment result includes: When E≤E1, the energy efficiency decay of the agent is judged to be normal diffusion, and energy efficiency calibration is not performed during the dynamic interference calibration process. When E1<E≤E2, the agent energy efficiency decay is determined to be slight decay. Energy efficiency calibration is performed during the dynamic interference calibration process. The base target spray volume Qbase is increased by the first spray volume compensation coefficient ε1. ε1=1.08 is set to obtain the first calibrated target spray volume Qcal1. Qcal1=Qbase×ε1 is set to replace the value of the base target spray volume Qbase with the value of the first calibrated target spray volume Qcal1. When E > E2, the agent's energy efficiency is determined to be severely degraded. Energy efficiency calibration is performed during the dynamic interference calibration process. The base target spray volume Qbase is increased by the second spray volume compensation coefficient ε. ε2 is set to 1.18 to obtain the second calibrated target spray volume Qcal2. Qcal2 is set to Qbase × ε2. The theoretical moving speed vbase is replaced with the compensation speed vcomp. vcomp is set to 0.85 × vbase. The value of the base target spray volume Qbase is replaced with the value of the second calibrated target spray volume Qcal2, and the value of the theoretical moving speed vbase is replaced with the value of the compensation speed vcomp.

8. The hotel pest control device capable of autonomously planning its elimination path according to claim 7, characterized in that, The efficacy feedback optimization module acquires the surface material adsorption rate and performs adhesion correction during the energy efficiency calibration process based on the surface material adsorption rate. It calculates the surface material adsorption rate ηs based on the carpet material coefficient λc, wood flooring coefficient λw, and tile smoothness coefficient λt in the environmental perception data. The module sets ηs = κc × λc + κw × λw + κt × λt, where κc, κw, and κt are the weight coefficients for different materials. The surface material adsorption rate ηs is compared with the preset adsorption threshold ηs0. The adhesion risk is judged based on the comparison result, and the adhesion correction is performed during the energy efficiency calibration process based on the judgment result.

9. The hotel pest control device capable of autonomously planning its elimination path according to claim 8, characterized in that, The process of comparing the surface material adsorption rate ηs with the preset adsorption threshold ηs0, judging the adhesion risk based on the comparison result, and correcting the adhesion in the energy efficiency calibration process based on the judgment result includes: When ηs≥ηs0, the adhesion risk is determined to be in the high adsorption range, and no adhesion correction is performed during the energy efficiency calibration process. When ηs < ηs0, the adhesion risk is determined to be in the low adsorption range. Adhesion correction is applied to the energy efficiency calibration process, the increase in spray volume in the energy efficiency calibration is canceled, and the target spray volume Qcal after calibration is reduced by the adhesion correction coefficient ζ. ζ = 0.92 is set to obtain the corrected target spray volume Qrev. Qrev = Qcal × ζ is set, and the value of the target spray volume Qcal after calibration is replaced with the value of the corrected target spray volume Qrev.

10. The hotel pest control device capable of autonomously planning its elimination path according to claim 9, characterized in that, When the solution push module pushes the final disinfection path solution, it pushes the final disinfection path solution to the distributed mobile disinfection terminal through the wireless communication network.

Images