Intelligent elevator control system and data processing method

Abstract

The invention discloses an intelligent elevator control system and a data processing method, and the intelligent elevator control system comprises a face recognition module which is used for recognizing the face of an elevator user and outputting a face authentication result; the voice recognition module is used for acquiring voice information of an elevator user, and the voice recognition module is used for converting the voice information into instruction information; the edge computing terminal is used for determining the use permission of an elevator user according to the face authentication result of the face recognition module; and the edge computing terminal controls the elevator to work according to the use permission and the instruction information. Through the face recognition module, the system can recognize and verify the identity of a user, it is ensured that only authorized personnel can use the elevator, and therefore the safety of a building is improved. And the edge computing terminal determines the user permission according to the face recognition result, unauthorized personnel are limited to access a specific floor, and illegal invasion is prevented. And through the language recognition module, the user can control the elevator through a voice instruction, physical contact with elevator buttons is not needed, and convenience is provided especially when objects are carried or hands are inconvenient.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent elevator control system and a data processing method. Background Art

[0002] With the intensification of urbanization, the number of elevators in my country has increased rapidly. Elevators have become an indispensable tool for urban construction. Elevators are frequently used and have a complex user population. Traditional elevators are usually controlled by physical button panels, and users need to directly touch the buttons to select floors.

[0003] However, in traditional elevator control methods, physical buttons are prone to wear and damage due to frequent use, and need to be replaced and maintained regularly, which not only increases operating costs but also affects the normal operation of the elevator. On the other hand, this contact method has potential risks in terms of hygiene and health, especially during epidemics of infectious diseases, where physical contact increases the possibility of virus transmission. Summary of the invention

[0004] In order to overcome the above technical defects, the present invention provides an intelligent elevator control system and a data processing method.

[0005] In order to solve the above problems, the present invention is implemented according to the following technical solutions:

[0006] The present invention provides an intelligent elevator control system, the intelligent elevator control system comprising:

[0007] A face recognition module, which is used for face recognition of elevator users and outputs face authentication results;

[0008] A voice recognition module, the voice recognition module is used to obtain voice information of the elevator user, and the voice recognition module converts the voice information into instruction information;

[0009] The edge computing terminal determines the use authority of the elevator user according to the face authentication result of the face recognition module; the edge computing terminal controls the operation of the elevator according to the use authority and instruction information.

[0010] In combination with the first aspect, the present invention further provides a first specific implementation of the first aspect, specifically, the voice recognition module converts the voice information into a voice control instruction, specifically comprising the following steps:

[0011] Preprocessing the voice information;

[0012] Converting the voice information into text information;

[0013] Extracting a number of keywords from the text information;

[0014] Match several keywords in the preset command vocabulary to obtain command information that adapts to the elevator user's intention.

[0015] In combination with the first aspect, the present invention further provides a second specific implementation of the first aspect, specifically, the intelligent elevator control system further includes:

[0016] A sensor assembly, the sensor assembly comprising a photoelectric sensor, a temperature sensor, a power detection sensor and an infrared sensor;

[0017] A fault detection module is connected to the sensor component, and the fault detection module generates fault alarm information according to the detection signal collected by the sensor component.

[0018] In combination with the first aspect, the present invention further provides a third specific implementation of the first aspect, specifically, the intelligent elevator control system further includes:

[0019] A camera component, the camera component is used to obtain image data of the elevator, the image content of the image data is the internal scene of the elevator car;

[0020] Wherein, the camera component is connected to the edge computing terminal, and the edge computing terminal performs image analysis on the image data.

[0021] In a second aspect, the present invention further provides a data processing method for the intelligent elevator control system as described in the first aspect, characterized in that the edge computing terminal is used to execute the data processing method, and the data processing method includes:

[0022] Acquire image data of the camera assembly, wherein the image content of the image data is an internal scene of an elevator car;

[0023] Screening the image data to obtain target image data;

[0024] Identify whether there is a prohibited object in the target image data, wherein the prohibited object is an object prohibited from entering the elevator;

[0025] When there is a prohibited target, a corresponding warning message is generated and issued.

[0026] In combination with the second aspect, the present invention further provides a first specific implementation manner of the second aspect, specifically, the screening of the image data comprises the following steps:

[0027] Read the current frame graphics data;

[0028] Calculate the optical flow of the current frame image data;

[0029] Determining whether the optical flow is greater than a preset threshold;

[0030] When the optical flow is greater than a preset threshold, the current frame of graphic data is used as target image data and output, and the next frame of image data is read;

[0031] When the optical flow is less than a preset threshold, the next frame of image data is read.

[0032] In combination with the second aspect, the present invention further provides a second specific implementation manner of the second aspect. Specifically, the optical flow is calculated by the following method:

[0033] Establishing the optical flow constraint equation: I x u+I y v+It=0; where I x and I y is the gradient of the image in the x and y directions, u and v are the components of the optical flow in the x and y directions, I t is the change of the image at time t; I x ,I y and I t It can be calculated from the image;

[0034] Establish the matrix equation of 9 pixels in the 3×3 range

[0035] The matrix equation Convert to tensor matrix

[0036]

[0037] According to the tensor matrix, the optical flow (u, v) of the pixel is calculated by accumulating the partial derivatives of the neighborhood pixels in three dimensions and performing matrix operations.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention provides an intelligent elevator control system, which includes: a face recognition module, which is used for face recognition of elevator users and outputs face recognition results; a voice recognition module, which is used to obtain voice information of elevator users and converts the voice information into instruction information; an edge computing terminal, which determines the use authority of the elevator user according to the face recognition result of the face recognition module; and the edge computing terminal controls the operation of the elevator according to the use authority and instruction information.

[0040] The present invention uses a face recognition module, and the system can identify and verify the identity of the user, ensuring that only authorized personnel can use the elevator, thereby improving the safety of the building. The edge computing terminal determines the user's authority based on the face recognition results, restricts unauthorized personnel from accessing specific floors, and prevents illegal intrusion. Through the language recognition module, users can control the elevator through voice commands without physically touching the elevator buttons, which is particularly convenient when carrying items or having difficulty with their hands. By reducing the reliance on physical buttons, the intelligent elevator system can reduce wear and damage caused by frequent use, thereby reducing the frequency and cost of replacement and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The specific embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings, wherein:

[0042] Figure 1 It is a flow chart of a data processing method of the present invention;

[0043] Figure 2 It is a system composition diagram of an intelligent elevator control system of the present invention. DETAILED DESCRIPTION

[0044] The following specifically illustrates the implementation mode of the present invention in conjunction with the accompanying drawings. The embodiments are provided for illustrative purposes only and cannot be understood as limiting the present invention. The accompanying drawings are provided for reference and illustration only and do not constitute a limitation on the scope of patent protection of the present invention, because many changes can be made to the present invention without departing from the spirit and scope of the present invention.

[0045] With the intensification of urbanization, the number of elevators in my country has increased rapidly. Elevators have become an indispensable tool for urban construction. Elevators are characterized by frequent use and complex user groups. Traditional elevators are usually controlled by physical button panels, and users need to directly touch the buttons to select floors. However, in traditional elevator control methods, physical buttons are prone to wear and damage due to frequent use, and need to be replaced and maintained regularly, which not only increases operating costs, but also affects the normal operation of elevators. On the other hand, this contact method has potential risks in terms of hygiene and health, especially during epidemics of infectious diseases, where physical contact increases the possibility of virus transmission.

[0046] For this purpose, refer to Figure 2The present invention provides an intelligent elevator control system, which includes: a face recognition module, which is used for face recognition of elevator users and outputs face recognition results; a voice recognition module, which is used to obtain voice information of elevator users and converts the voice information into instruction information; an edge computing terminal, which determines the use authority of the elevator user according to the face recognition result of the face recognition module; and the edge computing terminal controls the elevator operation according to the use authority and instruction information.

[0047] like Figure 2 As shown, the preferred structure of an intelligent elevator control system shown in the present invention. In the present invention, the intelligent elevator control system includes:

[0048] A face recognition module, which is used for face recognition of elevator users and outputs face authentication results;

[0049] A voice recognition module, the voice recognition module is used to obtain voice information of the elevator user, and the voice recognition module converts the voice information into instruction information;

[0050] The edge computing terminal determines the use authority of the elevator user according to the face authentication result of the face recognition module; the edge computing terminal controls the operation of the elevator according to the use authority and instruction information.

[0051] In a specific implementation, the face recognition module is used to identify the facial features of the elevator user and output the face recognition result. The face recognition module can be installed beside the elevator door for users entering the elevator to perform face recognition. It can also be placed in the elevator car for users in the elevator to perform face recognition.

[0052] In the specific implementation, the face recognition module captures the user's facial image through the camera. The face recognition module uses the face detection algorithm to quickly locate the face in the image, and extracts the facial feature points for subsequent comparison. Through the built-in face comparison database: store the facial feature data of registered users. Compare the recognition result with the data in the database and output the authentication result. The authentication result can be divided into authentication success, authentication failure, or authentication level, which corresponds to different permissions. For example, users are divided into ordinary users, maintenance users, management users, etc.

[0053] In a specific implementation, the face recognition module adopts OpenMV technology, which is a programmable camera module that can realize its logical functions through programming language. Its main function of face recognition is described by local binary pattern features. LBP extracts feature textures through operators that describe local features of images to complete face recognition. Face detection is to match and determine whether the object is in the image through the constructor of OpenMV. Through this function, facial recognition of people inside the elevator can be performed more accurately. OpenMV takes photos of each elevator user, matches the user's facial information with the face model in the database for similarity, and sends an authentication authority signal after a successful match to perform voice control of the elevator operation.

[0054] In the present invention, the voice recognition module converts the voice information into executable instruction information. Specifically, the voice recognition module converts the voice information into voice control instructions, which specifically includes the following steps:

[0055] Step 100: pre-process the voice information.

[0056] In this step, the purpose of preprocessing is to improve the accuracy and robustness of speech recognition. The specific implementation includes:

[0057] Denoising: Reduce background noise using a noise reduction algorithm, such as spectral subtraction or Wiener filtering.

[0058] Normalization: Normalize the speech signal to keep its volume within a certain range.

[0059] Pre-emphasis: Enhance the high-frequency part to compensate for the loss of high-frequency signals during transmission.

[0060] Frame segmentation: Divide the continuous speech signal into shorter frames to facilitate subsequent feature extraction.

[0061] Step 200: Convert the voice information into text information.

[0062] In this step, the speech signal is converted into a processable text format. The features of the speech frame, such as Mel-frequency cepstral coefficients (MFCC), are extracted. A deep learning model (such as the CTC connection time series classification model) or a traditional hidden Markov model (HMM) is used to map acoustic features to phonemes. The text is post-processed in conjunction with a language model to improve the accuracy of the conversion.

[0063] Step 300: extracting a number of keywords from the text information.

[0064] In this step, the goal is to identify key information in the user's instructions. The converted text is segmented to identify individual words or vocabulary. According to the context of elevator control, keywords related to elevator operation are extracted, such as "up", "down", "go", "floor", "open", "close", etc.

[0065] Step 400: Match several keywords in a preset command vocabulary to obtain instruction information that is adapted to the elevator user's intention.

[0066] In the present invention, the user's natural language intention is converted into control instructions that can be understood by the elevator system.

[0067] Command vocabulary construction: Build a vocabulary containing all possible elevator control commands.

[0068] Intent recognition: Identify the user's intent by matching keywords with entries in the command vocabulary.

[0069] Instruction generation: Generate corresponding elevator control instructions based on the recognized intention, such as "go to the 5th floor" is converted into the elevator's floor control instruction.

[0070] In a specific implementation, 4.1 Construction of command vocabulary:

[0071] Thesaurus design: Create a thesaurus containing all possible elevator control commands. This command thesaurus should contain all keywords and phrases related to elevator operation, such as "upstairs", "downstairs", "open door", "close door", "go to X floor", etc.

[0072] Classification label: Assign one or more classification labels to each entry in the command vocabulary. These labels represent the specific operations of the elevator, such as "up", "down", "open door", etc.

[0073] 4.2 Matching of keywords and command lexicon:

[0074] Matching algorithm: Develop or use existing string matching algorithms (such as edit distance algorithm, regular expression matching, fuzzy matching, etc.) to identify the degree of match between the extracted keywords and the entries in the command vocabulary.

[0075] Matching strategy: Determines the matching strategy, such as whether an exact match, partial match, or fuzzy match is required, and how to handle synonyms and polysemous words.

[0076] 4.3 Analysis of user intent:

[0077] Intent recognition: Based on the matching results, identify the user's intent. For example, if the keyword is "fifth floor", the intent may be "go to the fifth floor".

[0078] Contextual analysis: Consider contextual information, such as the elevator’s current floor, time, and user’s historical behavior, to more accurately parse user intent.

[0079] 4.4 Generation of instruction information:

[0080] Instruction construction: Construct elevator control instructions based on the recognized intent. For example, if the user intent is "go to the fifth floor", the generated instruction may be "MOVE_TO_FLOOR(5)".

[0081] Instruction verification: Verify whether the generated instructions comply with the elevator's operating logic and safety rules.

[0082] In a preferred implementation, the intelligent elevator control system further includes:

[0083] A sensor assembly, the sensor assembly comprising a photoelectric sensor, a temperature sensor, a power detection sensor and an infrared sensor;

[0084] A fault detection module is connected to the sensor component, and the fault detection module generates fault alarm information according to the detection signal collected by the sensor component.

[0085] In the present invention, the fault detection fault states are divided into four categories: abnormal elevator running speed, fire alarm fault, power failure fault and elevator door abnormality. When the system detects that the elevator is running abnormally, it will notify the external master control terminal to issue an alarm.

[0086] In the present invention, the edge computing terminal calculates the running speed of the elevator by receiving the elevator position signal and determines whether the elevator running state is abnormal. The running speed of the elevator is determined by the time it takes the elevator to pass through the floor. The average speed calculation formula of the elevator is as follows:

[0087]

[0088] Where d is the floor width; n is the number of floors the elevator passes from the start to the stop state; t i It is the time it takes for the elevator to pass each floor from the start of operation to the stop state.

[0089] Furthermore, the system has added a temperature sensor. When the temperature value is higher than the warning value, the system will send a fire alarm warning. The system's power detection circuit can determine whether the elevator is powered on. When it detects that the elevator is not powered on, it sends a power failure warning. A 12V lithium battery is added as an external power supply to ensure the normal operation of the warning function.

[0090] Preferably, an infrared sensor is added to the elevator door, which is turned on when the elevator is in a stopped state. If the microcontroller fails to detect the infrared sensor after a certain delay, an abnormal fault warning of the elevator door is sent.

[0091] In a preferred implementation, the intelligent elevator control system further includes:

[0092] A camera component, the camera component is used to obtain image data of the elevator, the image content of the image data is the internal scene of the elevator car;

[0093] Wherein, the camera component is connected to the edge computing terminal, and the edge computing terminal performs image analysis on the image data.

[0094] Among them, the edge computing terminal is implemented in the form of hardware and / or software to execute and implement the data processing method described in the present invention.

[0095] In the specific implementation, the camera component is responsible for real-time monitoring of the situation inside the elevator car, capturing passenger behavior, environmental changes, and potential safety hazards (such as fire, abnormal behavior, etc.). The camera should be installed on the top or side of the elevator car to ensure that it can cover the entire interior of the car and the elevator door. Among them, a high-definition camera is preferably used, which supports night vision function to adapt to different lighting conditions. It supports real-time video streaming and can transmit captured video data and image data to the edge computing terminal.

[0096] refer to Figure 1 The present invention provides a flow chart of a data processing method. The data processing method is executed by the above-mentioned elevator safety management system, which can be implemented in the form of hardware and / or software. Figure 1 As shown, the data processing method includes:

[0097] S100: Acquire image data of the camera assembly, where the image content of the image data is an internal scene of an elevator car;

[0098] S200: Screening the image data to obtain target image data;

[0099] S300: Identify whether there is a prohibited object in the target image data, where the prohibited object is an object prohibited from entering the elevator;

[0100] S400: When there is a prohibited target, a corresponding warning message is generated and issued.

[0101] Specifically, the present invention describes in detail each step of the data processing method:

[0102] S100: Acquire image data of the camera assembly, wherein the image content of the image data is an internal scene of an elevator car.

[0103] In the present invention, the purpose of the camera assembly is to capture the real-time video stream inside the elevator car and provide image data for subsequent safety detection.

[0104] S200: Screening the image data to obtain target image data.

[0105] In the present invention, by extracting key image data from the original video stream, unnecessary data processing can be reduced and system efficiency can be improved. It is understandable that the amount of data generated by elevator monitoring 24 hours a day is very large, including large sections of repeated images of empty cars, and it is not necessary to detect whether there are electric cars in all of them. To this end, the present invention provides a technical solution that can effectively reduce the amount of calculation of edge technology terminals by pre-screening these images using the optical flow method. The images are pre-screened based on the optical flow method, and only images that may contain prohibited targets are further detected.

[0106] To this end, the screening of the image data specifically includes the following steps:

[0107] S210: Read the current frame graphic data.

[0108] S220: Calculate the optical flow of the current frame image data.

[0109] In the present invention, the motion changes in the image sequence are detected by calculating the optical flow to determine whether a dynamic event occurs. The motion information of the object can be understood by calculating the optical flow generated between images. In the car, the camera is stationary relative to the car, so it can be considered that the movement of the object in the car is directly mapped to the monitoring video and becomes the optical flow. By calculating the optical flow of the elevator monitoring video, the movement of the object in the car can be obtained.

[0110] In some examples, a suitable optical flow algorithm is selected, such as the Lucas-Kanade method, the Horn-Schunck method, or a deep learning-based optical flow estimation method.

[0111] In a specific implementation, the optical flow is calculated by the following method:

[0112] S221: Establishing the optical flow constraint equation: I x u+I y v+I t =0; where I x and I y is the gradient of the image in the x and y directions, u and v are the components of the optical flow in the x and y directions, I t is the change of the image at time t; I x ,I y and I t It can be calculated from the image.

[0113] S222: Establish the matrix equation of 9 pixels in the 3×3 range

[0114] It can be understood that a linear system of equations is built for each pixel and its 8 surrounding neighbors. For each pixel, a 3×3 neighborhood around it is selected. For each pixel in the neighborhood, a linear equation is constructed using the optical flow constraint equation.

[0115] S223: Transform the matrix equation Convert to tensor matrix

[0116]

[0117] It can be understood that the linear equations of the local neighborhood are extended to the entire image to form a tensor matrix. The specific implementation includes:

[0118] Tensor construction: Combine all the linear equations of the local neighborhood to form a large tensor using matrix A and vector b:

[0119]

[0120] The dimension of A is 3N×3, where N is the number of pixels in the image; the dimension of b is 3N×1.

[0121] S224: Using a matrix based on the tensor, the optical flow (u, v) of the pixel is calculated by accumulating partial derivatives of the neighboring pixel points in three dimensions and performing matrix operations.

[0122] Partial derivative accumulation: For each pixel, accumulate the partial derivatives of all pixels in its neighborhood in the x, y, and time dimensions.

[0123] Matrix operations: Use least squares or other numerical optimization techniques to solve for a tensor using matrix A and vector b:

[0124] The resulting u and v vectors represent the optical flow components of each pixel.

[0125] S230: Determine whether the optical flow is greater than a preset threshold.

[0126] It can be understood that by comparing the optical flow size with the preset threshold, it is determined whether the current frame image contains significant motion changes. The optical flow size threshold is set according to the elevator car scene and experience.

[0127] S240: When the optical flow is greater than a preset threshold, the current frame of graphic data is used as target image data and output, and the next frame of image data is read.

[0128] S250: When the optical flow is less than a preset threshold, read the next frame of image data.

[0129] Specifically, the calculated optical flow is compared with a preset threshold value. If the optical flow size exceeds the threshold value, it is considered that the current frame image contains significant motion changes.

[0130] The purpose of screening images is to filter out images that do not need to be detected, improve detection efficiency, and save detection calculations. When observing the surveillance video of the elevator, it is found that there are long periods of static images with no people. Performing forbidden target detection on such images will cause a lot of resource waste. Therefore, it is necessary to improve detection efficiency and avoid repeated detection by eliminating the images that are continuously static, that is, the images without moving targets.

[0131] S300: Identify whether there is a prohibited object in the target image data, where the prohibited object is an object that is prohibited from entering the elevator.

[0132] In the present invention, prohibited items mainly include electric vehicles, motorcycles, electric vehicle batteries, etc. In the specific implementation, an object recognition algorithm can be used to identify whether there are electric vehicles, motorcycles, and electric vehicle batteries in the image data. The object recognition algorithm can train the model by using a large amount of image data of marked prohibited objects to more accurately identify prohibited objects.

[0133] S400: When there is a prohibited target, a corresponding warning message is generated and issued.

[0134] In specific implementation, the edge computing terminal will transmit the warning information to the monitoring background, and then notify the elevator maintenance personnel and safety management personnel. This information may include the type of warning, the time of occurrence, the specific location of the elevator, etc., so that the management personnel can respond quickly and take appropriate measures.

[0135] In another implementation, while issuing a warning message, the elevator's stop mechanism can also be triggered to prevent the elevator from continuing to operate before the prohibited object is removed. This can be achieved by sending a signal to the elevator control system, which instructs the elevator to stop or safely stop at the nearest floor and open the elevator door.

[0136] In another preferred implementation, the data processing method further includes a door blocking event detection step, and the door blocking event detection step includes:

[0137] S1000: Obtain the door opening detection signal sent by the elevator.

[0138] In specific implementation, the elevator control system will send a door opening detection signal when the door is fully opened. The safety management system receives this signal through the interface with the elevator control system. The interface can be a hardware signal line or a network communication interface to ensure the real-time and accuracy of the signal.

[0139] In elevator technology, the open and closed status of elevator doors is usually detected by sensors on the door. These sensors can be photoelectric sensors, infrared sensors or mechanical switches, etc., which trigger signals when the door is opened. The elevator control system generates a door opening detection signal based on the input of the sensor or the feedback signal of the door motor.

[0140] S2000: The timing starts when a door opening detection signal is obtained and stops when a door closing detection signal from the elevator is obtained.

[0141] After receiving the door open detection signal, the edge computing terminal of the security management system starts a timer. The timer will continue to run until the system receives the door close detection signal, which usually occurs when the door is fully closed. The accuracy of the timer needs to be high enough to ensure the accuracy of time measurement. After receiving the door open detection signal, the security management system starts a timer. The timer will continue to run until the system receives the door close detection signal, which usually occurs when the door is fully closed. The accuracy of the timer needs to be high enough to ensure the accuracy of time measurement.

[0142] It is understood that, similar to the door opening detection, the closed state of the elevator door is usually detected by sensors installed on the door. These sensors may include contact switches, photoelectric sensors or proximity sensors, etc., which trigger signals when the door is fully closed. The elevator control system generates a door closing detection signal based on the input of the sensor or the feedback signal of the door motor.

[0143] In another preferred implementation, when a door opening command is received from the elevator, the timer is reset. The door opening command refers to a command from a user to open the elevator door by pressing a button on a control panel in the elevator car. The edge computing terminal is connected to the control system of the elevator so as to detect the action of the user pressing the door opening button.

[0144] S3000: Determine in real time whether the accumulated time value is greater than a preset time threshold, where the time value is in seconds.

[0145] In the present invention, the safety management system continuously monitors the time value of the timer and compares it with a preset time threshold. This threshold is set by the system according to the elevator usage specifications and safety requirements. For example, if the elevator door does not close within 30 seconds, it is considered that a door blocking event or elevator door failure has occurred.

[0146] S4000: When the time value is greater than a preset time threshold, identify whether there is an obstacle at a preset target position in the image data, where the preset target position is a door of an elevator car.

[0147] In a specific implementation, once the door remains open for longer than a preset threshold, the system triggers the camera component to capture image data at the elevator door. Then, image processing techniques (such as edge detection, color recognition, and deep learning models) are used to analyze the image to identify whether there is an object blocking the door from closing.

[0148] In a specific implementation, identifying whether there is an obstacle at a preset target position in the image data specifically includes:

[0149] S4100: Retrieve the preset cropping template.

[0150] S4200: Crop the image data using a preset cropping template to obtain a cropped image.

[0151] S4300: Identify the outline of an elevator door in the cropped image.

[0152] In the specific implementation, it is identified by the following method:

[0153] The cropped image is converted to a grayscale image and output to reduce computational complexity.

[0154] The guided filter is used to enhance the edge information in the grayscale image in order to maintain the sharpness of the edge. In a specific implementation, a guided filter is applied to process the grayscale image, and the filter can adjust the filter strength according to the local statistical characteristics of the image to enhance the edge information.

[0155] The Canny algorithm is used to detect the elevator door edge in the grayscale image and output the elevator door contour image. The Canny edge detection algorithm is a popular edge detection method that can provide good detection results and robustness.

[0156] S4400: Compare the standard profile information with the elevator door profile to obtain an obstacle detection result.

[0157] In a specific implementation, the standard profile information is usually generated based on the actual profile of the elevator door. In a specific implementation, the detected elevator door profile is compared with the standard profile using an image comparison algorithm, such as template matching, feature point matching, or geometric comparison method. The difference between the elevator door profile and the standard profile is analyzed, including deviations in shape, size, and position.

[0158] In another specific implementation, the contour matching algorithm uses dynamic time warping (DTW) or Hausdorff distance to measure the similarity between two contours. This method can determine the degree of contour matching by calculating the minimum distance between the two.

[0159] Hausdorff distance: Calculate the maximum and minimum distance between the standard contour and the elevator door contour. If the distance is less than the preset threshold, it is considered that there is no obstacle in the elevator door.

[0160] S5000: When an obstacle is identified at the preset target position, a door blocking event detection result is generated.

[0161] It is understood that if the image analysis confirms that there is an obstruction, the system will generate a detection result of the door blocking event. This may include detailed information such as the type, size, and location of the obstruction. The system will record this information in the log, trigger an alarm or notify the management staff to prompt the user's improper behavior.

[0162] S6000: When it is identified that there is no obstacle at the preset target position, an elevator door abnormality result is generated.

[0163] In another specific implementation, if the image analysis shows that there is no obstruction, but the door is still not closed for a long time, the system will generate a detection result of an abnormal elevator door. This may indicate a mechanical failure or other non-blocking problem with the elevator door. The system will also record this information and notify maintenance personnel to check and repair it.

[0164] In the entire door blocking event detection process, the safety management system needs to be closely integrated with the elevator control system and be able to respond and process the state changes of the elevator door in real time. Through this automated detection and response mechanism, the operation safety of the elevator can be effectively improved and safety accidents caused by door blocking events or car door failures can be reduced.

[0165] In a preferred implementation, the data processing method further includes a fire event detection step, and the fire event detection step includes:

[0166] S10: Acquire image data of the camera assembly in real time.

[0167] S20: Preprocessing the image data.

[0168] In one specific implementation, preprocessing includes:

[0169] Denoising: Use a filter (such as a Gaussian filter) to reduce the noise in an image.

[0170] Enhance: Adjust the contrast and brightness of an image to make features more distinct.

[0171] Normalization: Normalize the image to reduce the impact of lighting changes on subsequent detection.

[0172] S30: Identify smoke texture features and flame color features in the image data.

[0173] In a specific implementation, image processing techniques (such as gray level co-occurrence matrix GLCM, Laplacian transform, etc.) are used to extract the texture features of the smoke. On the other hand, the typical color (such as red, yellow) area of ​​the flame is identified through color space conversion (such as from RGB to HSV), and the color histogram or other color features are extracted.

[0174] In the specific implementation, smoke texture feature extraction:

[0175] Gray Level Co-occurrence Matrix (GLCM): Calculates the gray level co-occurrence matrix of an image, which is a statistical method used to describe the spatial relationship of pixels in an image. Extract texture features such as contrast, uniformity, energy, and entropy from the GLCM.

[0176] Laplacian transform: Apply the Laplacian operator to the image for edge detection to identify the outline of smoke. The density and distribution of edges are counted as supplementary features of smoke texture.

[0177] In the specific implementation, flame color feature extraction:

[0178] Color space conversion: Convert the image from RGB color space to HSV color space because color features in HSV space are easier to extract.

[0179] Color threshold segmentation: Set thresholds for typical flame colors (such as red and yellow) to identify possible flame areas in the image. Extract the color histogram within these areas and count the red and yellow pixel ratios.

[0180] Color moment: Calculate the color moment (such as mean, standard deviation) of the flame area to describe the characteristics of the color distribution.

[0181] S40: Combining the smoke texture feature and the flame color feature into a feature vector.

[0182] In the specific implementation, the extracted smoke texture features and flame color features are combined into a feature vector. This vector will be used as the input of the fire detection model, containing all the necessary information for fire probability assessment.

[0183] In one example, four features (contrast, uniformity, energy, and entropy) have been extracted from the smoke texture features, and three features (red pixel ratio, yellow pixel ratio, and color variance) have been extracted from the flame color features, and these features are synthesized into a feature vector.

[0184] S50: Inputting the feature vector into a preset fire detection model to obtain a fire probability value output by the preset fire detection model.

[0185] In a specific implementation, selecting or training a machine learning model suitable for fire detection, such as a support vector machine (SVM), a random forest or a deep learning model, is a common technical means in the art.

[0186] S60: Outputting a fire detection result based on the fire probability value.

[0187] In the specific implementation, a fire probability threshold (e.g. 0.5) is set. When the fire probability value output by the model exceeds this threshold, a fire is considered to be detected. If the fire probability exceeds the threshold, the system will output the fire detection result and trigger the corresponding alarm and emergency response measures. The time, probability value and related image data of the fire detection are recorded for subsequent audit and system optimization.

[0188] For other structures of the intelligent elevator control system and data processing method described in this embodiment, refer to the prior art.

[0189] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Therefore, any modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. An intelligent elevator control system, characterized in that: The intelligent elevator control system comprises: A face recognition module, which is used for face recognition of elevator users and outputs face authentication results; A voice recognition module, the voice recognition module is used to obtain voice information of the elevator user, and the voice recognition module converts the voice information into instruction information; The edge computing terminal determines the use authority of the elevator user according to the face authentication result of the face recognition module; the edge computing terminal controls the operation of the elevator according to the use authority and instruction information.

2. The intelligent elevator control system according to claim 1, characterized in that: The voice recognition module converts the voice information into a voice control instruction, specifically comprising the following steps: Preprocessing the voice information; Converting the voice information into text information; Extracting a number of keywords from the text information; Match several keywords in the preset command vocabulary to obtain command information that adapts to the elevator user's intention.

3. The intelligent elevator control system according to claim 1, characterized in that: The intelligent elevator control system also includes: A sensor assembly, the sensor assembly comprising a photoelectric sensor, a temperature sensor, a power detection sensor and an infrared sensor; A fault detection module is connected to the sensor component, and the fault detection module generates fault alarm information according to the detection signal collected by the sensor component.

4. An intelligent elevator control system according to any one of claims 1 to 3, characterized in that: The intelligent elevator control system also includes: A camera component, the camera component is used to obtain image data of the elevator, the image content of the image data is the internal scene of the elevator car; Wherein, the camera component is connected to the edge computing terminal, and the edge computing terminal performs image analysis on the image data.

5. A data processing method for an intelligent elevator control system according to claim 4, characterized in that: The edge computing terminal is used to execute the data processing method, and the data processing method includes: Acquire image data of the camera assembly, wherein the image content of the image data is an internal scene of an elevator car; Screening the image data to obtain target image data; Identify whether there is a prohibited object in the target image data, wherein the prohibited object is an object prohibited from entering the elevator; When there is a prohibited target, a corresponding warning message is generated and issued.

6. A data processing method according to claim 5, characterized in that: The screening of the image data specifically comprises the following steps: Read the current frame graphics data; Calculate the optical flow of the current frame image data; Determining whether the optical flow is greater than a preset threshold; When the optical flow is greater than a preset threshold, the current frame of graphic data is used as target image data and output, and the next frame of image data is read; When the optical flow is less than a preset threshold, the next frame of image data is read.

7. A data processing method according to claim 6, characterized in that: The optical flow is calculated by the following method: Establishing the optical flow constraint equation: I x u+I y v+I t =0; where I x and I y is the gradient of the image in the x and y directions, u and v are the components of the optical flow in the x and y directions, I t is the change of the image at time t; I x ,I y and I t It can be calculated from the image; Establish the matrix equation of 9 pixels in the 3×3 range The matrix equation Convert to tensor matrix According to the tensor matrix, the optical flow (u, v) of the pixel is calculated by accumulating the partial derivatives of the neighborhood pixels in three dimensions and performing matrix operations.

Images