Search

CN-122009940-A - Elevator door motor control method and system

CN122009940ACN 122009940 ACN122009940 ACN 122009940ACN-122009940-A

Abstract

The invention discloses a control method and a system for an elevator door machine, which are characterized in that a landing door image of an elevator is obtained, the landing door image is subjected to image preprocessing to obtain a landing door corrosion image, the landing door corrosion image is subjected to feature extraction to obtain a feature vector of a corrosion image, the landing door corrosion degree is obtained by classifying the corrosion degree based on K-means clustering and the feature vector, vibration time sequence data of the elevator are obtained, the vibration time sequence data are subjected to data preprocessing to obtain vibration abnormal data, the elevator is subjected to abnormal detection according to the vibration abnormal data to obtain an abnormal detection result, and control parameters of the elevator door machine are determined based on the landing door corrosion degree and the abnormal detection result, so that the safety detection efficiency and accuracy of the elevator are improved, and the reliability of elevator operation control is ensured.

Inventors

  • BAI WEICHAO
  • Cai Ziqiao

Assignees

  • 深圳市经纬纵横科技有限公司

Dates

Publication Date
20260512
Application Date
20260204

Claims (10)

  1. 1. The elevator door machine control method is characterized by comprising the following steps: acquiring a landing door image of an elevator, and carrying out image preprocessing on the landing door image to obtain a landing door corrosion image; Extracting features of the landing door corrosion image to obtain a feature vector of the corrosion image, and classifying the corrosion degree based on K-means clustering and the feature vector to obtain landing door corrosion degree; Acquiring vibration time sequence data of the elevator, and performing data preprocessing on the vibration time sequence data to obtain vibration abnormal data; And carrying out abnormality detection on the elevator according to the vibration abnormality data to obtain an abnormality detection result, and determining control parameters of the elevator door machine based on the landing door corrosion degree and the abnormality detection result.
  2. 2. The elevator door machine control method according to claim 1, wherein the steps of acquiring a landing door image of an elevator, and performing image preprocessing on the landing door image to obtain a landing door corrosion image, comprise: Image graying processing is carried out on the layer door image, and image filtering is carried out on the image subjected to the image graying processing by adopting a mean value filtering algorithm, and the method comprises the following steps: For images Summing and averaging the pixel points of the images, comparing the pixels by using a template, filtering the images to obtain a processing result, and outputting the pixels of the images by using the average filtering: (1) Wherein, the The pixel points of the image are represented, Representing the corresponding pixel point Is used for the display of the display panel, Representing the image after the mean value filtering, Representing a neighborhood set, M representing the sum of pixel points in a neighborhood range; the template adopts an eight-neighborhood template, and the expression of mean filtering is: (2)。
  3. 3. The elevator door machine control method according to claim 2, further comprising: expanding and corroding the filtered image, expanding a background pixel point meeting the conditions in a target adjacent area in the image into a target area, expanding the target area, presetting R2 to represent a two-dimensional integer space, wherein two random sets A and B exist in the R2 space, and defining A to be expanded by B as a mathematical expression The corresponding mathematical expression is: (3) Wherein A is expanded by B, i.e. B is along the A-origin The erosion is the inverse operation of the expansion operation, the binarization erosion operation is the reduction operation of the binarization image to the target area along a certain mathematical form, A is represented by B erosion The corresponding mathematical expression is: (4) when small miscellaneous points exist in the image, after expansion corrosion operation treatment, the white area in the binary image is wholly contracted and filtered, and the small miscellaneous points are phagocytized to finish the image treatment.
  4. 4. The elevator door machine control method according to claim 1, wherein the feature extraction of the landing door corrosion image to obtain a feature vector of the corrosion image, and the classification of the corrosion degree based on K-means clustering and the feature vector of the corrosion image to obtain the landing door corrosion degree, comprises: extracting four characteristic values of the landing door corrosion image to construct a characteristic vector of the elevator landing door corrosion image, wherein the first characteristic value is the area proportion of the total area of each corrosion area to the whole image; the method comprises the steps of obtaining multiple colors through the change and superposition of three color channels of red, green and blue, adopting the average value of the red, green and blue of all pixels of a corrosion area in a layer door image as the remaining three characteristic values, and correspondingly expressing the following expression: (5) Wherein, the The area ratio of the corroded area is represented by M, the pixel number of the image is represented by N, the pixel number of the corroded area is represented by the area of the elevator landing door surface image, 、 、 Respectively representing the average value of red R, green G and blue B components of N pixel points in the corrosion area, wherein the value range is ; Constructing feature vectors representing each of the door erosion images according to the four feature values in the formula (5) : (6)。
  5. 5. The elevator door machine control method of claim 4, wherein the classification of the corrosion degree using K-means clustering comprises: (1) Randomly selecting K corrosion feature vectors in the feature matrix as clustering centers; (2) Respectively calculating the distances between other corrosion feature vectors in the feature matrix and the four clustering centers, and squaring the corrosion distances among the vectors As distance measurement indexes, and respectively taking the characteristic vectors as the categories of the nearest clustering centers, The calculation process of (1) is as follows: Wherein, the , ; Represents the m-th layer door corrosion characteristic vector, Representing an nth door corrosion feature vector; (3) Averaging the distances in each category according to the classified feature vectors, and solving new four cluster centroids; (4) Comparing the new four cluster centroids with the four cluster centroids obtained by the previous calculation, if the cluster centroids change, turning to a process (2), otherwise turning to a process (5); (5) And stopping and outputting the clustering result when the centroid does not change.
  6. 6. The elevator door machine control method according to claim 1, wherein the steps of acquiring vibration time series data of the elevator, and performing data preprocessing on the vibration time series data to obtain vibration abnormality data, include: each variable in the multi-variable time sequence has corresponding value at different time points, and the mutual influence among each variable can be expressed as , , Wherein X represents the entire time series, Is a single time series, T is the number of time series, m is the dimension of each time series; Carrying out data normalization processing on the vibration time sequence data to remove missing values in the elevator time sequence data, and carrying out normalization processing on each dimension of the time sequence data by adopting a maximum-minimum normalization method, wherein the expression is as follows: (7) Wherein, the Is the value of the normalized multivariate time series, And Is the maximum and minimum vector in the time series.
  7. 7. The elevator door machine control method according to claim 6, wherein abnormality detection is performed on the elevator according to the vibration abnormality data to obtain an abnormality detection result, and the control parameters of the elevator door machine are determined based on the landing door corrosion degree and the abnormality detection result, comprising: abnormality detection and presetting of elevator by using transducer model Representing input data having n features and for the input data Performing linear transformation to obtain a query vector Q, a key value vector K and a value vector V, wherein the expression of an attention score matrix obtained through a self-attention mechanism is as follows: (8) Wherein, the Representing the dimension of a key vector sequence, wherein the multi-head attention mechanism is to map a query vector Q, a key value vector K and a value vector V through a fully connected neural network and then input the mapped query vector Q, the mapped key value vector K and the mapped value vector V into a plurality of self-attention mechanism modules; splicing the outputs of the self-attention mechanism modules, integrating the outputs through a full connection layer, wherein the attention score matrixes are respectively shown in formulas (9) and (10): (9) (10) Wherein, the 、 、 、 Parameter matrix representing full connected layer, dimension of key value vector Representing the unified input dimension of the transducer model, h representing the number of self-attention mechanism modules; The transducer model appends sequence information to the raw data using position-coding functions with the expressions (11) and (12), respectively: (11) (12) Wherein, the The location of the data is indicated and, Representing the dimensions of the vector, i representing the dimensions of the position code, For the even dimension in the position coding, For odd dimensions, according to vectors at arbitrary positions Can be expressed as And is not limited by the length of the sequence, a trigonometric function is chosen as the encoding function.
  8. 8. The elevator door machine control method according to claim 6, wherein the result of the abnormality detection is expressed by a size of a sliding window, an original time series is defined as a time window of length L, and a division expression of the sliding window is: (13) if the current time point t is smaller than the window length L, firstly calculating the difference value n between t and L, and making the time sequences of n t moments ; Time series of n times t Time window added to current partition In (1) the original time sequence The sliding window sequence converted to length L is expressed as: 。
  9. 9. The elevator door machine control method of claim 8, wherein the input sequence is expressed as , wherein, When a time sequence with a sliding window length of L is input to the encoder, position coding of a transducer is added to provide time information of the input sequence, and the corresponding expressions are (14) and (15): (14) (15) Wherein, the D refers to the size of the transducer encoder, , ; The feature representations are stacked in layer order to define a final output representation of the encoder The corresponding expression is: (16) Decoder structure for receiving an output representation of an encoder As the input sequences of the same length pass through the sliding step Input to an encoder to predict an output sequence Wherein , , Is the predictive field of view of the algorithm.
  10. 10. An elevator door machine control system, characterized by being applied to an elevator door machine control method as claimed in any one of claims 1 to 9, comprising: the image acquisition unit is used for acquiring a landing door image of the elevator and carrying out image preprocessing on the landing door image to obtain a landing door corrosion image; The feature extraction unit is used for extracting features of the landing door corrosion image to obtain feature vectors of the corrosion image, and classifying the corrosion degree based on K-means clustering and the feature vectors to obtain landing door corrosion degree; The data acquisition unit is used for acquiring vibration time sequence data of the elevator and carrying out data preprocessing on the vibration time sequence data to obtain vibration abnormal data; the abnormality detection unit is used for carrying out abnormality detection on the elevator according to the vibration abnormality data to obtain an abnormality detection result, and determining the control parameters of the elevator door machine based on the landing door corrosion degree and the abnormality detection result.

Description

Elevator door motor control method and system Technical Field The invention belongs to the technical field of intelligent elevator control. In particular to a control method and a control system for an elevator door motor. Background The elevator landing door can be closed in a state just before the elevator car leaves the landing position, personnel at the landing are separated, falling and shearing accidents are prevented, and related statistical data show that the elevator landing door system has high accident frequency and fault frequency, and accidents caused by the elevator landing door system account for more than 80% of all elevator accidents. Due to the influence of high salt, high humidity or high wind speed, the landing door corrosion of the elevator can lead to insufficient structural strength of the landing door, and cause the running failure of the elevator or the accident of special equipment which leads to the falling of passengers. At present, in the aspect of detecting corrosion products of material appearance, the color of the artificial observation products is mainly used, and the detection results of the corrosion degree of the same layer door are often inconsistent due to the fact that people only have differences, experience and other factors. Meanwhile, the judgment of the corrosion degree of the elevator landing door is mainly qualitative, quantitative analysis is not carried out, and the corrosion degree of the elevator landing door is judged very darkly and accurately, so that the normal operation of the elevator is influenced. Therefore, there is a need to provide a door control method for an elevator to solve the above-mentioned problems. Disclosure of Invention In view of the above, the invention provides a method and a system for controlling an elevator door machine, which improve the safety detection efficiency and accuracy of an elevator and ensure the reliability of elevator operation control. In a first aspect. The invention provides an elevator door machine control method, which comprises the following steps: acquiring a landing door image of an elevator, and carrying out image preprocessing on the landing door image to obtain a landing door corrosion image; Extracting features of the landing door corrosion image to obtain a feature vector of the corrosion image, and classifying the corrosion degree based on K-means clustering and the feature vector to obtain landing door corrosion degree; Acquiring vibration time sequence data of the elevator, and performing data preprocessing on the vibration time sequence data to obtain vibration abnormal data; And carrying out abnormality detection on the elevator according to the vibration abnormality data to obtain an abnormality detection result, and determining control parameters of the elevator door machine based on the landing door corrosion degree and the abnormality detection result. As the optimization of the technical scheme, the method for acquiring the landing door image of the elevator and carrying out image preprocessing on the landing door image to obtain the landing door corrosion image comprises the following steps: Image graying processing is carried out on the layer door image, and image filtering is carried out on the image subjected to the image graying processing by adopting a mean value filtering algorithm, and the method comprises the following steps: For images Summing and averaging the pixel points of the images, comparing the pixels by using a template, filtering the images to obtain a processing result, and outputting the pixels of the images by using the average filtering: (1) Wherein, the The pixel points of the image are represented,Representing the corresponding pixel pointIs used for the display of the display panel,Representing the image after the mean value filtering,Representing a neighborhood set, M representing the sum of pixel points in a neighborhood range; the template adopts an eight-neighborhood template, and the expression of mean filtering is: (2)。 As the optimization of the technical scheme, the expansion and corrosion treatment is carried out on the filtered image, the background pixel points meeting the conditions in the target adjacent area in the image are expanded into the target area, the expansion operation of the target area is carried out, R2 is preset to represent a two-dimensional integer space, two random sets A and B exist in the R2 space, and the expansion mathematical expression of A and B is defined as The corresponding mathematical expression is: (3) Wherein A is expanded by B, i.e. B is along the A-origin The erosion is the inverse operation of the expansion operation, the binarization erosion operation is the reduction operation of the binarization image to the target area along a certain mathematical form, A is represented by B erosionThe corresponding mathematical expression is: (4) when small miscellaneous points exist in the image, after expansion corrosion operatio