An image-based elevator component wear-out early warning method and system

Abstract

The application relates to the technical field of image processing, and discloses an elevator part wear-out early warning method and system based on images, which comprises the following steps: determining image data sets by taking pictures of a to-be-detected elevator part from several directions; determining a target part image based on a part image model and an initial part image; establishing a wear mark of the to-be-detected elevator part based on all pixel points; determining a first wear value according to a wear pixel value of the wear pixel point; determining a wear detection area based on the part position; acquiring environment characteristic data corresponding to each environment acquisition point; determining a second wear value based on all the environment characteristic data; comparing a comprehensive wear value with a historical data set; determining whether to adjust the comprehensive wear value according to a comparison result; and issuing a warning to the to-be-detected elevator part according to the target wear value. The application improves the reliability of the warning by processing the image data sets.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically, to an image-based method and system for early warning of wear and tear on elevator components. Background Technology

[0002] With urbanization, single-story houses are gradually being replaced by high-rise buildings, and elevators have become an indispensable tool in high-rise buildings. Their safe operation is directly related to the safety of users. During long-term operation, elevator components inevitably experience wear due to mechanical stress, friction, fatigue, and other factors. This wear not only affects the normal operation of the elevator but also reduces its service life, potentially leading to safety accidents. Currently, the detection and early warning of elevator component wear mainly relies on manual inspection. However, manual inspection is susceptible to subjective judgment based on human experience, resulting in low reliability of detection and early warning. Furthermore, the limited space of the elevator shaft restricts the efficiency and time-consuming nature of inspections, further negatively impacting the detection and early warning system.

[0003] Therefore, it is necessary to design an image-based method and system for early warning of elevator component wear in order to solve the problems existing in the current technology. Summary of the Invention

[0004] In view of this, the present invention proposes an image-based method and system for early warning of wear of elevator components, aiming to solve the problems that manual inspection is susceptible to subjective judgment based on human experience, resulting in low reliability of detection and early warning. Furthermore, the inspection personnel are limited by the space of the elevator shaft, resulting in low inspection efficiency and long inspection cycles, which further have an adverse impact on the detection and early warning.

[0005] In one aspect, the present invention proposes an image-based method for early warning of wear on elevator components, comprising:

[0006] The elevator component to be inspected is identified, and images of the component are taken from several angles to determine an image dataset. The image dataset is then processed to determine an initial part image. Based on the component image model and the initial part image, a target part image is determined.

[0007] Extract all pixels from the target part image, establish wear marks for the elevator component to be inspected based on all pixels, extract wear pixels from the target part image based on the established wear marks, and determine a first wear value based on the wear pixel value of the wear pixels.

[0008] The component location of the elevator component to be inspected is determined, and a wear detection area is determined based on the component location. Several environmental acquisition points are deployed in the wear detection area, and environmental feature data corresponding to each environmental acquisition point is acquired. A second wear value is determined based on all environmental feature data.

[0009] The comprehensive wear value of the elevator component to be inspected is determined based on the first wear value and the second wear value. The comprehensive wear value is compared with the historical dataset. Based on the comparison result, it is determined whether to adjust the comprehensive wear value. When it is determined to adjust the comprehensive wear value, a target wear value is determined based on the historical dataset. An early warning is issued to the elevator component to be inspected based on the target wear value.

[0010] Furthermore, when processing the image dataset to determine the initial part image, the process includes:

[0011] The image dataset is preprocessed, including image denoising and geometric correction;

[0012] Feature points are extracted from the preprocessed image dataset, and the extracted feature points are matched and registered. The registered image datasets are then merged to determine the merged image.

[0013] The merged image is processed to determine the initial part image, and the image processing includes adjusting the image contrast and sharpening the image edges.

[0014] Furthermore, when determining the target part image based on the component image model and the initial part image, the process includes:

[0015] Obtain an image feature set and divide the image feature set into an image training set and an image test set according to a sampling ratio. Pre-select a convolutional neural network model, train the convolutional neural network model according to the image training set, test the trained convolutional neural network model according to the image test set, and determine the component image model according to the test results.

[0016] The initial part image is substituted into the component image model to output the verification value of the initial part image. If the verification value is greater than the verification threshold, the initial part image is determined as the target part image. If the verification value is less than or equal to the verification threshold, the image dataset of the elevator component to be detected is reacquired.

[0017] Furthermore, when establishing wear marks for the elevator components under inspection based on all pixels, the process includes:

[0018] Determine the target qualified part image corresponding to the target part image, extract all pixels of the target part image, and extract all qualified pixels from the target qualified part image, with each pixel corresponding to a qualified pixel;

[0019] Determine the pixel value corresponding to all pixels and the qualified pixel value corresponding to all qualified pixels. Establish wear marks for the elevator parts to be inspected based on the relationship between pixel values ​​and qualified pixel values.

[0020] The wear marks include qualified marks, wear marks, and suspected wear marks.

[0021] Furthermore, when extracting wear pixels from the target part image based on the established wear marks, and determining the first wear value based on the wear pixel values ​​of the wear pixels, the process includes:

[0022] The worn pixels include a first worn pixel and a second worn pixel;

[0023] If the wear mark established for the elevator component to be inspected is the suspected wear mark, the pixel point corresponding to the pixel value greater than the qualified pixel value in the target part image is recorded as the first wear pixel point, and the pixel point corresponding to the pixel value less than the qualified pixel value in the target part image is recorded as the second wear pixel point.

[0024] Extract all first wear pixels and all second wear pixels from the target part image, and determine the first wear value based on all first wear pixels and all second wear pixels.

[0025] Furthermore, when determining the wear detection area based on the location of the component, the process includes:

[0026] Extending outwards from the geometric center of the component's location, a cuboid space with length and width s and height h is formed. Based on the component's location, the cuboid space is reduced to determine the wear detection area.

[0027] Furthermore, several environmental acquisition points are deployed in the wear detection area, and environmental feature data corresponding to each environmental acquisition point is acquired. When determining the second wear value based on all environmental feature data, the process includes:

[0028] The environmental feature data of each environmental acquisition point is compared with the standard environmental feature data. If the environmental feature data of the environmental acquisition point is greater than the standard environmental feature data, the environmental acquisition point is recorded as the first acquisition point. If the environmental feature data of the environmental acquisition point is less than the standard environmental feature data, the environmental acquisition point is recorded as the second acquisition point. If one or more environmental feature data of the environmental acquisition point are greater than the standard environmental feature data, and one or more environmental feature data are less than or equal to the standard environmental feature data, the environmental acquisition point is recorded as the pending acquisition point.

[0029] The first quantity of the first acquisition point and the second quantity of the second acquisition point are counted, and the second wear value is determined based on the first quantity and the second quantity.

[0030] Furthermore, when determining the comprehensive wear value of the elevator component to be tested based on the first wear value and the second wear value, comparing the comprehensive wear value with historical datasets, and determining whether to adjust the comprehensive wear value based on the comparison result, the process includes:

[0031] The comprehensive wear value is the product of the first wear value and the second wear value;

[0032] The historical dataset includes the historical average wear value, several historical average wear values, and several historical wear adjustment factors. Each historical average wear value corresponds to a historical wear adjustment factor. The average wear value is compared with the historical average wear value. If the average wear value is greater than or equal to the historical average wear value, it is determined that the average wear value will not be adjusted and the average wear value is determined as the target wear value. If the average wear value is less than the historical average wear value, it is determined that the average wear value will be adjusted.

[0033] Furthermore, when determining to adjust the overall wear value, a target wear value is determined based on the historical dataset, and an early warning is issued to the elevator component to be inspected based on the target wear value, including:

[0034] The comprehensive wear value is compared with the historical dataset. If a historical comprehensive wear value equal to the comprehensive wear value exists in the historical dataset, the comprehensive wear value is adjusted using the historical wear adjustment factor corresponding to that historical comprehensive wear value. If no historical comprehensive wear value equal to the comprehensive wear value exists in the historical dataset, the historical comprehensive wear value with the highest similarity to the comprehensive wear value is obtained, and the comprehensive wear value is adjusted using the historical wear adjustment factor corresponding to that historical comprehensive wear value.

[0035] The target wear value is the product of the historical wear adjustment factor and the comprehensive wear value;

[0036] The target wear value is compared with the target wear threshold. If the target wear value is greater than or equal to the target wear threshold, it is determined that the elevator component under test has wear and an early warning is issued. If the target wear value is less than the target wear threshold, it is determined that the elevator component under test has no wear and no early warning is issued.

[0037] Compared with existing technologies, the advantages of this invention are as follows: By constructing an image dataset through imaging from multiple angles and combining it with a component image model to accurately process the initial part images, wear marks can be established based on pixel-level analysis. This allows for the precise extraction of wear pixels and the calculation of a first wear value. Changes in pixel values ​​reflect the degree of wear, effectively avoiding subjective errors and improving the reliability of early warnings. Through an automated image acquisition and analysis process, early warnings can be issued for the wear condition of elevator components under inspection. Simultaneously, by deploying several environmental acquisition points in the wear detection area, the relevant environment of the elevator components under inspection can be comprehensively covered. Environmental feature data can be acquired in real time, and a second wear value can be calculated, realizing the monitoring of the operating environment of the elevator components under inspection. A comprehensive wear value is determined by comprehensively considering image analysis and the operating environment. The comprehensive wear value is dynamically adjusted based on historical datasets to determine a target wear value that conforms to the actual operating conditions. This data-driven dynamic early warning mechanism can track the wear condition of the elevator components under inspection in real time, thereby avoiding safety accidents caused by wear accumulation and providing reliable protection for the safe operation of elevators.

[0038] On the other hand, this application also provides an image-based elevator component wear early warning system for applying the above-mentioned image-based elevator component wear early warning method, including:

[0039] The data acquisition module is configured to identify elevator components to be inspected, take pictures of the elevator components to be inspected from several angles to determine an image dataset, process the image dataset to determine an initial part image, and determine a target part image based on the component image model and the initial part image.

[0040] The first processing module is configured to extract all pixels of the target part image, establish wear marks on the elevator parts to be inspected based on all pixels, extract wear pixels of the target part image based on the established wear marks, and determine a first wear value based on the wear pixel value of the wear pixels.

[0041] The second processing module is configured to determine the component location of the elevator component to be inspected, determine the wear detection area based on the component location, deploy several environmental acquisition points in the wear detection area, acquire environmental feature data corresponding to each environmental acquisition point, and determine a second wear value based on all environmental feature data.

[0042] The comprehensive early warning module is configured to determine the comprehensive wear value of the elevator component to be inspected based on the first wear value and the second wear value, compare the comprehensive wear value with a historical dataset, determine whether to adjust the comprehensive wear value based on the comparison result, and when it is determined to adjust the comprehensive wear value, determine a target wear value based on the historical dataset and issue an early warning to the elevator component to be inspected based on the target wear value.

[0043] It is understandable that the above-mentioned image-based elevator component wear early warning method and system have the same beneficial effects, and will not be elaborated further here. Attached Figure Description

[0044] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0045] Figure 1 A flowchart illustrating an image-based elevator component wear early warning method provided in an embodiment of the present invention;

[0046] Figure 2 This is a functional block diagram of an image-based elevator component wear early warning system provided in an embodiment of the present invention. Detailed Implementation

[0047] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0048] In some embodiments of this application, see Figure 1 As shown, an image-based method for early warning of elevator component wear includes:

[0049] S100: Identify the elevator component to be inspected, and take pictures of the elevator component from several angles to determine the image dataset. Process the image dataset to determine the initial part image, and determine the target part image based on the component image model and the initial part image.

[0050] S200: Extract all pixels from the target part image, establish wear marks on the elevator parts to be inspected based on all pixels, extract wear pixels from the target part image based on the established wear marks, and determine the first wear value based on the wear pixel value of the wear pixels.

[0051] S300: Determine the location of the elevator component to be inspected, determine the wear detection area based on the component location, deploy several environmental acquisition points in the wear detection area, acquire the environmental feature data corresponding to each environmental acquisition point, and determine the second wear value based on all the environmental feature data.

[0052] S400: Determine the comprehensive wear value of the elevator component to be inspected based on the first wear value and the second wear value. Compare the comprehensive wear value with the historical dataset. Determine whether to adjust the comprehensive wear value based on the comparison result. When it is determined to adjust the comprehensive wear value, determine the target wear value based on the historical dataset and issue an early warning for the elevator component to be inspected based on the target wear value.

[0053] Specifically, image acquisition devices such as machine vision and infrared cameras are used to capture images of the elevator components to be inspected from multiple angles, including up, down, front, back, left, and right. This ensures that the image dataset completely includes the components, preventing blind spots from a single angle from causing omissions. Because of the use of different angles for shooting, the narrow environment of the elevator shaft may affect the data acquisition process, causing some interference to the image dataset. Therefore, the image dataset is processed to determine the initial component images, eliminating the influence of environmental noise and other interference factors. However, the initial part image after processing may not be complete. Verification is performed based on the part image model to avoid incomplete images affecting the final judgment result, laying the foundation for subsequent wear detection. All pixels in the target part image are extracted to establish wear marks. Changes in pixel values ​​are used to identify potential wear. When a component wears, changes in surface morphology are reflected in differences in pixel values ​​on the image. Through detailed analysis of pixels, the worn pixels in the target part image can be accurately determined, thus quantifying the wear of the elevator component to be inspected and determining the first wear value. The elevator component to be inspected may be located in the car, traction device, and guide device, etc. The actual environment these components come into contact with may not be entirely the same. For example, the car is the carrying space for passengers or goods, while the traction device is located in the elevator shaft. Therefore, the wear detection area is determined based on the component location, and several environmental acquisition points are deployed in the wear detection area. The preferred number of environmental acquisition points is 20, but the specific number can be dynamically set according to the wear detection area. Environmental characteristic data, including temperature, air pressure, and humidity sensors, is acquired using data collection devices such as humidity, air pressure, and dust levels. These are key environmental factors for stable elevator operation. By acquiring this data, the impact on component wear is quantified, thereby determining a second wear value and improving the comprehensiveness and reliability of wear warnings. A comprehensive wear value is determined by combining the first and second wear values ​​and compared with historical datasets. Using historical data as a reference, comparative analysis ensures the stability of the warning system.

[0054] Understandably, image-based detection and early warning avoids errors caused by subjective human judgment, improves the reliability of detection and early warning, and the detection process is not limited by the elevator space, thereby improving detection efficiency and shortening the detection cycle, providing strong protection for the safe operation of elevators and effectively reducing safety accidents caused by wear and tear of parts.

[0055] In some embodiments of this application, when processing an image dataset to determine an initial part image, the process includes: preprocessing the image dataset, the preprocessing including image denoising and geometric correction; extracting feature points from the preprocessed image dataset and matching and registering the extracted feature points; merging the registered image datasets to determine a merged image; and performing image processing on the merged image to determine the initial part image, the image processing including adjusting image contrast and sharpening image edges.

[0056] Specifically, image denoising utilizes filtering algorithms and other techniques to remove noise generated during the shooting process due to factors such as lighting and equipment, improving the clarity of the image dataset and preventing noise from affecting the subsequent identification of the elevator parts to be inspected. Geometric correction uses mathematical transformations to correct image distortion caused by shooting angle and lens distortion, ensuring the accuracy of the size and shape of the elevator parts to be inspected. Feature point extraction employs algorithms such as SIFT or SURF to find representative points (feature points) in the image dataset for subsequent image matching. Feature point matching and registration are performed by matching feature points from images taken from different angles to eliminate perspective differences. Feature point matching can be performed using the RANSAC algorithm, while registration can be performed using weighted average and multi-band fusion techniques. Image merging combines the registered image datasets into an image that can completely present the overall appearance of the elevator parts to be inspected. Image processing is performed on the merged image to determine the initial part image. Adjusting the image contrast enhances the detail differences on the surface of the elevator parts to be inspected, and sharpening the image edges highlights the contours of the elevator parts to be inspected, laying the data foundation for subsequent processing.

[0057] In some embodiments of this application, when determining the target part image based on the component image model and the initial part image, the process includes: acquiring an image feature set and dividing the image feature set into an image training set and an image test set according to a sampling ratio; pre-selecting a convolutional neural network model; training the convolutional neural network model based on the image training set; testing the trained convolutional neural network model based on the image test set; determining the component image model based on the test results; substituting the initial part image into the component image model to output the verification value of the initial part image; if the verification value is greater than the verification threshold, then the initial part image is determined as the target part image; if the verification value is less than or equal to the verification threshold, then the image dataset of the elevator component to be detected is reacquired.

[0058] Specifically, an image feature set is acquired, which includes key data such as texture features (metallic luster and processing patterns, etc.), color features (color distribution), and size features (length, width, height, and spacing, etc.) of the elevator components to be detected. The image feature set is divided into an image training set and an image test set according to a sampling ratio, typically 7:3, to ensure that both sets contain diverse data to improve the model's generalization ability. A convolutional neural network (CNN) model is pre-selected. This CNN model contains multiple convolutional layers, pooling layers, and fully connected layers, designed to capture complex relationships in the data. The image training set is used to train the CNN model, and the image test set is used to test and evaluate the trained CNN model. Test metrics include accuracy, F1 score, and recall to measure the model's performance.

[0059] Understandably, during each training iteration, the model attempts to learn patterns and relationships in the data to improve its prediction or classification capabilities. If the test value of the currently trained convolutional neural network (CNN) model is lower than that of the previously trained model, it indicates a decline in model performance. In this case, the learning rate needs to be reduced by decreasing the magnitude of the gradient change. This is achieved by adjusting the learning rate using cosine annealing and continuing training until the test value is greater than or equal to that of the previously trained model. This helps the model stably approach the global optimum. If the test value of the currently trained CNN model is greater than or equal to that of the previously iterated model, it indicates that the model's performance has stabilized and improved. At this point, iterative training can be stopped, and the currently trained CNN model can be designated as the component image model. The component image model can learn and distinguish key features of images. Since the component image model is based on the CNN model, the CNN model can output the image verification probability (i.e., the verification value of the initial component image) using the softmax function. Based on the verification value, it dynamically determines whether to reacquire the image dataset of the elevator components to be detected, ensuring the stability and reliability of the target component images.

[0060] In some embodiments of this application, when establishing wear marks on elevator parts to be inspected based on all pixels, the process includes: determining a target qualified part image corresponding to a target part image; extracting all pixels from the target part image; extracting all qualified pixels from the target qualified part image, with each pixel corresponding to a qualified pixel; determining the pixel value corresponding to all pixels; determining the qualified pixel value corresponding to all qualified pixels; and establishing wear marks on the elevator parts to be inspected based on the relationship between pixel values ​​and qualified pixel values. The wear marks include qualified marks, wear marks, and suspected wear marks.

[0061] Specifically, the target qualified part image is obtained based on qualified elevator parts to be inspected, specifically determined by whether they are unopened and in use or have provided certificates of conformity. Each pixel corresponds to a qualified pixel, ensuring the correspondence between pixel values ​​and qualified pixel values. When establishing wear marks for elevator parts to be inspected based on the relationship between pixel values ​​and qualified pixel values, if all pixel values ​​are equal to qualified pixel values, a qualified mark is established for the elevator parts to be inspected. A qualified mark indicates that the elevator parts to be inspected have not experienced wear and can continue to be used without issuing a warning. If all pixel values ​​are not equal to qualified pixel values, a wear mark is established for the elevator parts to be inspected. A wear mark indicates that the elevator parts to be inspected have experienced wear and need to be stopped from use and a warning is issued directly. If one or more pixel values ​​are not equal to qualified pixel values, a suspected wear mark is established for the elevator parts to be inspected. A suspected wear mark indicates that the elevator parts to be inspected may have experienced wear and need to be analyzed, ensuring the adaptability and stability of the wear warning system.

[0062] In some embodiments of this application, when extracting wear pixels from a target part image based on established wear marks and determining a first wear value based on the wear pixel values ​​of the wear pixels, the process includes: the wear pixels include first wear pixels and second wear pixels; if the wear mark established for the elevator component to be inspected is a suspected wear mark, the pixel corresponding to the pixel value greater than the qualified pixel value in the target part image is recorded as the first wear pixel, and the pixel corresponding to the pixel value less than the qualified pixel value in the target part image is recorded as the second wear pixel; all first wear pixels and all second wear pixels in the target part image are extracted, and the first wear value is determined based on all first wear pixels and all second wear pixels.

[0063] Specifically, all first worn pixels and all second worn pixels are extracted. The first maximum worn pixel value and the first minimum worn pixel value among all first worn pixels are determined. The second maximum worn pixel value and the second minimum worn pixel value among all second worn pixels are determined. The difference between the first maximum worn pixel value and the first minimum worn pixel value is determined as the first difference. The difference between the second maximum worn pixel value and the second minimum worn pixel value is determined as the second difference. The product of the first difference and the second difference is determined as the first wear value.

[0064] Understandably, when elevator components under inspection show potential wear, changes in surface morphology are reflected in pixel values ​​on the image. Using acceptable pixel values ​​as a benchmark, pixels are categorized into first-wear pixels and second-wear pixels. First-wear pixels with values ​​greater than the acceptable value correspond to surface protrusions and material buildup on the inspected elevator components, while second-wear pixels with values ​​less than the acceptable value correspond to surface depressions and material peeling. By extracting all two types of wear pixels and determining their maximum and minimum wear pixel values, a first difference and a second difference are calculated. The product of the first and second differences is then used to determine the first wear value. By using the difference in pixel values ​​to quantify the degree of wear, and comprehensively considering the range and relative differences of pixel value variations, the actual wear situation can be reflected more comprehensively and meticulously, thus issuing timely warnings and ensuring the safe operation of the elevator.

[0065] In some embodiments of this application, when determining the wear detection area based on the component location, the method includes: extending outwards from the geometric center of the component location to form a cuboid space with a length and width of s and a height of h, and then reducing the cuboid space based on the component location to determine the wear detection area.

[0066] Specifically, when reducing the size of a cuboid space, the reduction is dynamically determined based on the component's location. When a component is located within the elevator car (potentially at the top, middle, or bottom), the wear detection area is ultimately determined within the car's interior during the outward extension process, without extending beyond the car's perimeter. For example, if the component is at the top of the car, the outward extension forms a cuboid space. When the extension height h reaches the bottom of the car, the area encompassing the car's interior is designated as the wear detection area, and other areas are reduced. The same principle applies to the middle and bottom of the car. When a component is outside the car (inside the elevator shaft), the outward extension process defines the area covering the elevator shaft's interior as the wear detection area, while other areas are reduced. This ensures the adaptability and flexibility of dynamically determining the wear detection area based on the component's location, thereby dynamically detecting the component's environment and improving the comprehensiveness and stability of the early warning system.

[0067] In some embodiments of this application, several environmental acquisition points are deployed in the wear detection area, and environmental feature data corresponding to each environmental acquisition point is acquired. When determining the second wear value based on all environmental feature data, the method includes: comparing the environmental feature data of each environmental acquisition point with standard environmental feature data; if the environmental feature data of the environmental acquisition point is greater than the standard environmental feature data, the environmental acquisition point is recorded as the first acquisition point; if the environmental feature data of the environmental acquisition point is less than the standard environmental feature data, the environmental acquisition point is recorded as the second acquisition point; if one or more environmental feature data of the environmental acquisition point are greater than the standard environmental feature data, and one or more environmental feature data are less than or equal to the standard environmental feature data, the environmental acquisition point is recorded as the pending acquisition point; the first quantity of the first acquisition point and the second quantity of the second acquisition point are counted; and the second wear value is determined based on the first quantity and the second quantity.

[0068] Specifically, 20 environmental acquisition points are preferred, and the exact number can be dynamically set according to the wear detection area. Environmental characteristic data, including temperature, humidity, air pressure, and dust, are acquired using humidity sensors, temperature sensors, and air pressure sensors. These are environmental factors essential for the stable operation of the elevator. Standard environmental characteristic data are determined according to the user manuals of the elevator components under test. The product of the first and second quantities is determined as the second wear value. A higher second wear value indicates a greater deviation of the elevator components from the standard environment, and a higher probability of wear. Determining the second wear value based on the first and second quantities ensures the comprehensiveness and reliability of the early warning system for elevator components.

[0069] In some embodiments of this application, when determining the comprehensive wear value of the elevator component to be tested based on the first wear value and the second wear value, comparing the comprehensive wear value with a historical dataset, and determining whether to adjust the comprehensive wear value based on the comparison result, the following steps are taken: the comprehensive wear value is the product of the first wear value and the second wear value; the historical dataset includes the historical average comprehensive wear value, several historical comprehensive wear values, and several historical wear adjustment factors, and each historical comprehensive wear value corresponds to a historical wear adjustment factor; the comprehensive wear value is compared with the historical average comprehensive wear value; if the comprehensive wear value is greater than or equal to the historical average comprehensive wear value, it is determined that the comprehensive wear value will not be adjusted, and the comprehensive wear value is determined as the target wear value; if the comprehensive wear value is less than the historical average comprehensive wear value, it is determined that the comprehensive wear value will be adjusted.

[0070] Specifically, multiplying the first wear value (image analysis result) with the second wear value (environmental factor influence) is essentially a comprehensive analysis of both the presence of wear and environmental adaptation. When both the first and second wear values ​​are high, the product amplifies the wear assessment of the elevator component under inspection. Conversely, when one is low, it can appropriately reduce the false alarm rate. For example, if the first wear value is low but the second wear value is high (due to harsh environmental conditions), the product will still indicate a high wear risk. The historical average wear value represents the average wear level of similar components. Using it as a benchmark reflects the group characteristics of the elevator components under inspection. If the current average wear value is lower than the historical average wear value, it may mean that the wear characteristics are not obvious, there are individual differences in the components, and the potential impact of environmental factors has not yet fully manifested. Therefore, it needs to be adjusted. By introducing the historical average wear value as a reference benchmark, the accuracy and automation of adjusting the average wear value are improved, the interference and error of human judgment are reduced, and the adaptability to different conditions and the comprehensiveness of the early warning are enhanced.

[0071] In some embodiments of this application, when determining to adjust the comprehensive wear value, the method of determining the target wear value based on the historical dataset and issuing an early warning for the elevator component to be inspected based on the target wear value includes: comparing the comprehensive wear value with the historical dataset; if there is a historical comprehensive wear value in the historical dataset that is equal to the comprehensive wear value, adjusting the comprehensive wear value with the historical wear adjustment factor corresponding to the historical comprehensive wear value; if there is no historical comprehensive wear value in the historical dataset that is equal to the comprehensive wear value, obtaining the historical comprehensive wear value with the highest similarity to the comprehensive wear value, and adjusting the comprehensive wear value with the historical wear adjustment factor corresponding to the historical comprehensive wear value; the target wear value is the product of the historical wear adjustment factor and the comprehensive wear value; comparing the target wear value with the target wear threshold; if the target wear value is greater than or equal to the target wear threshold, determining that the elevator component to be inspected has wear and issuing an early warning; if the target wear value is less than the target wear threshold, determining that the elevator component to be inspected has no wear and not issuing an early warning.

[0072] Specifically, by determining the degree of matching between the comprehensive wear value and historical datasets, when identical historical comprehensive wear values ​​are found, these data can be directly used to determine the historical wear adjustment factor, thus ensuring the reliability and consistency of the adjustment. If multiple identical historical comprehensive wear values ​​exist, the average of the historical wear adjustment factors corresponding to each historical comprehensive wear value is used to adjust the comprehensive wear value, with the target wear value being the product of this average and the comprehensive wear value. For cases where the comprehensive wear value does not perfectly match the historical comprehensive wear value, the corresponding historical wear adjustment factor is determined by identifying the historical comprehensive wear value with the highest similarity. Similarity can be determined through Euclidean distance or cosine similarity, etc. Determining the corresponding historical wear adjustment factor based on the historical comprehensive wear value with the highest similarity can effectively address changes in the adjustment. Through data-driven automated adjustment, reliance on human experience and intuition is reduced, lowering the uncertainty and detection risks brought about by human judgment. This improves the automation and accuracy of early warning for elevator components under inspection, and by comprehensively utilizing a large amount of historical data, rich reference information is provided for the adjustment of the comprehensive wear value. The target wear threshold is used to distinguish whether there is wear on the elevator parts to be inspected. The target wear threshold can be dynamically set. By comparing the target wear value with the target wear threshold, and taking into account the results of image analysis, the influence of environmental factors and historical data, the risks of low efficiency and long inspection cycle of manual inspection are avoided, and the accuracy and reliability of wear warning for the elevator parts to be inspected are improved.

[0073] In summary, the beneficial effects of this invention are as follows: By constructing an image dataset through imaging from multiple angles and combining it with a component image model to accurately process the initial part images, wear marks can be established based on pixel-level analysis. This allows for the precise extraction of wear pixels and the calculation of a first wear value. Changes in pixel values ​​reflect the degree of wear, effectively avoiding subjective errors and improving the reliability of early warnings. Through an automated image acquisition and analysis process, early warnings can be issued for the wear condition of elevator components under inspection. Simultaneously, by deploying several environmental acquisition points in the wear detection area, the relevant environment of the elevator components under inspection can be comprehensively covered. Environmental feature data can be acquired in real time, and a second wear value can be calculated, realizing the monitoring of the operating environment of the elevator components under inspection. By comprehensively considering image analysis and the operating environment, a comprehensive wear value is determined, and the comprehensive wear value is dynamically adjusted based on historical datasets to determine a target wear value that conforms to the actual operating conditions. This data-driven dynamic early warning mechanism can track the wear condition of the elevator components under inspection in real time, thereby avoiding safety accidents caused by wear accumulation and providing reliable protection for the safe operation of elevators.

[0074] In another preferred embodiment based on the above embodiments, see [reference] Figure 2As shown, this embodiment provides an image-based elevator component wear early warning system for applying the above-described image-based elevator component wear early warning method, including:

[0075] The data acquisition module is configured to identify the elevator parts to be inspected, take pictures of the elevator parts from several angles to determine the image dataset, process the image dataset to determine the initial part image, and determine the target part image based on the part image model and the initial part image.

[0076] The first processing module is configured to extract all pixels from the target part image, establish wear marks on the elevator parts to be inspected based on all pixels, extract wear pixels from the target part image based on the established wear marks, and determine a first wear value based on the wear pixel value of the wear pixels.

[0077] The second processing module is configured to determine the component location of the elevator parts to be inspected, determine the wear detection area based on the component location, deploy several environmental acquisition points in the wear detection area, acquire environmental feature data corresponding to each environmental acquisition point, and determine the second wear value based on all the environmental feature data.

[0078] The comprehensive early warning module is configured to determine the comprehensive wear value of the elevator component to be inspected based on the first wear value and the second wear value, compare the comprehensive wear value with the historical dataset, and determine whether to adjust the comprehensive wear value based on the comparison result. When it is determined to adjust the comprehensive wear value, a target wear value is determined based on the historical dataset, and an early warning is issued for the elevator component to be inspected based on the target wear value.

[0079] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0080] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0081] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0082] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An image-based method for early warning of wear on elevator components, characterized in that, include: The elevator component to be inspected is identified, and images of the component are taken from several angles to determine an image dataset. The image dataset is then processed to determine an initial part image. Based on the component image model and the initial part image, a target part image is determined. Extract all pixels from the target part image, establish wear marks for the elevator component to be inspected based on all pixels, extract wear pixels from the target part image based on the established wear marks, and determine a first wear value based on the wear pixel value of the wear pixels. The component location of the elevator component to be inspected is determined, and a wear detection area is determined based on the component location. Several environmental acquisition points are deployed in the wear detection area, and environmental feature data corresponding to each environmental acquisition point is acquired. A second wear value is determined based on all environmental feature data. The comprehensive wear value of the elevator component to be inspected is determined based on the first wear value and the second wear value. The comprehensive wear value is compared with the historical dataset. Based on the comparison result, it is determined whether to adjust the comprehensive wear value. When it is determined to adjust the comprehensive wear value, a target wear value is determined based on the historical dataset. An early warning is issued to the elevator component to be inspected based on the target wear value. The first wear value is determined based on the wear pixel value of the worn pixel, including: The pixels in the target part image that have a pixel value greater than the qualified pixel value are recorded as the first wear pixel, and the pixels in the target part image that have a pixel value less than the qualified pixel value are recorded as the second wear pixel; all the first wear pixels and all the second wear pixels in the target part image are extracted, and the first wear value is determined based on all the first wear pixels and all the second wear pixels; When determining the second wear value based on all environmental characteristic data, it includes: The environmental feature data of each environmental acquisition point is compared with the standard environmental feature data. If the environmental feature data of each environmental acquisition point is greater than the standard environmental feature data, the environmental acquisition point is recorded as the first acquisition point. If the environmental feature data of each environmental acquisition point is less than the standard environmental feature data, the environmental acquisition point is recorded as the second acquisition point. If one or more environmental feature data of an environmental acquisition point are greater than the standard environmental feature data, and one or more environmental feature data are less than or equal to the standard environmental feature data, the environmental acquisition point is recorded as the pending acquisition point. The first quantity of the first acquisition point and the second quantity of the second acquisition point are counted, and the second wear value is determined based on the first quantity and the second quantity.

2. The image-based elevator component wear early warning method according to claim 1, characterized in that, When processing the image dataset to determine the initial part image, the process includes: The image dataset is preprocessed, including image denoising and geometric correction; Feature points are extracted from the preprocessed image dataset, and the extracted feature points are matched and registered. The registered image datasets are then merged to determine the merged image. The merged image is processed to determine the initial part image, and the image processing includes adjusting the image contrast and sharpening the image edges.

3. The image-based elevator component wear early warning method according to claim 2, characterized in that, Determining the target part image based on the part image model and the initial part image includes: Obtain an image feature set and divide the image feature set into an image training set and an image test set according to a sampling ratio. Pre-select a convolutional neural network model, train the convolutional neural network model according to the image training set, test the trained convolutional neural network model according to the image test set, and determine the component image model according to the test results. The initial part image is substituted into the component image model to output the verification value of the initial part image. If the verification value is greater than the verification threshold, the initial part image is determined as the target part image. If the verification value is less than or equal to the verification threshold, the image dataset of the elevator component to be detected is reacquired.

4. The image-based elevator component wear early warning method according to claim 3, characterized in that, When establishing wear marks for the elevator components under inspection based on all pixels, the process includes: Determine the target qualified part image corresponding to the target part image, extract all pixels of the target part image, and extract all qualified pixels from the target qualified part image, with each pixel corresponding to a qualified pixel; Determine the pixel value corresponding to all pixels and the qualified pixel value corresponding to all qualified pixels. Establish wear marks for the elevator parts to be inspected based on the relationship between pixel values ​​and qualified pixel values. The wear marks include qualified marks, wear marks, and suspected wear marks.

5. The image-based elevator component wear early warning method according to claim 4, characterized in that, The worn pixels include a first worn pixel and a second worn pixel.

6. The image-based elevator component wear early warning method according to claim 5, characterized in that, When determining the wear detection area based on the location of the component, the following is included: Extending outwards from the geometric center of the component's location, a cuboid space with length and width s and height h is formed. Based on the component's location, the cuboid space is reduced to determine the wear detection area.

7. The image-based elevator component wear early warning method according to claim 1, characterized in that, When determining the comprehensive wear value of the elevator component to be inspected based on the first wear value and the second wear value, comparing the comprehensive wear value with historical datasets, and determining whether to adjust the comprehensive wear value based on the comparison result, the process includes: The comprehensive wear value is the product of the first wear value and the second wear value; The historical dataset includes the historical average wear value, several historical average wear values, and several historical wear adjustment factors. Each historical average wear value corresponds to a historical wear adjustment factor. The average wear value is compared with the historical average wear value. If the average wear value is greater than or equal to the historical average wear value, it is determined that the average wear value will not be adjusted and the average wear value is determined as the target wear value. If the average wear value is less than the historical average wear value, it is determined that the average wear value will be adjusted.

8. The image-based elevator component wear early warning method according to claim 7, characterized in that, When determining to adjust the overall wear value, a target wear value is determined based on the historical dataset. When issuing a warning to the elevator component under inspection based on the target wear value, the process includes: The comprehensive wear value is compared with the historical dataset. If a historical comprehensive wear value equal to the comprehensive wear value exists in the historical dataset, the comprehensive wear value is adjusted using the historical wear adjustment factor corresponding to that historical comprehensive wear value. If no historical comprehensive wear value equal to the comprehensive wear value exists in the historical dataset, the historical comprehensive wear value with the highest similarity to the comprehensive wear value is obtained, and the comprehensive wear value is adjusted using the historical wear adjustment factor corresponding to that historical comprehensive wear value. The target wear value is the product of the historical wear adjustment factor and the comprehensive wear value; The target wear value is compared with the target wear threshold. If the target wear value is greater than or equal to the target wear threshold, it is determined that the elevator component under test has wear and an early warning is issued. If the target wear value is less than the target wear threshold, it is determined that the elevator component under test has no wear and no early warning is issued.

9. An image-based elevator component wear early warning system, used to apply the image-based elevator component wear early warning method as described in any one of claims 1-8, characterized in that, include: The data acquisition module is configured to identify elevator components to be inspected, take pictures of the elevator components to be inspected from several angles to determine an image dataset, process the image dataset to determine an initial part image, and determine a target part image based on the component image model and the initial part image. The first processing module is configured to extract all pixels of the target part image, establish wear marks on the elevator parts to be inspected based on all pixels, extract wear pixels of the target part image based on the established wear marks, and determine a first wear value based on the wear pixel value of the wear pixels. The second processing module is configured to determine the component location of the elevator component to be inspected, determine the wear detection area based on the component location, deploy several environmental acquisition points in the wear detection area, acquire environmental feature data corresponding to each environmental acquisition point, and determine a second wear value based on all environmental feature data. The comprehensive early warning module is configured to determine the comprehensive wear value of the elevator component to be inspected based on the first wear value and the second wear value, compare the comprehensive wear value with a historical dataset, determine whether to adjust the comprehensive wear value based on the comparison result, and when it is determined to adjust the comprehensive wear value, determine a target wear value based on the historical dataset and issue an early warning to the elevator component to be inspected based on the target wear value. The determination of the first wear value based on the wear pixel value of the worn pixel includes: The pixels in the target part image that have a pixel value greater than the qualified pixel value are recorded as the first wear pixel, and the pixels in the target part image that have a pixel value less than the qualified pixel value are recorded as the second wear pixel; all the first wear pixels and all the second wear pixels in the target part image are extracted, and the first wear value is determined based on all the first wear pixels and all the second wear pixels; When determining the second wear value based on all environmental characteristic data, it includes: The environmental feature data of each environmental acquisition point is compared with the standard environmental feature data. If the environmental feature data of each environmental acquisition point is greater than the standard environmental feature data, the environmental acquisition point is recorded as the first acquisition point. If the environmental feature data of each environmental acquisition point is less than the standard environmental feature data, the environmental acquisition point is recorded as the second acquisition point. If one or more environmental feature data of an environmental acquisition point are greater than the standard environmental feature data, and one or more environmental feature data are less than or equal to the standard environmental feature data, the environmental acquisition point is recorded as the pending acquisition point. The first quantity of the first acquisition point and the second quantity of the second acquisition point are counted, and the second wear value is determined based on the first quantity and the second quantity.

Images