Intelligent elevator control method and system based on Internet of Things

Abstract

The invention relates to the technical field of intelligent elevator control, in particular to an intelligent elevator control method and system based on the Internet of Things, and the method comprises the following steps: S1, monitoring human body three-dimensional coordinate data of an elevator waiting area of each floor in real time, and collecting multi-modal sensing data in an elevator car; s2, generating a dynamic people flow distribution thermodynamic diagram of each floor, and constructing a time-space weight matrix; s3, generating an elevator dynamic scheduling strategy; s4, the abnormal state type of the lift car is judged by combining the gas component spectrum and the vibration spectrum; s5, a grading alarm instruction is sent to the central controller through the LoRaWAN protocol; and S6, automatically correcting the opening and closing speed curve of the portal crane. According to the method, efficient operation and safety guarantee of the elevator system in a complex people flow environment are achieved by fusing multi-source sensing, dynamic scheduling and an abnormal response mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator intelligent control, and particularly to an elevator intelligent control method and system based on the Internet of Things. Background Art

[0002] With the rapid development of Internet of Things technology, large-scale sensor networks, cloud computing, and big data analysis means have been widely applied in the field of smart buildings; as an important device for guiding the flow of people in high-rise buildings and public places, the safety and operating efficiency of the elevator system directly affect the comfort and travel experience of passengers; existing elevator dispatching schemes often rely on fixed logics or simple group control algorithms, lacking comprehensive perception and prediction of real-time passenger flow density and car interior status, and it is difficult to respond in a timely manner to the passenger flow explosion during peak riding periods and equipment abnormalities under special circumstances; especially in high-rise or super high-rise buildings, how to reduce the passenger waiting time and ensure safety alarms and quick responses in case of sudden failures has become a pain point problem that the industry urgently needs to solve.

[0003] However, most traditional elevator control systems only rely on floor call signals and simple weight detection devices, and are unable to obtain and integrate multi-modal sensing information to comprehensively and accurately analyze the passenger flow distribution and car operating status, resulting in problems such as low dispatching efficiency and lagging anomaly handling during actual operation; if a real-time multi-source data perception system cannot be established and an effective dynamic dispatching mechanism cannot be formed, it will be difficult to make full use of the advantages of Internet of Things technology to improve the refined management and safety stability of the elevator system. Summary of the Invention

[0004] Based on the above purposes, the present invention provides an elevator intelligent control method and system based on the Internet of Things.

[0005] An elevator intelligent control method based on the Internet of Things includes the following steps:

[0006] S1: Real-time monitor the three-dimensional human body coordinate data in the waiting areas on each floor of the building through a millimeter-wave radar array deployed on each floor, and simultaneously collect multi-modal sensing data in the elevator car, including pressure distribution matrix, gas component spectrum, and vibration spectrum;

[0007] S2: Input the three-dimensional human body coordinate data obtained in S1 into a spatial density clustering algorithm to generate a dynamic passenger flow distribution heat map for each floor, and construct a time-space weight matrix based on historical riding records;

[0008] S3: Generate an elevator dynamic dispatching strategy based on the dynamic passenger flow distribution heat map and time-space weight matrix in S2, and the dispatching strategy includes the target floor sequence and stop time window for each elevator;

[0009] S4: Perform singular value decomposition on the pressure distribution matrix obtained in S1. When it is detected that the principal component variance exceeds the preset threshold within three consecutive sampling periods, trigger multimodal data fusion analysis, and combine the gas component spectrum and the vibration spectrum to determine the type of abnormal state of the car;

[0010] S5: Based on the abnormal state determination result in S4, send a hierarchical alarm instruction to the central controller through the LoRaWAN protocol, and at the same time dynamically adjust the docking time window parameter in the elevator scheduling strategy generated in S3 according to the alarm level;

[0011] S6: Embed an adaptive impedance algorithm in the elevator door control module. When the docking time window adjusted in S5 does not match the real-time passenger flow density, automatically correct the door opening and closing speed curve to ensure that the actual boarding and alighting time deviates from the scheduling strategy by less than ±1.5 seconds.

[0012] Optionally, the specific steps of S1 include:

[0013] S11: Install two groups of millimeter-wave radar modules in a diagonal layout on the ceiling of the waiting areas on each floor of the building. Each group of modules contains 3 FMCW radar units operating in the 76 - 81 GHz band, forming an overlapping detection area, and output three-dimensional point cloud data including distance, azimuth angle, and elevation angle with a period of 250 ms;

[0014] S12: Perform moving clutter suppression processing on the original point cloud data output in S11. Use a recursive least squares filter to eliminate the static background reflection signal, retain the human motion feature points, and generate a denoised three-dimensional coordinate data set;

[0015] S13: Embeddedly lay a 16×24 array of piezoresistive thin-film sensors under the floor of the elevator car, and obtain the pressure distribution matrix at a sampling rate of 100 Hz. The matrix element value represents the pressure value at the corresponding grid position;

[0016] S14: Install a tunable diode laser absorption spectrometer in the center of the car roof, and synchronously collect the absorption spectra of CO, CO2, and CH4 through wavelength scanning;

[0017] S15: Arrange three-axis MEMS accelerometers at the four corners of the car bottom, extract the vibration spectrum characteristics through a Butterworth band-pass filter, and generate a spectrum vector containing the vibration energy in the X / Y / Z axes every 200 ms;

[0018] S16: Establish a unified timestamp server to perform time domain alignment on the three-dimensional coordinate data in S12, the pressure distribution matrix in S13, the gas absorption spectrum in S14, and the vibration spectrum in S15.

[0019] Optionally, the specific steps of S2 include: S21: Perform a planar projection transformation on the three-dimensional human body coordinate dataset obtained in S1 to eliminate the height dimension and generate a two-dimensional coordinate set; S22: Dynamically calculate the spatial density clustering radius parameter based on the current moment and the historical moments of the previous cycle ; S23: Use the DBSCAN clustering algorithm to cluster the two-dimensional coordinate set and output a set of clustering clusters; S24: Remove the abnormal points with a duration less than 5 seconds in the set of clustering clusters, retain the stable clustering clusters, and form a set of central coordinates; S25: Map the set of central coordinates to the rasterized waiting elevator area plane, construct an original density distribution map, and perform spatial interpolation using the thin plate spline interpolation algorithm to obtain the interpolated pedestrian flow density value , and then generate a continuous pedestrian flow distribution heat map.

[0025] Optionally, S2 further includes:

[0026] S26: Extract the historical elevator riding records of the past 30 days and construct a three-dimensional tensor, the expression of which is: , where represents the elevator riding frequency of the th grid at week , the rd hour, is the total number of grids, , ;

[0027] S27: Perform Tucker decomposition on the tensor to obtain a time factor matrix and a spatial factor matrix, and then reconstruct their outer product to obtain a reduced-dimensional time-space weight matrix, the formula of which is: , where is the tensor outer product operation, is the weight matrix.

[0028] Optionally, S3 specifically includes:

[0029] S31: Obtain the dynamic pedestrian flow distribution heat map and the time-space weight matrix generated in S2, and determine the total elevator riding demand of each floor during the current scheduling cycle based on the two;

[0030] S32: Match the elevator riding demand of each floor with the available capacity of the elevator group, use a group control algorithm based on minimizing the waiting time of the objective function to establish a priority ranking for all floors, and determine the target floor sequence of each elevator;

[0031] S33: For the target floor sequence of each elevator in S32, in combination with the aforementioned elevator ride demand prediction results, allocate the docking time window;

[0032] S34: Record in the generated elevator dispatching strategy the respective elevators and their corresponding target floor sequences and docking time windows.

[0033] Optionally, S33 specifically includes:

[0034] S331: According to the target floor sequences of each elevator determined in S32, count the total elevator ride demand for each target floor, and in combination with the maximum carrying capacity of the elevator, calculate the up and down passenger ratio coefficient for this floor ;

[0035] S332: According to the door opening and closing lag time of the elevator door machine and the average boarding and alighting time required for each passenger, in combination with the ratio coefficient , calculate the recommended docking time window for the elevator on the floor. The formula is: , where is the docking time window for the floor, is the fixed time delay for the elevator door to open and close, is the average boarding and alighting time for a single person, represents the number of people expected to board and alight;

[0036] S333: Assign the docking time windows of each target floor to the corresponding elevator's dispatching instruction set in sequence, and synchronize it with the current system time axis to ensure that there are no conflicts in the time windows of multiple elevators during the dispatching cycle, and form a final executable docking time schedule.

[0037] Optionally, S4 specifically includes:

[0038] S41: Receive the in-car pressure distribution matrix obtained in S1, first perform two-dimensional matrix processing on the original pressure values within the same sampling period to obtain matrix data arranged in two dimensions of rows and columns;

[0039] S42: Perform singular value decomposition on the matrix-processed pressure data, decompose the pressure matrix into three matrices, which respectively represent the orthogonal information in the row direction, the diagonal singular value information, and the orthogonal information in the column direction;

[0040] S43: Among the several principal components obtained from the singular value decomposition, select the principal component with the highest strength, use the variance proportion it occupies as the characterization index, and compare its values in three consecutive sampling periods. When this proportion continuously exceeds the preset threshold, it is determined that there is an abnormal fluctuation in the in-car pressure distribution;

[0041] S44: After determining that an abnormal fluctuation has occurred, based on the gas component spectrum and vibration frequency spectrum data collected in S1, respectively retrieve the characteristic peak positions of the gas absorption spectrum and the concentration of vibration energy in different frequency bands. If there are changes in gas concentration or vibration intensity that exceed the normal range, it is determined that an abnormality exists;

[0042] S45: Synthesize the abnormal pressure detection results and the analysis information of the gas spectrum and vibration spectrum, and identify the type of abnormal state inside the car according to the comparison of abnormal characteristics, including gas leakage, abnormal vibration or pressure fluctuation.

[0043] Optionally, the specific steps of S5 include:

[0044] S51: Receive the abnormal state result determined in S4, generate an alarm level identifier based on the abnormal type, and digitally encapsulate the alarm information to form an alarm data packet with priority fields and abnormal type fields;

[0045] S52: Send the alarm data packet to the network receiving end of the central controller through the gateway module of the LoRaWAN protocol. The central controller parses the priority field and abnormal type field in the alarm data packet, maps them to a pre-configured alarm level table, and triggers the corresponding emergency handling strategy;

[0046] S53: According to the identified alarm level, retrieve the elevator dispatching strategy generated by S3 from the central controller, read the stop time window parameters of each elevator, and dynamically adjust the time window of the elevator or floor in the alarm state to cooperate with the on-site emergency disposal;

[0047] S54: Rewrite the adjusted stop time window parameters into the elevator dispatching instruction set, and the elevator control center issues them to the real-time control unit of the corresponding elevator.

[0048] Optionally, the specific steps of S6 include:

[0049] S61: After detecting a deviation between the stop time window adjusted by S5 and the real-time passenger flow density, obtain the actual opening and closing durations of the door machine, compare them with the preset boarding and alighting times in the elevator dispatching strategy, and calculate the boarding and alighting time error;

[0050] S62: When the absolute value of the boarding and alighting time error is greater than 1.5 seconds, enter the adaptive impedance correction process. Obtain the force information of the door machine during opening and closing through the torque sensor of the door machine control module, and combine the real-time passenger flow situation to adjust the opening and closing speed target curve;

[0051] S63: Write the corrected speed target curve into the instruction buffer of the door controller, so that the door controller automatically adjusts the torque and motion trajectory during subsequent opening or closing actions, ensuring that the deviation between the actual boarding and alighting time and the scheduling strategy is maintained within ±1.5 seconds.

[0052] An elevator intelligent control system based on the Internet of Things for implementing the above-mentioned elevator intelligent control method based on the Internet of Things, including:

[0053] Millimeter-wave radar sensing module: Installed in the elevator waiting areas on each floor of the building, used to collect real-time three-dimensional coordinates of the passenger flow and send the data to the central controller;

[0054] Multi-modal sensing module: Deployed inside the elevator car, used to obtain the pressure distribution matrix, gas component spectrum, and vibration spectrum data, and maintain data interaction with the central controller by wired or wireless means;

[0055] Timestamp server: Connected to the millimeter-wave radar sensing module and the multi-modal sensing module, used to perform unified time-domain alignment on data from different sources to form a time-synchronized sensing data packet;

[0056] Central controller: Communicates with the timestamp server, millimeter-wave radar sensing module, and multi-modal sensing module, used to comprehensively analyze the passenger flow distribution information, in-car sensing data, and historical elevator riding records, and generate elevator scheduling strategies and abnormal state determination results;

[0057] LoRaWAN communication module: Connected to the central controller, used to receive the abnormal state determination result and send an alarm instruction with a priority identifier to the outside, and at the same time receive the scheduling update command from the central controller;

[0058] Historical database: Interacts with the central controller, used to store elevator riding records, alarm logs, and passenger flow spatio-temporal feature information;

[0059] Door control module: Connected to the central controller, used to adjust the opening and closing speed curve of the door in real time according to the docking time window parameter set in the elevator scheduling strategy and the correction instruction output by the adaptive impedance algorithm.

[0060] Advantages of the present invention:

[0061] In the present invention, by deploying millimeter-wave radar sensing devices in the elevator waiting areas, multi-modal sensing devices inside the car, and building a big data analysis model at the system end, it is possible to obtain multi-source information such as passenger flow distribution, in-car pressure, gas components, and vibration characteristics in real time, and combine it with historical elevator riding records to form a refined dynamic scheduling strategy; this not only significantly shortens the average waiting time of passengers, but also improves the response ability to sudden large passenger flows and the resource allocation efficiency.

[0062] In the present invention, when it is detected that the car load state or the operating environment is abnormal, a hierarchical alarm mechanism can be triggered immediately, and the opening and closing speed of the elevator door machine can be adjusted by an adaptive impedance algorithm, so that the disposal process in an emergency and normal operation can be taken into account simultaneously, thereby maintaining the safety and reliability of the elevator system. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only those of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0064] Figure 1 Schematic diagram of the elevator intelligent control method according to an embodiment of the present invention;

[0065] Figure 2 Schematic diagram of the elevator intelligent control system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] The present invention will be described in detail below with reference to the drawings and specific embodiments. At the same time, it should be noted here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also adopt other alternative methods for implementation; moreover, the drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0067] It should be pointed out that in the specification, when referring to "an embodiment", "embodiments", "exemplary embodiments", "some embodiments", etc., it indicates that the described embodiments may include specific features, structures or characteristics, but not necessarily every embodiment includes such specific features, structures or characteristics. In addition, when combining embodiments to describe specific features, structures or characteristics, implementing such features, structures or characteristics in combination with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the relevant art.

[0068] Generally, the terms can be understood at least in part from their use in the context. For example, at least in part depending on the context, the term "one or more" used herein can be used to describe any feature, structure or characteristic in a singular sense, or can be used to describe a combination of features, structures or characteristics in a plural sense. In addition, the term "based on" can be understood not necessarily to convey a set of exclusive factors, but rather, at least in part depending on the context, to allow for the existence of other factors that may not be explicitly described.

[0069] Such as Figure 1As shown in the figure, an elevator intelligent control method based on the Internet of Things includes the following steps:

[0070] S1: Real-time monitor the three-dimensional coordinate data of the human body in the waiting areas of each floor of the building through a millimeter-wave radar array deployed on each floor of the building, and at the same time collect multi-modal sensing data in the elevator car, including the pressure distribution matrix, gas composition spectrum, and vibration spectrum;

[0071] S2: Input the three-dimensional coordinate data of the human body obtained in S1 into the spatial density clustering algorithm to generate a heat map of the dynamic passenger flow distribution on each floor, and construct a time-space weight matrix based on historical elevator riding records;

[0072] S3: Based on the heat map of the dynamic passenger flow distribution and the time-space weight matrix in S2, generate an elevator dynamic scheduling strategy, and the scheduling strategy includes the target floor sequence and stop time window of each elevator;

[0073] S4: Perform singular value decomposition on the pressure distribution matrix in the car obtained in S1. When it is detected that the variance of the main component exceeds the preset threshold in three consecutive sampling periods, trigger multi-modal data fusion analysis, and combine the gas composition spectrum and vibration spectrum to judge the type of abnormal state in the car;

[0074] S5: Based on the abnormal state determination result in S4, send a hierarchical alarm instruction to the central controller through the LoRaWAN protocol, and at the same time dynamically adjust the stop time window parameter in the elevator scheduling strategy generated in S3 according to the alarm level;

[0075] S6: Embed an adaptive impedance algorithm in the elevator door control module. When the stop time window adjusted in S5 does not match the real-time passenger flow density, automatically correct the opening and closing speed curve of the door machine to ensure that the actual boarding and alighting time deviates from the scheduling strategy by less than ±1.5 seconds.

[0076] S1 specifically includes:

[0077] S11: Install two groups of millimeter-wave radar modules in a diagonal layout on the ceiling of the waiting areas on each floor of the building. Each group of modules contains 3 FMCW radar units operating in the 76-81 GHz band to form a cross-coverage detection area, and output three-dimensional point cloud data including distance, azimuth angle, and elevation angle with a period of 250 ms;

[0078] S12: Perform moving clutter suppression processing on the original point cloud data output in S11, use a recursive least squares filter to eliminate the static background reflection signal, retain the human body movement feature points, and generate a denoised three-dimensional coordinate data set;

[0079] S13: Embedded laying a 16×24 array of piezoresistive thin-film sensors under the elevator car floor, obtaining a pressure distribution matrix at a sampling rate of 100 Hz. The matrix element value represents the pressure value at the corresponding grid position, and the grid size is 8 cm×8 cm;

[0080] S14: Install a tunable diode laser absorption spectrometer (TDLAS) at the center of the car top. Synchronously collect the absorption spectra of three gases, CO, CO2, and CH4, by wavelength scanning. The spectral resolution is set to 0.1 nm, and each scan takes 50 ms;

[0081] S15: Arrange triaxial MEMS accelerometers at the four corners of the car bottom. Extract the vibration spectral features through a Butterworth band-pass filter (passband 1 - 500 Hz), and generate a spectral vector containing the vibration energy in the X / Y / Z axial directions every 200 ms;

[0082] S16: Establish a unified timestamp server to perform time-domain alignment on the three-dimensional coordinate data in S12, the pressure distribution matrix in S13, the gas absorption spectra in S14, and the vibration spectra in S15; Each sub-step of the above S1 constructs a multi-modal input data basis for the elevator control system through the collaborative deployment of millimeter-wave radar, human-computer interaction sensors, and environmental perception devices, ensuring comprehensive and time-effective consistent perception of the flow state, load conditions inside the car, gas changes, and vibration characteristics, and significantly improving the response accuracy and robustness of the system in scheduling optimization and anomaly detection.

[0083] S2 specifically includes:

[0084] S21: Perform a planar projection transformation on the human body three-dimensional coordinate data set obtained in S1, eliminate the height dimension, and generate a two-dimensional coordinate set, expressed as: , where, and respectively represent the horizontal position and vertical position of the th target in the waiting area plane coordinate system, is the total number of people currently monitored;

[0085] S22: Based on the current moment and the historical moment of the previous cycle, dynamically calculate the spatial density clustering radius parameter , the formula is: , where, is the neighborhood radius, is the area of the waiting area, is the number of people monitored at the current moment , is the number of people monitored at the historical moment , is the maximum value operation;

[0086] S23: Cluster the two-dimensional coordinate set using the DBSCAN clustering algorithm and set the minimum neighborhood point number threshold for the core object. The setting formula is: where, is the minimum neighborhood sample number, is defined as above, is the ceiling function, and the output cluster set is expressed as: where, is the number of clusters, represents the th cluster;

[0087] S24: Eliminate the abnormal points with a duration less than 5 seconds in the cluster set, retain the stable clusters, and form the central coordinate set, which is expressed as: where, is the central point coordinate of the th stable cluster, is the number of clusters after filtering;

[0088] S25: Map the central coordinate set to the rasterized waiting elevator area plane, construct the original density distribution map, and perform spatial interpolation using the thin plate spline interpolation algorithm to obtain the interpolated pedestrian flow density value , and then generate a continuous pedestrian flow distribution heat map. Among them, and are plane coordinate variables, is the pedestrian flow density value at the coordinate .

[0089] S2 also includes:

[0090] S26: Extract the historical elevator riding records in the past 30 days and construct a three-dimensional tensor. The expression is: where, represents the elevator riding frequency of the th grid on week , at the th hour, is the total number of grids, , ;

[0091] S27: Perform Tucker decomposition on the tensor to obtain the time factor matrix and the space factor matrix , and then reconstruct their outer product to obtain the dimension-reduced time-space weight matrix. The formula is: where, is the tensor outer product operation, is the weight matrix, which reflects the comprehensive influence weights of different time and space positions on the elevator-taking demand; through the above steps, the real-time modeling of the passenger flow density in the waiting area is realized, and the time-space weight matrix after dimensionality reduction is constructed based on historical elevator-taking behaviors, significantly improving the elevator system's understanding and response capabilities to the characteristics of passenger flow distribution, and providing high-timeliness and high-precision prediction support for the subsequent scheduling strategy formulation.

[0092] S3 specifically includes:

[0093] S31: Obtain the dynamic passenger flow distribution heat map and the time-space weight matrix generated by S2, and determine the total elevator-taking demand of each floor within the current scheduling period based on the two. The total elevator-taking demand represents the predicted value of the number of waiting passengers on the corresponding floor during peak or off-peak passenger flow periods; let the total elevator-taking demand be , and its calculation formula is: , where is the total elevator-taking demand of the th floor, represents the set of coordinate points covered by the waiting area on the corresponding th floor, is the weight value of the time-space corresponding coordinates, representing the product of the corresponding positions;

[0094] S32: Match the elevator-taking demand of each floor with the available capacity of the elevator group, and use the group control algorithm based on minimizing the waiting time of the objective function to establish a priority ranking for all floors and determine the target floor sequence of each elevator. The target floor sequence of the elevator is obtained by optimizing the following objective function: , where is the floor scheduling sequence of the elevator, is the total elevator-taking demand of the th floor, is the maximum number of passengers that the elevator can carry, is the time required for the elevator to reach the th floor from the current position;

[0095] S33: For the target floor sequence of each elevator in S32, combine the aforementioned elevator-taking demand prediction results and allocate a docking time window. This docking time window is calculated based on the expected number of passengers getting on and off and the elevator door opening and closing lag time to ensure effective passenger diversion by the elevator during peak periods;

[0096] S34: Record the corresponding target floor sequences and stop time windows of each elevator in the generated elevator dispatching strategy; through the sub-steps of S3 above, the obtained passenger flow distribution information in step S2 is fully combined with the historical weight data, and the target floor sequences and stop time windows of the elevators are calculated based on real-time demand prediction, significantly improving the elevator's ability to respond to passenger flow fluctuations at different times, shortening the waiting time of passengers, and improving the overall operation efficiency of the elevator.

[0097] The specific process of allocating the stop time window in S33 includes:

[0098] S331: According to the target floor sequences of each elevator determined in S32, count the total demand for taking the elevator on each target floor, and combine with the maximum carrying capacity of the elevator to calculate the proportion coefficient of passengers getting on and off at this floor , and the formula is: , where is the passenger saturation ratio of the th floor, is the total demand for taking the elevator, is the maximum carrying capacity of the elevator;

[0099] S332: According to the door opening and closing lag time of the elevator door machine and the average boarding and alighting time of each passenger, combined with the proportion coefficient , calculate the recommended stop time window of the elevator on the th floor, and the formula is: , where is the stop time window of the th floor, is the fixed time delay for the elevator door to open and close, is the average boarding and alighting time for a single person, represents the expected number of people to board and alight;

[0100] S333: Assign the stop time windows of each target floor to the corresponding elevator's dispatching instruction set in sequence, and synchronize it with the current system time axis to ensure that there is no conflict in the time windows of multiple elevators within the dispatching cycle, forming a final executable stop time scheduling table; through the above sub-steps of S33, based on the demand for taking the elevator, the elevator's carrying capacity, and operation lag factors, the dynamic quantitative allocation of the stop time of each floor of the elevator is realized, ensuring the optimal operation efficiency of the elevator on the premise of meeting the boarding and alighting needs of passengers, and effectively reducing the elevator backlog and delay problems caused by unreasonable stop time settings.

[0101] S4 specifically includes:

[0102] S41: Receive the car interior pressure distribution matrix obtained in S1. First, perform two-dimensional matrix processing on the original pressure values within the same sampling period to obtain matrix data arranged in two dimensions of rows and columns, enabling it to enter the subsequent decomposition and analysis process;

[0103] S42: Perform singular value decomposition on the matrix-processed pressure data. Decompose the pressure matrix into three matrices, which respectively represent the orthogonal information in the row direction, the diagonal singular value information, and the orthogonal information in the column direction. Through this decomposition method, the overall pressure distribution can be disassembled into several principal components to characterize the most significant load change trend inside the car;

[0104] S43: Among the several principal components obtained from the singular value decomposition, select the principal component with the highest strength. Use the variance proportion it occupies as a characterization index, and compare its values within three consecutive sampling periods. When this proportion continuously exceeds the preset threshold, it is judged that the pressure distribution inside the car has an abnormal fluctuation, and multi-modal data fusion analysis is initiated to confirm the true cause of this abnormal fluctuation;

[0105] S44: After determining that an abnormal fluctuation has occurred, based on the gas component spectrum and vibration spectrum data collected in S1, respectively retrieve the characteristic peak positions of the gas absorption spectrum and the concentration of vibration energy in different frequency bands. If there are changes beyond the normal range in terms of gas concentration or vibration intensity, it is judged that there is an abnormality;

[0106] S45: Integrate the pressure anomaly detection results with the analysis information of the gas spectrum and vibration spectrum. According to the comparison of abnormal characteristics, identify the types of abnormal states inside the car, including gas leakage, vibration abnormality, or pressure fluctuation, and output the corresponding abnormal identification to the alarm module of the elevator control center for subsequent processing and recording.

[0107] The steps for performing singular value decomposition are as follows:

[0108] First, for the car interior pressure distribution matrix obtained in S1 Perform sampling and preprocessing, and integrate the original pressure data within the same sampling period into matrix form, denoted as , where and respectively represent the number of rows and columns of the pressure sensor array, is the sampling time;

[0109] Then, perform singular value decomposition on the matrix , and the formula is: , where and are orthogonal matrices respectively, is the diagonal singular value matrix, and denote as the singular values arranged in descending order;

[0110] Finally, select the principal component among the singular values , and calculate the variance of this principal component according to the following formula: , where represents the proportion of the variance occupied by the principal component within the current sampling period ; detect for three consecutive sampling periods. When is satisfied, it is determined that the pressure distribution has an abnormal fluctuation, triggering multimodal data fusion analysis is the preset variance threshold.

[0111] S5 specifically includes:

[0112] S51: Receive the abnormal status result determined by S4, generate an alarm level identifier based on the abnormal type, and digitally encapsulate the alarm information to form an alarm data packet with a priority field and an abnormal type field;

[0113] S52: Send the alarm data packet to the network receiving end of the central controller through the gateway module of the LoRaWAN protocol. The central controller parses the priority field and the abnormal type field in the alarm data packet, maps them to the pre-configured alarm level table, and triggers the corresponding emergency handling strategy;

[0114] Table 1 Alarm Level Mapping

[0115] Alarm type number Abnormal type Priority field Alarm level Suggestions for dispatching strategy adjustment A1 Abnormal gas concentration High Level 1 The elevator on the floor where the alarm occurs immediately stops running, issues a forced broadcast alarm, sets the docking time window to 0, and notifies the maintenance personnel to intervene A2 Abnormal gas concentration Medium Level 2 The docking time window of the elevator on the floor where the alarm occurs is shortened by 50%, a prompt sound is issued, and passengers are restricted from getting on and off A3 Abnormal gas concentration Low Level 3 The elevator operation remains unchanged, and only an abnormal log is generated in the control center B1 Vibration severe shaking High Level 1 The elevator is suspended from dispatching, removed from service, and other elevators on nearby floors are given priority B2 Vibration slight anomaly Medium Level 2 The elevator continues to run, but the docking time window for all floors is uniformly increased by 2 seconds to facilitate the safe boarding and alighting of passengers C1 Load abnormal mutation High Level 1 Stop new passengers from entering, issue a forced load limit reminder in the car, and suspend the up and down dispatching C2 Load fluctuation medium Medium Level 2 The docking time window of the elevator in the current cycle is increased by 3 seconds to avoid overcrowding

[0116] In the above Table 1, the alarm type number is the unique identifier internally recognized by the central controller according to the abnormal type field; the abnormal type is determined by step S4 and includes gas leakage, vibration anomaly, and pressure fluctuation;

[0117] The priority field is the severity determined by S4 and is used to assist in determining the emergency level of the scheduling strategy; the alarm level represents the preset response level of the system, usually divided into level 1 (emergency handling), level 2 (intervention adjustment), and level 3 (recording and warning); the scheduling strategy adjustment suggestion represents the specific scheduling strategy adjustment logic executed by the central controller according to the corresponding level, mainly acting on the adjustment of the "target floor stop time window" parameter of the scheduling strategy generated in S3.

[0118] S53: According to the identified alarm level, retrieve the elevator scheduling strategy generated by S3 from the central controller, read the stop time window parameters of each elevator, and dynamically adjust the time window of the elevator or floor in the alarm state, increasing or shortening the stop time to cooperate with the on-site emergency disposal and minimize the interference to normal operation;

[0119] S53 specifically includes:

[0120] S531: Analyze the alarm data packet received in S52, and read the alarm level identifier and the corresponding abnormal type information, and find the corresponding relationship between the alarm level and the adjustment coefficient in the alarm level mapping table, denoted as ;

[0121] S532: Retrieve the elevator set and floor set in the alarm state from the elevator dispatching strategy generated in S3. For any elevator or any floor, obtain its current stop time window ;

[0122] S533: Calculate the new stop time window in the alarm state ;

[0123] If the adjustment strategy is to increase the stop duration, then execute: ;

[0124] If the adjustment strategy is to shorten the stop duration, then execute: , where is the adjustment coefficient corresponding to the alarm level , and its value is uniquely determined by the alarm level mapping table, is the original stop time window.

[0125] S54: Rewrite the adjusted stop time window parameters into the elevator dispatching instruction set, and send them from the elevator control center to the real-time control unit of the corresponding elevator; through the above S5 sub-steps, after detecting potential safety hazards in the car or related areas, the alarm information can be transmitted to the central controller through the LoRaWAN network in the first time, and the elevator stop time window can be flexibly adjusted in combination with the alarm level, so that the elevator can run orderly during the abnormality and cooperate with subsequent emergency measures, significantly improving the system's abnormality handling efficiency and safety.

[0126] S6 specifically includes:

[0127] S61: After detecting a deviation between the stop time window adjusted by S5 and the real-time passenger flow density, obtain the actual opening and closing durations of the door machine, and compare them with the preset boarding and alighting times in the elevator dispatching strategy to calculate the boarding and alighting time error;

[0128] S62: When the absolute value of the boarding and alighting time error is greater than 1.5 seconds, enter the adaptive impedance correction process, obtain the force information of the door machine during the opening and closing process through the torque sensor of the door machine control module, and adjust the opening and closing speed target curve in combination with the real-time passenger flow situation;

[0129] The specific adjustment steps are as follows:

[0130] S621: Obtain the real-time passenger flow density value of the current elevator stop floor, denoted as , which is derived from the dynamic passenger flow heat map in step S2 and represents the passenger number density per unit waiting area for the elevator; at the same time, read the target passenger flow density value set for this floor from the scheduling strategy, denoted as , representing the expected service passenger flow density corresponding to the current time window;

[0131] S622: Calculate the actual passenger flow deviation rate , which is used to measure the difference between the scheduling and the actual situation. The formula is: , where is the passenger flow density deviation rate, is the real-time passenger flow density, is the density set by the strategy; when is greater than the preset threshold , trigger the speed curve correction mechanism;

[0132] S623: Assume that the standard speed curve for the door opening process of the door machine is , and the speed curve for the door closing process is , and correct them to the target speed curves and ; specifically:

[0133] When (the passenger flow density is higher than expected), execute the strategy of decelerating the door opening and delaying the door closing, and adjust the target curve to: ;

[0134] When (the passenger flow density is lower than expected), execute the strategy of accelerating the door opening and closing the door in advance, and adjust the target curve to: , where and are speed correction factors adaptively adjusted based on , satisfying , and are dynamically adjusted by the control module according to the error to ensure that the total door opening and closing time is within a controllable range;

[0135] S624: Send the corrected target speed curve to the door machine controller, which adjusts the motor torque according to the real-time position feedback to achieve the corresponding speed curve tracking, so that the actual boarding and alighting time error is controlled within seconds.

[0136] S63: Write the corrected speed target curve into the instruction buffer of the door controller, so that the door machine automatically adjusts the torque and motion trajectory during subsequent opening or closing actions, shortening the boarding and alighting time error and maintaining the safe boarding and alighting rhythm of the elevator, ensuring that the deviation between the actual boarding and alighting time and the scheduling strategy is maintained within ±1.5 seconds; through the above S6 sub-step, when the docking time window does not match the actual passenger flow density, the door machine adaptive impedance algorithm can be triggered in a timely manner to dynamically correct the speed and torque of the door machine, realizing precise control of the elevator boarding and alighting duration, taking into account passenger safety and system efficiency, and ensuring that the scheduling strategy can still maintain high reliability and low delay rate under abnormal conditions.

[0137] As Figure 2 shown, an elevator intelligent control system based on the Internet of Things, used to implement the above-mentioned elevator intelligent control method based on the Internet of Things, includes:

[0138] Millimeter-wave radar sensing module: Installed in the waiting areas of each floor of the building, used to collect real-time three-dimensional coordinates of the passenger flow and send the data to the central controller;

[0139] Multimodal sensing module: Deployed inside the elevator car, used to obtain the pressure distribution matrix, gas component spectrum, and vibration spectrum data, and maintain data interaction with the central controller through wired or wireless means;

[0140] Timestamp server: Connected to the millimeter-wave radar sensing module and the multimodal sensing module, used to perform unified time-domain alignment on data from different sources, forming a time-synchronized sensing data packet;

[0141] Central controller: Communicates with the timestamp server, millimeter-wave radar sensing module, and multimodal sensing module, used to comprehensively analyze the passenger flow distribution information, in-car sensing data, and historical elevator riding records, generating elevator scheduling strategies and abnormal state determination results;

[0142] LoRaWAN communication module: Connected to the central controller, used to receive the abnormal state determination result and send an alarm instruction with a priority identifier to the outside, and at the same time receive the scheduling update command from the central controller;

[0143] Historical database: Interacts with the central controller, used to store elevator riding records, alarm logs, and passenger flow spatio-temporal feature information, providing data support for the central controller to perform dynamic scheduling calculations and abnormal analysis;

[0144] Door machine control module: Connected to the central controller, used to adjust the opening and closing speed curve of the door machine in real time according to the docking time window parameters set in the elevator scheduling strategy and the correction instruction output by the adaptive impedance algorithm, ensuring that the deviation between the boarding and alighting process and the scheduling strategy is maintained within the preset threshold.

[0145] The present invention encompasses any alternatives, modifications, equivalent methods, and solutions that are made to the essence and scope of the present invention. For the public to have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention. However, those skilled in the art can also fully understand the present invention without the description of these details. In addition, well-known methods, processes, procedures, components, and circuits are not described in detail to avoid unnecessary confusion to the essence of the present invention.

[0146] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An elevator intelligent control method based on the Internet of Things, characterized in that It includes the following steps: S1: Real-time monitor the three-dimensional coordinate data of the human body in the elevator waiting areas on each floor of the building through a millimeter-wave radar array deployed on each floor, and at the same time collect multi-modal sensing data in the elevator car, including the pressure distribution matrix, gas component spectrum, and vibration spectrum; S2: Input the three-dimensional coordinate data of the human body obtained in S1 into the spatial density clustering algorithm to generate a heat map of the dynamic passenger flow distribution on each floor, and construct a time-space weight matrix based on historical elevator riding records; S3: Generate an elevator dynamic scheduling strategy based on the dynamic passenger flow distribution heat map and the time-space weight matrix in S2. The scheduling strategy includes the target floor sequence and docking time window of each elevator; S4: Perform singular value decomposition on the pressure distribution matrix in the car obtained in S1. When it is detected that the variance of the principal component exceeds a preset threshold within 3 consecutive sampling periods, trigger multi-modal data fusion analysis, and combine the gas component spectrum and vibration spectrum to judge the type of abnormal state in the car; S5: Based on the abnormal state determination result in S4, send a hierarchical alarm instruction to the central controller through the LoRaWAN protocol, and at the same time dynamically adjust the docking time window parameter in the elevator scheduling strategy generated in S3 according to the alarm level; S6: Embed an adaptive impedance algorithm in the elevator door control module. When the docking time window adjusted in S5 does not match the real-time passenger flow density, automatically correct the door opening and closing speed curve to ensure that the actual boarding and alighting time deviates from the scheduling strategy by less than ±1.5 seconds.

2. The intelligent elevator control method based on the Internet of Things according to claim 1, characterized in that The specific content of S1 includes: S11: Install two groups of millimeter-wave radar modules diagonally on the ceiling of the elevator waiting areas on each floor of the building. Each group of modules contains 3 FMCW radar units operating in the 76-81 GHz band, forming an overlapping detection area, and output three-dimensional point cloud data containing distance, azimuth, and elevation angles at a cycle of 250 ms; S12: Perform moving clutter suppression processing on the original point cloud data output in S11. Use a recursive least squares filter to eliminate the static background reflection signal, retain the human body movement feature points, and generate a denoised three-dimensional coordinate data set; S13: Embeddedly lay a 16×24 array of piezoresistive thin-film sensors under the floor of the elevator car, and obtain the pressure distribution matrix at a sampling rate of 100 Hz. The matrix element value represents the pressure value at the corresponding grid position; S14: Install a tunable diode laser absorption spectrometer in the center of the car roof, and synchronously collect the absorption spectra of CO, CO2, and CH4 gases by wavelength scanning; S15: Arrange triaxial MEMS accelerometers at the four corners of the car bottom, extract the vibration spectrum characteristics through a Butterworth band-pass filter, and generate a spectrum vector containing the vibration energy in the X / Y / Z axes every 200 ms; S16: Establish a unified timestamp server to perform time-domain alignment on the three-dimensional coordinate data in S12, the pressure distribution matrix in S13, the gas absorption spectrum in S14, and the vibration spectrum in S15.

3. An elevator intelligent control method based on the Internet of Things according to claim 1, characterized in that, The specific content of S2 includes: S21: Perform plane projection conversion on the three-dimensional coordinate data set of the human body obtained in S1 to eliminate the height dimension and generate a two-dimensional coordinate set; S22: Dynamically calculate the spatial density clustering radius parameter based on the current moment and the historical moment of the previous cycle ; S23: Cluster the two-dimensional coordinate set using the DBSCAN clustering algorithm and output the set of clustering clusters; S24: Remove the abnormal points with a duration less than 5 seconds in the set of clustering clusters, retain the stable clustering clusters, and form a set of central coordinates; S25: Map the set of central coordinates to the rasterized waiting area plane, construct the original density distribution map, and perform spatial interpolation using the thin plate spline interpolation algorithm to obtain the interpolated pedestrian flow density values , and then generate a continuous pedestrian flow distribution heat map.

4. The elevator intelligent control method based on the Internet of Things according to claim 3, characterized in that, S2 further includes: S26: Extract the historical elevator riding records of the past 30 days and construct a three-dimensional tensor. The expression is: , where represents the elevator riding frequency of the -th grid on week , at the -th hour, is the total number of grids, , ; S27: Perform Tucker decomposition on the tensor to obtain the temporal factor matrix and the spatial factor matrix, and then reconstruct its outer product to obtain the reduced-dimensional time-space weight matrix. The formula is as follows: , where is the tensor outer product operation, and is the weight matrix.

5. The elevator intelligent control method based on the Internet of Things according to claim 1, characterized in that, S3 specifically includes: S31: Obtain the dynamic passenger flow distribution heat map and the time-space weight matrix generated by S2, and determine the total elevator demand for each floor within the current scheduling period based on the two; S32: Match the elevator demand of each floor with the available capacity of the elevator group, use the group control algorithm based on minimizing the waiting time of the objective function, establish a priority ranking for all floors, and determine the target floor sequence of each elevator; S33: For the target floor sequence of each elevator in S32, combine the foregoing elevator demand prediction results and allocate a docking time window; S34: Record the corresponding target floor sequence and docking time window of each elevator in the generated elevator scheduling strategy.

6. The elevator intelligent control method based on the Internet of Things according to claim 1, characterized in that S33 specifically includes: S331: According to the elevator target floor sequences determined in S32, count the total elevator ride demand for each target floor, and calculate the up and down passenger ratio coefficient for this floor in combination with the maximum carrying capacity of the elevator. ; S332: Calculate the recommended stop time window of the elevator on the th floor based on the door opening and closing lag time of the elevator door machine, the average boarding and alighting time required for each passenger, and in combination with the proportionality coefficient . The formula is: , where is the stop time window on the th floor, is the fixed time delay for the elevator door to open and close, is the average boarding and alighting time for a single person, and represents the expected number of passengers to board and alight. S333: Assign the stop time windows of each target floor to the scheduling instruction sets of the corresponding elevators in sequence, and synchronize them with the current system timeline to ensure that there are no conflicts in the time windows of multiple elevators within the scheduling cycle, thus forming a final executable stop time scheduling table.

7. The elevator intelligent control method based on the Internet of Things according to claim 1, wherein, S4 specifically includes: S41: Receive the in-car pressure distribution matrix obtained by S1, first perform two-dimensional matrix processing on the original pressure values within the same sampling period to obtain matrix data arranged in two dimensions of rows and columns; S42: Perform singular value decomposition on the matrixed pressure data, decompose the pressure matrix into three matrices, which respectively represent the orthogonal information in the row direction, the diagonal singular value information, and the orthogonal information in the column direction; S43: Among the several principal components obtained by the singular value decomposition, select the principal component with the highest intensity, use the variance ratio it occupies as a characterization index, and compare its values in three consecutive sampling periods. When this ratio continuously exceeds the preset threshold, it is judged that the in-car pressure distribution has an abnormal fluctuation; S44: After determining the abnormal fluctuation, based on the gas component spectrum and vibration spectrum data collected by S1, respectively retrieve the characteristic peak position of the gas absorption spectrum and the concentration of vibration energy in different frequency bands. If there are changes in the gas concentration or vibration intensity that exceed the normal range, it is judged that there is an abnormality; S45: Integrate the pressure anomaly detection results and the analysis information of the gas spectrum and vibration spectrum, and identify the abnormal state type in the car according to the comparison of abnormal characteristics, including gas leakage, vibration anomaly or pressure fluctuation.

8. The elevator intelligent control method based on the Internet of Things according to claim 1, characterized in that S5 specifically includes: S51: Receive the abnormal state result determined by S4, generate an alarm level identifier based on the abnormal type, and digitally encapsulate the alarm information to form an alarm data packet with a priority field and an abnormal type field; S52: Send the alarm data packet to the network receiving end of the central controller through the gateway module of the LoRaWAN protocol. The central controller parses the priority field and the abnormal type field in the alarm data packet, maps them to the pre-configured alarm level table, and triggers the corresponding emergency handling strategy; S53: According to the identified alarm level, retrieve the elevator scheduling strategy generated by S3 from the central controller, read the docking time window parameters of each elevator, and dynamically adjust the time window of the elevator or the floor belonging to the alarm state to cooperate with the on-site emergency disposal; S54: Rewrite the adjusted dock time window parameters back into the elevator dispatching instruction set, and send them from the elevator control center to the real-time control unit of the corresponding elevator.

9. The elevator intelligent control method based on the Internet of Things according to claim 1, characterized in that The specific steps of S6 are as follows: S61: After detecting a deviation between the dock time window adjusted in S5 and the real-time passenger flow density, obtain the actual opening and closing durations of the door machine, compare them with the preset boarding and alighting times in the elevator dispatching strategy, and calculate the boarding and alighting time error. S62: When the absolute value of the boarding and alighting time error is greater than 1.5 seconds, enter the adaptive impedance correction process. Obtain the force information of the door machine during the opening and closing processes through the torque sensor of the door machine control module, and combine the real-time passenger flow situation to adjust the opening and closing speed target curve. S63: Write the corrected speed target curve into the instruction buffer of the door machine controller, so that the door machine automatically adjusts the torque and motion trajectory during the subsequent opening or closing actions, ensuring that the deviation between the actual boarding and alighting time and the dispatching strategy is maintained within ±1.5 seconds.

10. An elevator intelligent control system based on the Internet of Things, which is used to implement an elevator intelligent control method based on the Internet of Things according to any one of claims 1-9, and is characterized in that, It includes: Millimeter-wave radar sensing module: Installed in the elevator waiting areas on each floor of the building, used to collect the three-dimensional coordinates of the real-time passenger flow and send the data to the central controller. Multimodal sensing module: Deployed inside the elevator car, used to obtain the pressure distribution matrix, gas component spectrum, and vibration spectrum data, and maintain data interaction with the central controller through wired or wireless means. Timestamp server: Connected to the millimeter-wave radar sensing module and the multimodal sensing module, used to perform unified time-domain alignment on data from different sources to form a time-synchronized sensing data packet. Central controller: Communicates with the timestamp server, the millimeter-wave radar sensing module, and the multimodal sensing module, used to comprehensively analyze the passenger flow distribution information, in-car sensing data, and historical elevator riding records, and generate elevator dispatching strategies and abnormal state determination results. LoRaWAN communication module: Connected to the central controller, used to receive the abnormal state determination results and send warning instructions with priority identifiers to the outside, and at the same time receive the dispatching update commands from the central controller. Historical database: Interacts with the central controller, used to store elevator riding records, warning logs, and passenger flow spatio-temporal feature information. Door machine control module: Connected to the central controller, used to adjust the opening and closing speed curve of the door machine in real time according to the dock time window parameters set in the elevator dispatching strategy and the correction instructions output by the adaptive impedance algorithm.

Images