Intelligent slag salvaging system fault diagnosis method

Abstract

The invention provides an intelligent slag salvaging system fault diagnosis method, and relates to the technical field of fault diagnosis, and the method comprises the steps: collecting the temperature, pressure and flow parameters of a slag salvaging system in real time through a sensor, collecting the image information of a key part through an image recognition technology, carrying out the filtering processing of the parameters collected by the sensor, and obtaining a fault diagnosis result; performing enhancement processing on the image information, extracting temperature change rate, pressure fluctuation amplitude, flow change trend, color feature information, texture feature information and shape feature information from the processed data, fusing the extracted feature information based on a preset fault diagnosis model, and diagnosing whether the slag salvaging system has a fault based on a fusion result. Generating an early warning signal based on the diagnosis result; according to the method, various sensor data and image recognition technologies are comprehensively utilized, and the efficiency and the accuracy of fault diagnosis of the slag salvaging system are improved through reasonable data processing and feature extraction methods and establishment of an effective fault diagnosis model.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a fault diagnosis method for an intelligent slag dredging system. Background Art

[0002] In the industrial production process, the slag removal system is an indispensable part of many industrial facilities (such as the boiler system of thermal power plants, etc.). It is mainly responsible for cleaning the waste slag generated during the combustion process to ensure the normal operation of the system. With the continuous development of industrial automation technology, the slag removal system has gradually changed from traditional manual operation to automatic control.

[0003] Early fault diagnosis of slag dredging systems mainly relied on manual inspections and experience-based judgments. Operators needed to regularly inspect the slag dredging machine and its related equipment, and judge whether there was a fault by observing the operating status, sound, temperature and other intuitive phenomena of the equipment. This method was not only inefficient, but also required a high level of experience from the operator, and was prone to misjudgment and missed judgments. With the development of sensor technology, more and more sensors are being applied to slag dredging systems to monitor the operating parameters of the system in real time, such as temperature sensors, pressure sensors and flow sensors. These sensors can obtain a large amount of real-time data, providing a richer source of information for fault diagnosis. However, there are still limitations to fault diagnosis based solely on single sensor data. For example, temperature anomalies may be caused by a variety of reasons, such as equipment wear, blockage, heat dissipation problems, etc. It is difficult to accurately determine the specific cause and location of the fault using only temperature sensor data.

[0004] Therefore, the field urgently needs an intelligent slag removal system fault diagnosis method to solve the problems existing in the prior art, such as low efficiency and poor accuracy of the diagnosis method, large limitations of single sensor data diagnosis, difficulty in accurately determining the cause and location of the fault, and image recognition technology being greatly affected by interference in complex industrial environments and inaccurate feature extraction. Summary of the invention

[0005] The present invention provides an intelligent slag dredging system fault diagnosis method, which aims to comprehensively utilize multiple sensor data and image recognition technology, solve the problems existing in the above-mentioned prior art through reasonable data processing and feature extraction methods, and establish an effective fault diagnosis model, and improve the efficiency and accuracy of slag dredging system fault diagnosis.

[0006] The present invention provides an intelligent slag dredging system fault diagnosis method, comprising:

[0007] Step 1: Use sensors to collect temperature, pressure, and flow parameters in the slag removal system in real time, and use image recognition technology to collect image information of key parts;

[0008] Step 2: filter the parameters collected by the sensor and enhance the image information;

[0009] Step 3: extracting temperature change rate, pressure fluctuation amplitude, flow rate change trend, and color feature, texture feature, and shape feature information from the processed data;

[0010] Step 4: Based on the preset fault diagnosis model, the extracted feature information is fused, and based on the fusion result, it is diagnosed whether the slag removal system has a fault;

[0011] Step five: Generate a warning signal based on the diagnosis results.

[0012] According to an intelligent slag dredging system fault diagnosis method provided by the present invention, in step one, the temperature, pressure and flow parameters are obtained by temperature sensors, pressure sensors and flow sensors installed in the slag dredging system; image acquisition is performed by a camera to obtain the appearance image of the key parts, that is, the image information.

[0013] According to an intelligent slag removal system fault diagnosis method provided by the present invention, in step 2, the process of filtering the parameters collected by the sensor adopts a median filtering algorithm, and the process of enhancing the image information adopts a histogram equalization algorithm.

[0014] According to an intelligent slag removal system fault diagnosis method provided by the present invention, in step three, the temperature change rate extraction process is:

[0015] Get the temperature values ​​and corresponding time intervals at adjacent time points;

[0016] Calculate the temperature change rate T at the current moment by the first calculation formula rate ;

[0017] The first calculation formula is:

[0018]

[0019] Among them, T rate is the temperature change rate, T t+Δt and T t are the temperatures at time t+Δt and t respectively, Δt is the time interval, n is the number of time points between time t-Δt and t, and s is a smoothing factor used to balance the current and historical temperature changes and reduce the impact of noise; the smoothing factor s is determined by trial and error or based on historical data analysis.

[0020] According to an intelligent slag removal system fault diagnosis method provided by the present invention, in step three, the process of extracting the pressure fluctuation amplitude is:

[0021] Get the current pressure value and the pressure value in the previous preset period;

[0022] Calculate the pressure fluctuation amplitude P at the current moment by the second calculation formula std ;

[0023] The second calculation formula is:

[0024]

[0025] Among them, P std is the pressure fluctuation amplitude; P i is the pressure value at the current moment i, is the average pressure in the preset period, and n is the total number of time points corresponding to the preset period; P max is the maximum pressure value within the preset period, P min is the minimum pressure value within the preset period, P avg is the ideal pressure value rated by the system; ω is the pressure stability weight, which is used to reflect the system's requirements for pressure stability and adjust the focus of pressure fluctuation assessment; the pressure stability weight ω is determined based on empirical settings or optimization algorithms.

[0026] According to an intelligent slag removal system fault diagnosis method provided by the present invention, in step three, the process of extracting the flow change trend is:

[0027] Obtaining the flow value at the current moment and the flow value within a preset time limit; the preset time limit is less than the preset period in the extraction process of the pressure fluctuation amplitude;

[0028] The current flow rate change trend Q is calculated by the third calculation formula tre ;

[0029] The third calculation formula is:

[0030]

[0031] Among them, is the flow change trend; Q is the flow value at the current time t, is the derivative of the flow rate at the current moment with respect to time; δt is the preset time limit, that is, a preset time interval; m is the number of time points between time t-δt and t, is the time derivative of the flow at the jth time point; q is the smoothness of flow change, which is used to integrate current and historical flow changes and reduce the impact of sudden flow changes; the flow change smoothness q is determined based on system requirements or data-driven methods.

[0032] According to an intelligent slag removal system fault diagnosis method provided by the present invention, in step three, the process of extracting the color feature, texture feature and shape feature information includes:

[0033] The color feature information is represented by a color histogram, specifically by dividing the color space of the image information into m intervals, counting the number of pixels in each interval, and obtaining a color histogram, i.e., the color feature information H a ={h 1 ,h 2 , ..., h m}, where h i is the number of pixels in the i-th color interval;

[0034] The texture feature information is extracted using a gray level co-occurrence matrix, specifically by calculating the gray level co-occurrence matrix of the image information, i.e., the texture feature information P b ={i, j, d, θ}, where i and j are grayscale values, d is the distance between pixels, and θ is the direction;

[0035] The shape feature information is represented by Fourier descriptors, specifically, for a shape contour C in the image information, it is represented as a series of complex coordinates (x k ,y k ), k = 0, 1, ..., N-1; where N is the number of points on the contour C; the contour is subjected to discrete Fourier transform, and the transformation formula is:

[0036]

[0037] The Fourier descriptor, namely the shape feature information F(u), is obtained through the transformation formula.

[0038] According to an intelligent slag removal system fault diagnosis method provided by the present invention, in step 4, the preset fault diagnosis model is:

[0039]

[0040] Among them, G is the fusion parameter index, which represents the result of fusing the temperature change rate, pressure fluctuation amplitude, flow change trend, and color, texture, and shape features through a linear combination; α, β, γ, δ a ,∈ b , θ are weight coefficients, which are used to adjust the importance of different features in the fault diagnosis process. Their values ​​are determined based on the training and optimization of the fault diagnosis model based on historical data. The result of the linear combination is mapped to between 0 and 1 through 1 / (1+exp{-x}) to obtain the fault probability value, that is, the fusion parameter index G;

[0041] When the fusion parameter index G is closer to 1, it indicates that the possibility of failure in the slag removal system is greater; when the fusion parameter index G is closer to 0, it indicates that the possibility of normal operation of the slag removal system is greater.

[0042] According to an intelligent slag dredging system fault diagnosis method provided by the present invention, the process of diagnosing whether the slag dredging system has a fault based on the fusion result includes:

[0043] Preset fault threshold G E ; If the fusion parameter index G is greater than or equal to the fault threshold G E When the current slag removal system is judged to have a fault; if the fusion parameter index G is greater than the fault threshold G E , it is judged that there is no fault in the current slag removal system.

[0044] Compared with the prior art, the beneficial effects of this application are:

[0045] This application comprehensively utilizes sensor data such as temperature, pressure, flow, and color, texture, and shape features obtained by image recognition technology, which can describe the operating status of the slag removal system from multiple angles and avoid the limitations of a single data source; for sensor data, a feature extraction formula of smoothing factor, pressure stability weight, and flow change smoothness parameter is adopted to better adapt to the dynamic changes of the system and reduce the influence of noise and outliers, thereby improving the accuracy of feature extraction; a fault diagnosis model based on the idea of ​​logistic regression is established, and weight coefficients are set through training results, various feature information is fused, and the fusion results are mapped to between 0 and 1 to represent the fault probability, which can better capture the complex relationship between features and improve the accuracy of fault diagnosis.

[0046] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0047] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0049] Figure 1 It is a flow chart of an intelligent slag dredging system fault diagnosis method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0051] Embodiment 1:

[0052] The present invention provides an intelligent slag removal system fault diagnosis method, please refer to Figure 1 ,include:

[0053] Step 1: Use sensors to collect temperature, pressure, and flow parameters in the slag removal system in real time, and use image recognition technology to collect image information of key parts;

[0054] Step 2: filter the parameters collected by the sensor and enhance the image information;

[0055] Step 3: extracting temperature change rate, pressure fluctuation amplitude, flow rate change trend, and color feature, texture feature, and shape feature information from the processed data;

[0056] Step 4: Based on the preset fault diagnosis model, the extracted feature information is fused, and based on the fusion result, it is diagnosed whether the slag removal system has a fault;

[0057] Step five: Generate a warning signal based on the diagnosis results.

[0058] The principle of the above embodiment is: the physical parameters and image information of the slag dredging system are collected respectively through sensors and image recognition technology, and the collected data are filtered and enhanced to improve the data quality; then a variety of feature information is extracted from the processed data, including dynamic features related to temperature, pressure, and flow, and color, texture and shape features of the image; then these different types of feature information are fused using a preset fault diagnosis model, and it is judged whether there is a fault in the slag dredging system based on the fusion result, and if there is a fault, a warning signal is generated; the principle is to comprehensively and accurately judge the operating status of the slag dredging system by integrating various data features, and timely discover potential faults.

[0059] The beneficial effects of the above embodiments are: it is able to collect various parameters and image information of the slag dredging system in real time, and timely grasp the system operation status; filter the sensor data and enhance the image information, thereby improving the accuracy and reliability of the data; comprehensively diagnose faults based on various information such as temperature change rate, pressure fluctuation amplitude, flow change trend, and color, texture and shape characteristics of the image, thereby improving the accuracy and comprehensiveness of the diagnosis; generate early warning signals based on the diagnosis results, which can promptly remind the staff to handle the faults, avoid the expansion of faults, and reduce equipment damage and production losses.

[0060] In order to further optimize the above embodiment, in step one, the temperature, pressure and flow parameters are obtained by temperature sensors, pressure sensors and flow sensors installed in the slag removal system; image acquisition is performed by a camera to obtain the appearance image of the key parts, that is, the image information.

[0061] It should be noted that the present embodiment uses specific sensors and equipment to accurately obtain information on different aspects of the slag dredging system for subsequent analysis and fault diagnosis; the temperature sensor detects the temperature at a specific position in the slag dredging system and uses its physical properties to convert temperature changes into measurable data such as electrical signals, thereby obtaining temperature parameters; the pressure sensor also converts the pressure value into a readable signal according to the effect of the pressure in the system to obtain pressure parameters; the flow sensor measures the flow of the fluid in the pipeline, such as flow velocity, flow rate, etc., to provide the system with flow parameters; and the camera uses the principle of optical imaging to shoot key parts of the slag dredging system to obtain its appearance image information, which can reflect the state, shape, color and other characteristics of the key parts;

[0062] Specifically, select the appropriate type of temperature sensor, such as thermocouple, thermal resistor, etc., and install it according to the characteristics of the slag dredger system and the location that needs to be monitored; for example, it can be installed in the key transmission parts of the slag dredger, slurry pipelines, etc. to monitor the temperature changes of these parts; during installation, ensure that the sensor is in good contact with the measured part to ensure the accuracy of the measurement; at the same time, consider the protection and durability of the sensor to avoid being affected by the harsh environment in the slag dredger system;

[0063] According to the pressure distribution in the slag dredging system, select the appropriate installation location of the pressure sensor; generally, it can be installed at the outlet of the slurry pump, key nodes of the pipeline, etc.; when installing, pay attention to the connection method and sealing performance of the sensor to ensure the accuracy and stability of pressure measurement;

[0064] According to the flow characteristics of the fluid in the slag dredging system, select the appropriate flow sensor type, such as electromagnetic flowmeter, ultrasonic flowmeter, etc.; when installed in the slurry pipeline, ensure that the installation position of the sensor complies with the principles of fluid mechanics to avoid factors such as eddy currents that affect measurement accuracy;

[0065] Furthermore, it is first necessary to determine the key parts in the slag removal system, such as the scraper, chain, transmission device, etc. of the slag removal machine; select a suitable location near the key parts to install the camera to ensure that the appearance image of the key parts can be clearly captured; the installation position of the camera should take into account factors such as light conditions and viewing angle range to obtain high-quality images; the angle and position of the camera can be adjusted using equipment such as a fixed bracket or a pan / tilt head; set the camera acquisition parameters such as resolution and frame rate according to actual needs; a higher resolution can provide clearer image details, but it will also take up more storage space and processing resources; the frame rate setting should be determined based on the system's operating speed and monitoring requirements to ensure that changes in key parts can be captured in a timely manner.

[0066] In order to further optimize the above embodiment, in step 2, the process of filtering the parameters collected by the sensor adopts a median filtering algorithm, and the process of enhancing the image information adopts a histogram equalization algorithm.

[0067] It should be noted that median filtering is a nonlinear digital filtering technology; its basic principle is to replace the value of a point in a digital image or signal with the median of the values ​​of each point in a neighborhood of the point; for parameters collected by sensors (such as temperature, pressure, flow, etc.), median filtering can effectively remove noise interference; because noise usually appears as random outliers, and median filtering can suppress these outliers to a certain extent by taking the median, retaining values ​​that are closer to the real signal;

[0068] The implementation method is as follows: determine the filter window size: first determine the window size of the median filter; the selection of the window size depends on the characteristics of the signal and the intensity of the noise; generally speaking, the larger the window, the stronger the filtering effect, but it will also cause the loss of signal details; for the sensor parameters of the slag removal system, the appropriate window size can be determined through experiments or experience; for the parameter sequence collected by the sensor, traverse each data point in turn; for each data point, take the data points in the surrounding window with it as the center; then sort these data points and take the median value as the result of filtering for this point; repeat the above steps until all data points are processed to obtain the filtered sensor parameter sequence;

[0069] The purpose of histogram equalization is to adjust the grayscale distribution of the image so that the grayscale of the image is more evenly distributed in the entire grayscale range, thereby enhancing the contrast and details of the image; its basic principle is to redistribute the grayscale of the image according to the histogram information of the image so that the probability of each grayscale appearing in the image is roughly equal; this can make the originally darker or brighter areas in the image clearer and improve the visual effect of the image;

[0070] The implementation method is as follows: first, the histogram of the input image is calculated, that is, the frequency of occurrence of each gray level in the image is counted; the cumulative distribution function is calculated based on the histogram, and the CDF represents the ratio of the number of pixels with a gray level less than or equal to a certain value to the total number of pixels in the image; according to the cumulative distribution function, the gray level of each pixel in the image is redistributed; the specific method is to map the original gray level of each pixel to a new gray level so that the new gray level is more evenly distributed in the entire gray range; according to the redistributed gray levels, an enhanced image is generated;

[0071] By performing median filtering on the parameters collected by the sensor and performing histogram equalization enhancement on the image information, the quality of the data and the visual effect of the image can be improved, providing more accurate and reliable information for subsequent fault diagnosis.

[0072] In order to further optimize the above embodiment, in step 3, the temperature change rate extraction process is:

[0073] Get the temperature values ​​and corresponding time intervals at adjacent time points;

[0074] Calculate the temperature change rate T at the current moment by the first calculation formula rate ;

[0075] The first calculation formula is:

[0076]

[0077] Among them, T rate is the temperature change rate, T t+Δt and T t are the temperatures at time t+Δt and t respectively, Δt is the time interval, n is the number of time points between time t-Δt and t, and s is the smoothing factor, which is used to balance the current and historical temperature changes and reduce the impact of noise; the smoothing factor s is determined by trial and error or based on historical data analysis;

[0078] The extraction process of pressure fluctuation amplitude is:

[0079] Get the current pressure value and the pressure value in the previous preset period;

[0080] Calculate the pressure fluctuation amplitude P at the current moment by the second calculation formula std ;

[0081] The second calculation formula is:

[0082]

[0083] Among them, P std is the pressure fluctuation amplitude; P i is the pressure value at the current moment i, is the average pressure in the preset period, and n is the total number of time points corresponding to the preset period; P max is the maximum pressure value within the preset period, P min is the minimum pressure value within the preset period, P avg is the ideal pressure value rated by the system; ω is the pressure stability weight, which is used to reflect the system's requirements for pressure stability and adjust the focus of pressure fluctuation assessment; the pressure stability weight ω is determined based on experience or optimization algorithm;

[0084] The process of extracting traffic change trend is as follows:

[0085] Obtain the flow value at the current moment and the flow value within a preset time limit; the preset time limit is less than the preset period of time in the extraction process of the pressure fluctuation amplitude;

[0086] The current flow rate change trend Q is calculated by the third calculation formula tre ;

[0087] The third calculation formula is:

[0088]

[0089] Among them, is the flow change trend; Q is the flow value at the current time t, is the derivative of the flow rate at the current moment with respect to time; δt is the preset time limit, that is, a preset time interval; m is the number of time points between time t-δt and t, is the time derivative of the flow at the jth time point; q is the smoothness of flow change, which is used to integrate current and historical flow changes and reduce the impact of sudden flow changes; the flow change smoothness q is determined based on system requirements or data-driven methods.

[0090] It should be noted that the smoothing factor can adjust the weight of the current temperature change rate and the average temperature change rate over the past period of time; it allows the calculation of the temperature change rate to take into account both the current temperature change situation and the historical temperature change trend, avoiding instability and misjudgment caused by relying solely on the current instantaneous change; there are various noise interferences in the temperature data, and the smoothing factor can play a certain smoothing role, reducing the impact of noise on the temperature change rate calculation, making the result more reliable; according to the different working conditions of the slag removal system and its sensitivity to temperature changes, the smoothing factor can be adjusted to meet specific fault diagnosis needs; for example, when the system is more sensitive to temperature changes, the smoothing factor can be appropriately reduced to respond to temperature changes more quickly; and when a more stable temperature change rate is required, the smoothing factor can be increased;

[0091] Specifically, the value is obtained as follows:

[0092] Trial and error method: You can start with an initial value, such as 0.5, and then adjust this value according to the results of fault diagnosis. If you find that the current temperature change rate is too sensitive to the fault diagnosis results, you can reduce it appropriately to increase the weight of the average temperature change rate over the past period of time. If you find that the response to temperature changes is not timely enough, you can increase the value. By trying different values ​​many times and observing the changes in the diagnosis results, find a more appropriate smoothing factor.

[0093] Based on data analysis: collect a large amount of temperature data and corresponding fault conditions, and determine a suitable smoothing factor range through data analysis; calculate the temperature change rate under different smoothing factors, and compare it with the actual fault conditions, and select the smoothing factor value that can better distinguish between normal and fault conditions;

[0094] The pressure stability weight determines the relative importance of the standard deviation part and the ratio of the pressure range to the average pressure in the pressure fluctuation amplitude calculation. By adjusting this weight, the system can focus more on the stability of the pressure (standard deviation) or the range of pressure variation (ratio to the average pressure) according to the characteristics and needs of the system. Different slag removal systems have different requirements for pressure stability. For example, some systems pay more attention to the long-term stability of the pressure. In this case, the pressure stability weight can be increased to make the fault diagnosis focus more on the standard deviation of the pressure. For some systems that are more sensitive to the pressure variation range, the weight can be appropriately reduced to pay more attention to the ratio of the pressure range to the average pressure. Reasonable setting of the pressure stability weight can improve the accuracy of pressure fault diagnosis. By balancing the different characteristics of pressure, the pressure fluctuation can be more comprehensively evaluated, so as to more accurately determine whether the system has pressure-related faults.

[0095] Specifically, the value is obtained as follows:

[0096] Setting the empirical value: Based on the experience of domain experts, the importance of pressure stability is evaluated and an initial weight value is given. For example, if experts believe that the standard deviation of pressure is relatively more important in fault diagnosis, the initial weight value can be set to , indicating that the standard deviation is given more importance.

[0097] Optimization algorithm: Use optimization algorithms in machine learning, such as genetic algorithms or particle swarm optimization algorithms; use the calculation formula of pressure fluctuation amplitude as the objective function, and optimize by continuously adjusting the value of weights to minimize the prediction error; in the optimization process, a large amount of pressure data and known fault conditions are required as training data;

[0098] The flow change smoothness parameter can adjust the weight of the flow change rate at the current moment and the average flow change rate over the past period of time; it allows the calculation of the flow change trend to take into account both the current flow change situation and the historical flow change trend, making the evaluation of flow changes more comprehensive and accurate; sudden changes may occur in the flow data, and the smoothness parameter can play a certain smoothing role, reducing the impact of sudden flow changes on the flow change trend calculation and avoiding misjudgment caused by instantaneous flow changes; different slag removal systems have different flow change characteristics, such as stable flow, periodic flow changes or random flow fluctuations; by adjusting the flow change smoothness parameter, the fault diagnosis method can be better adapted to various flow change situations and improve the detection capability of flow-related faults;

[0099] Specifically, the value is obtained as follows:

[0100] Based on actual needs: Determine the initial value based on the specific operating characteristics of the slag removal system and the response requirements to flow changes; if the system is sensitive to sudden changes in flow, a smaller smoothness parameter may be required, for example, to pay more attention to the flow change rate at the current moment; if the system pays more attention to the stable change trend of flow, a larger value can be selected;

[0101] Data-driven method: Collect flow data and corresponding system status in different time periods, and determine the appropriate smoothness parameter by analyzing the characteristics of flow changes; calculate the flow change trend under different values ​​and compare it with the actual system status, and select the parameter value that can better reflect the change of system status.

[0102] In order to further optimize the above embodiment, in step 3, the process of extracting color feature, texture feature and shape feature information includes:

[0103] The color feature information is represented by a color histogram. Specifically, the color space of the image information is divided into m intervals, and the number of pixels in each interval is counted to obtain a color histogram, i.e., the color feature information H. a ={h 1 ,h 2 , ..., h m}, where h i is the number of pixels in the i-th color interval;

[0104] Texture feature information is extracted using the gray-level co-occurrence matrix, specifically by calculating the gray-level co-occurrence matrix of the image information, i.e., the texture feature information P b ={i, j, d, θ}, where i and j are grayscale values, d is the distance between pixels, and θ is the direction;

[0105] The shape feature information is represented by Fourier descriptors. Specifically, for a shape contour C in the image information, it is represented as a series of complex coordinates (x k ,y k ), k = 0, 1, ..., N-1; where N is the number of points on the contour C; the contour is subjected to discrete Fourier transform, and the transformation formula is:

[0106]

[0107] The Fourier descriptor, namely the shape feature information F(u), is obtained through the transformation formula.

[0108] It should be noted that the color histogram is a method of statistically analyzing the color distribution in an image. By dividing the color space of an image into multiple intervals (here, m intervals), and then counting the number of pixels in each interval, a histogram representing color features can be obtained. This histogram can reflect the proportion of different colors in the image.

[0109] First, determine the color space to be divided and the number of intervals m; for example, for the common RGB color space, each color channel (red, green, blue) can be divided into several intervals according to actual needs;

[0110] Then traverse each pixel in the image and determine the color interval to which the pixel belongs;

[0111] Finally, the number of pixels in each color interval is counted to obtain a color histogram;

[0112] Gray-level co-occurrence matrix is ​​a method used to describe the texture characteristics of an image. It reflects the texture information of an image by counting the number of pixel pairs with a specific gray value combination at a specific distance and direction.

[0113] Determine the parameters used to calculate the gray-level co-occurrence matrix, including gray-level values ​​i and j (usually the range of gray-level values ​​that may appear in the image), the distance d between pixels, and the direction θ;

[0114] Traverse each pixel in the image, for each pixel, find other pixels with a distance of d, a direction of θ and grayscale values ​​of i and j, and count the number of such pixel pairs to obtain the grayscale co-occurrence matrix;

[0115] Fourier descriptor is a shape description method based on frequency domain analysis; it represents the contour of the shape as a series of complex coordinates, and then obtains a set of coefficients through discrete Fourier transform, which can describe the characteristics of the shape; low-frequency coefficients reflect the overall contour of the shape, and high-frequency coefficients reflect the details of the shape; and Fourier descriptor belongs to the existing technical means, so it will not be elaborated in detail.

[0116] In order to further optimize the above embodiment, in step 4, the preset fault diagnosis model is:

[0117]

[0118] Among them, G is the fusion parameter index, which represents the result of fusing the temperature change rate, pressure fluctuation amplitude, flow change trend, and color, texture, and shape features through a linear combination; α, β, γ, δ a ,∈ b, θ are weight coefficients, which are used to adjust the importance of different features in the fault diagnosis process. Their values ​​are determined based on the training and optimization of the fault diagnosis model based on historical data. The result of the linear combination is mapped to between 0 and 1 through 1 / (1+exp{-x}) to obtain the fault probability value, that is, the fusion parameter index G;

[0119] When the fusion parameter index G is closer to 1, it indicates that the possibility of failure in the slag removal system is greater; when the fusion parameter index G is closer to 0, it indicates that the possibility of normal operation of the slag removal system is greater.

[0120] It should be noted that the model introduces more weight coefficients and adjustment coefficients, which can be adjusted more finely according to the characteristics of different slag removal systems and actual data; this enables the model to better adapt to various complex working environments and fault conditions;

[0121] Through the form of exponential function and logistic regression, nonlinear mapping of features is realized; this nonlinear mapping can better capture the complex relationship between features and improve the accuracy of fault diagnosis; for example, there may be interactions between some features, and nonlinear models can better capture such interaction effects;

[0122] The output of the model can be directly interpreted as the probability of a fault, which provides more intuitive information for fault diagnosis. Operators can judge the severity of the fault based on the probability value and take corresponding measures. Although the model is relatively complex, the weight coefficient and adjustment coefficient of each feature can be optimized and determined through training data. This makes the model interpretable to a certain extent, that is, it can analyze which features contribute more to fault diagnosis, thereby providing guidance for system maintenance and improvement.

[0123] At the same time, the model can adapt to different types of sensor data and image features, whether it is continuous numerical features (such as temperature, pressure, flow) or discrete features (such as color, texture, shape), and can be effectively integrated and diagnosed; this makes the model have good application prospects under different slag removal system configurations and monitoring requirements.

[0124] In order to further optimize the above embodiment, the process of diagnosing whether there is a fault in the slag removal system based on the fusion result includes:

[0125] Preset fault threshold G E ; If the fusion parameter index G is greater than or equal to the fault threshold G E When the current slag removal system is judged to have a fault; if the fusion parameter index G is greater than the fault threshold G E , it is judged that there is no fault in the current slag removal system.

[0126] It should be noted that the setting of the fault threshold is based on a large amount of experimental data and actual operating experience; during the system development and testing phase, statistical analysis is performed on the fusion parameter indicators of the slag dredging system under normal operation and known fault conditions; for example, a large amount of data during normal operation is collected, the distribution range of its fusion parameter indicators is calculated, and the fusion parameter indicator values ​​when different types of faults occur are collected at the same time; based on these data, a suitable fault threshold can be determined so that the fusion parameter indicator is usually less than the threshold during normal operation, and the fusion parameter indicator can exceed the threshold when a fault occurs; with the operation and use of the slag dredging system, its performance and operating environment may change; for example, the aging of the equipment may cause some parameters to change during normal operation, or new operating conditions may affect the manifestation of the fault; in this case, the fault threshold may need to be adjusted; the fusion parameter indicators of the system under normal and fault conditions can be re-evaluated regularly, and the appropriate fault threshold can be reset based on the new data to ensure the accuracy of fault diagnosis.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent slag removal system fault diagnosis method, characterized in that: include: Step 1: Use sensors to collect temperature, pressure, and flow parameters in the slag removal system in real time, and use image recognition technology to collect image information of key parts; Step 2: filter the parameters collected by the sensor and enhance the image information; Step 3: extracting temperature change rate, pressure fluctuation amplitude, flow rate change trend, and color feature, texture feature, and shape feature information from the processed data; Step 4: Based on the preset fault diagnosis model, the extracted feature information is fused, and based on the fusion result, it is diagnosed whether the slag removal system has a fault; Step five: Generate a warning signal based on the diagnosis results.

2. The intelligent slag removal system fault diagnosis method according to claim 1 is characterized in that: In step one, the temperature, pressure and flow parameters are obtained by temperature sensors, pressure sensors and flow sensors installed in the slag dredging system; and images are collected by cameras to obtain appearance images of key parts, namely, the image information.

3. The intelligent slag removal system fault diagnosis method according to claim 2 is characterized in that: In step 2, the process of filtering the parameters collected by the sensor adopts a median filtering algorithm, and the process of enhancing the image information adopts a histogram equalization algorithm.

4. The intelligent slag removal system fault diagnosis method according to claim 3 is characterized in that: In step three, the temperature change rate extraction process is: Get the temperature values ​​and corresponding time intervals at adjacent time points; Calculate the temperature change rate T at the current moment by the first calculation formula rate ; The first calculation formula is: Among them, T rate is the temperature change rate, T t+Δt and T t are the temperatures at time t+Δt and t respectively, Δt is the time interval, n is the number of time points between time t-Δt and t, and s is a smoothing factor used to balance the current and historical temperature changes and reduce the impact of noise; the smoothing factor s is determined by trial and error or based on historical data analysis.

5. The intelligent slag removal system fault diagnosis method according to claim 4 is characterized in that: In step three, the process of extracting the pressure fluctuation amplitude is as follows: Get the current pressure value and the pressure value in the previous preset period; Calculate the pressure fluctuation amplitude P at the current moment by the second calculation formula std ; The second calculation formula is: Among them, P std is the pressure fluctuation amplitude; P i is the pressure value at the current moment i, is the average value of the pressure in the preset period, and n is the total number of time points corresponding to the preset period; p max is the maximum pressure value within the preset period, P min is the minimum pressure value within the preset period, P avg is the ideal pressure value rated by the system; ω is the pressure stability weight, which is used to reflect the system's requirements for pressure stability and adjust the focus of pressure fluctuation assessment; the pressure stability weight ω is determined based on empirical settings or optimization algorithms.

6. The intelligent slag removal system fault diagnosis method according to claim 5 is characterized in that: In step three, the process of extracting the flow change trend is as follows: Get the current flow value and the flow value within the preset time limit; The preset time limit is less than a preset period of time during the extraction process of the pressure fluctuation amplitude; The current flow rate change trend Q is calculated by the third calculation formula tre ; The third calculation formula is: Among them, is the flow change trend; Q is the flow value at the current time t, is the derivative of the flow rate at the current moment with respect to time; δt is the preset time limit, that is, a preset time interval; m is the number of time points between time t-δt and t, is the time derivative of the flow at the jth time point; q is the smoothness of flow change, which is used to integrate current and historical flow changes and reduce the impact of sudden flow changes; the flow change smoothness q is determined based on system requirements or data-driven methods.

7. The intelligent slag removal system fault diagnosis method according to claim 6 is characterized in that: In step three, the process of extracting the color feature, texture feature and shape feature information includes: The color feature information is represented by a color histogram, specifically by dividing the color space of the image information into m intervals, counting the number of pixels in each interval, and obtaining a color histogram, i.e., the color feature information H a ={h1, h2, ..., h m }, where h i is the number of pixels in the i-th color interval; The texture feature information is extracted using a gray level co-occurrence matrix, specifically by calculating the gray level co-occurrence matrix of the image information, i.e., the texture feature information P b ={i, j, d, θ}, where i and j are grayscale values, d is the distance between pixels, and θ is the direction; The shape feature information is represented by Fourier descriptors, specifically, for a shape contour C in the image information, it is represented as a series of complex coordinates (x k ,y k ), k = 0, 1, ..., N-1; where N is the number of points on the contour C; the contour is subjected to discrete Fourier transform, and the transformation formula is: The Fourier descriptor, namely the shape feature information F(u), is obtained through the transformation formula.

8. The intelligent slag removal system fault diagnosis method according to claim 7 is characterized in that: In step 4, the preset fault diagnosis model is: Among them, G is the fusion parameter index, which represents the result of fusing the temperature change rate, pressure fluctuation amplitude, flow change trend, and color, texture, and shape features through a linear combination; α, β, γ, δ a ,∈ b , θ are weight coefficients, which are used to adjust the importance of different features in the fault diagnosis process. Their values ​​are determined based on the training and optimization of the fault diagnosis model based on historical data. The result of the linear combination is mapped to between 0 and 1 through 1 / (1+exp{-x}) to obtain the fault probability value, that is, the fusion parameter index G; When the fusion parameter index G is closer to 1, it indicates that the possibility of failure in the slag removal system is greater; when the fusion parameter index G is closer to 0, it indicates that the possibility of normal operation of the slag removal system is greater.

9. The intelligent slag removal system fault diagnosis method according to claim 8, characterized in that: The process of diagnosing whether the slag removal system has a fault based on the fusion result includes: Preset fault threshold G E ; If the fusion parameter index G is greater than or equal to the fault threshold G E When the current slag removal system is judged to have a fault; if the fusion parameter index G is greater than the fault threshold G E , it is judged that there is no fault in the current slag removal system.

Images