CN114187237A - A conveyor belt tearing and deviation detection method, device, equipment and storage medium - Google Patents
A conveyor belt tearing and deviation detection method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114187237A CN114187237A CN202111341853.8A CN202111341853A CN114187237A CN 114187237 A CN114187237 A CN 114187237A CN 202111341853 A CN202111341853 A CN 202111341853A CN 114187237 A CN114187237 A CN 114187237A
- Authority
- CN
- China
- Prior art keywords
- image data
- neural network
- image
- deep neural
- abnormal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 238000001514 detection method Methods 0.000 title claims description 48
- 230000002159 abnormal effect Effects 0.000 claims abstract description 90
- 238000003062 neural network model Methods 0.000 claims abstract description 81
- 238000012549 training Methods 0.000 claims abstract description 61
- 238000012706 support-vector machine Methods 0.000 claims abstract description 55
- 239000013598 vector Substances 0.000 claims abstract description 50
- 238000000034 method Methods 0.000 claims abstract description 38
- 230000006870 function Effects 0.000 claims description 30
- 238000012545 processing Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 15
- 230000004044 response Effects 0.000 claims description 7
- 230000000694 effects Effects 0.000 abstract description 3
- 239000003245 coal Substances 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000012360 testing method Methods 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Image Analysis (AREA)
Abstract
Description
技术领域technical field
本申请涉及工业自动化技术领域,尤其是涉及一种输送带撕裂和跑偏检测方法、装置、设备及存储介质。The present application relates to the technical field of industrial automation, and in particular, to a method, device, device and storage medium for detecting tearing and deviation of a conveyor belt.
背景技术Background technique
目前,输送带是电厂运载动力煤炭的重要工具,是电厂安全可靠运行的必要条件。然而,输送带在长时间的煤炭运输过程中,由于负荷的交替变化、疲劳损伤等原因,将会发生撕裂、跑偏等问题,严重的情况下会影响工厂的正常运转,甚至威胁到工人的作业安全。At present, the conveyor belt is an important tool for the power plant to carry thermal coal, and it is a necessary condition for the safe and reliable operation of the power plant. However, in the long-term coal transportation process, the conveyor belt will be torn, deviate and other problems due to alternating load changes and fatigue damage. In severe cases, it will affect the normal operation of the factory and even threaten workers. work safety.
具体地,输送带撕裂指输送带在运输煤炭的过程中,由于摩擦损伤、金属卡顿、外部压力、高速运转等原因引起的输送带撕裂现象,将会导致动力煤泄漏,裂纹不断扩展,严重时会造成机器故障、生产停滞、人员伤亡等重大事故,将对煤矿运输系统的安全运行形成重大威胁。现有的输送带撕裂检测方法主要有人工检测、传感器检测和基于图像处理的撕裂检测方法,但人工检测方式不能及时发现撕裂并停止运输;传感器检测容易受损导致检测失效、且准确性、稳定性和可靠性低;基于图像处理的撕裂检测方法通常只能针对单一场景下的撕裂进行检测,对撕裂趋势也缺乏早期预测,通用性和可靠性低。Specifically, the tearing of the conveyor belt refers to the tearing of the conveyor belt caused by friction damage, metal jamming, external pressure, high-speed operation, etc. during the process of transporting coal, which will lead to thermal coal leakage and continuous expansion of cracks. In severe cases, it will cause major accidents such as machine failure, production stagnation, casualties, etc., which will pose a major threat to the safe operation of the coal mine transportation system. Existing conveyor belt tear detection methods mainly include manual detection, sensor detection and tear detection methods based on image processing, but manual detection methods cannot detect tears in time and stop transportation; sensor detection is easily damaged, resulting in detection failure, and accurate detection. The tearing detection method based on image processing can only detect tearing in a single scene, and it also lacks early prediction of tearing trend, and its versatility and reliability are low.
输送带跑偏指输送带运转过程中其实际中心线与理论中心线距离过大,会导致动力煤在末端集聚,形成侧漏,造成环境污染、增加清理工作量、影响生产效率和输送带使用寿命,重则导致安全生产事故。现有的输送带跑偏检测方法主要有人工检测、光电检测和视觉图像检测方法,但人工检测方法效率低、不能及时发现跑偏并进行停工检修;光电检测设备造价昂贵、难以推广;视觉图像检测方法依赖于数学模型、且受复杂环境影响较大而易导致检测准确率不高。The deviation of the conveyor belt means that the distance between the actual center line and the theoretical center line of the conveyor belt is too large, which will lead to the accumulation of thermal coal at the end, resulting in side leakage, causing environmental pollution, increasing the cleaning workload, affecting production efficiency and conveyor belt usage. life, and it will lead to safety production accidents. Existing conveyor belt deviation detection methods mainly include manual detection, photoelectric detection and visual image detection methods, but the manual detection method is inefficient, cannot detect deviation in time and stop work for maintenance; photoelectric detection equipment is expensive and difficult to popularize; visual image detection Detection methods rely on mathematical models and are greatly affected by complex environments, which easily lead to low detection accuracy.
针对上述中的相关技术,发明人认为存在有现有的输送带撕裂或跑偏场景的识别准确率较低的缺陷。In view of the above-mentioned related technologies, the inventor believes that there is a defect that the recognition accuracy of the existing conveyor belt tearing or deviation scene is low.
发明内容SUMMARY OF THE INVENTION
为了提高输送带撕裂或跑偏场景的识别准确率,本申请提供了一种输送带撕裂和跑偏检测方法、装置、设备及存储介质。In order to improve the recognition accuracy of a conveyor belt tearing or deviation scene, the present application provides a conveyor belt tearing and deviation detection method, device, equipment and storage medium.
第一方面,本申请提供一种输送带撕裂和跑偏检测方法,具有对输送带进行撕裂和跑偏精确检测的特点。In a first aspect, the present application provides a method for detecting tearing and deviation of a conveyor belt, which has the characteristics of accurately detecting the tearing and deviation of a conveyor belt.
本申请是通过以下技术方案得以实现的:This application is achieved through the following technical solutions:
一种输送带撕裂和跑偏检测方法,包括以下步骤:A method for detecting tearing and deviation of a conveyor belt, comprising the following steps:
获取输送带的视频数据,构建图像数据库;Obtain the video data of the conveyor belt and build an image database;
使所述图像数据库的数据输入深度神经网络模型中进行训练,并输出异常图像数据特征;inputting the data of the image database into a deep neural network model for training, and outputting abnormal image data features;
基于支持向量机模型的使分类准确率最高的超参数组合,并使标注的所述异常图像数据特征的撕裂图像数据特征和跑偏图像数据特征输入支持向量机模型进行训练,获得该超参数组合下的支持向量机模型;Based on the combination of hyperparameters with the highest classification accuracy based on the support vector machine model, and inputting the torn image data features and deviated image data features of the marked abnormal image data features into the support vector machine model for training, the hyperparameters are obtained The support vector machine model under the combination;
使训练的所述深度神经网络模型和所述超参数组合下的所述支持向量机模型对输送带的视频数据进行检测;Make the described deep neural network model of training and described support vector machine model under described hyperparameter combination to detect the video data of conveyor belt;
当满足预设条件时,所述深度神经网络模型输出特征向量z,并将当前的输送带的视频数据判断为异常;When the preset conditions are met, the deep neural network model outputs the feature vector z, and the video data of the current conveyor belt is judged to be abnormal;
所述支持向量机模型基于所述特征向量z,输出输送带撕裂或输送带跑偏的异常结果。The support vector machine model outputs abnormal results of conveyor belt tearing or conveyor belt deviation based on the feature vector z.
本申请在一较佳示例中可以进一步配置为:所述预设条件包括所述深度神经网络模型输出的特征向量z与预设中心向量c的距离r大于预设阈值R。In a preferred example of the present application, it may be further configured that: the preset condition includes that the distance r between the feature vector z output by the deep neural network model and the preset center vector c is greater than the preset threshold R.
本申请在一较佳示例中可以进一步配置为:使所述图像数据库的数据输入深度神经网络模型中进行训练的步骤包括:In a preferred example, the present application can be further configured as: the step of inputting the data of the image database into the deep neural network model for training includes:
预设对比损失函数,所述对比损失函数包括,A preset contrast loss function, the contrast loss function includes,
其中,I1表示正常图像数据集合,I2表示异常图像数据集合,i表示正常图像数据集合I1或异常图像数据集合I2中的某一个图像,Zi表示图像i的特征向量,I1\i表示正常图像数据集合I1中的某一图像i以外的其他所有图像数据集合,j表示正常图像数据集合I1或异常图像数据集合I2中某一图像i以外的一个图像,Zj表示图像j的特征向量,I2\i表示异常图像数据集合I2中的某一图像i以外的其他所有图像数据集合,t为正常图像数据集合I1或异常图像数据集合I2中的一个图像,Zt为图像t的特征向量;使用所述对比损失函数对深度神经网络模型进行预训练。Among them, I 1 represents the normal image data set, I 2 represents the abnormal image data set, i represents an image in the normal image data set I 1 or the abnormal image data set I 2 , Z i represents the feature vector of the image i, and I 1 \i represents all image data sets other than a certain image i in the normal image data set I 1 , j represents an image other than a certain image i in the normal image data set I 1 or abnormal image data set I 2 , Z j Represents the feature vector of image j, I 2 \i represents all other image data sets except a certain image i in the abnormal image data set I 2 , t is one of the normal image data set I 1 or the abnormal image data set I 2 image, Z t is the feature vector of image t; the deep neural network model is pre-trained using the contrast loss function.
本申请在一较佳示例中可以进一步配置为:对深度神经网络模型进行预训练后,还包括以下步骤:In a preferred example, the present application can be further configured as: after pre-training the deep neural network model, the following steps are also included:
预设训练损失函数,所述训练损失函数包括,A preset training loss function, the training loss function includes,
其中,zi为输入图像i在深度神经网络模型上的输出向量,c为预训练得到的输出向量均值;Among them, zi is the output vector of the input image i on the deep neural network model, and c is the mean value of the output vector obtained by pre-training;
使用所述训练损失函数对经过预训练的深度神经网络模型进行优化。The pretrained deep neural network model is optimized using the training loss function.
本申请在一较佳示例中可以进一步配置为:获取所述支持向量机模型的使分类准确率最高的超参数组合的步骤包括:In a preferred example, the present application can be further configured as follows: the step of obtaining the hyperparameter combination of the support vector machine model with the highest classification accuracy includes:
预设若干组支持向量机模型的包括核函数、惩罚参数、松弛向量的超参数组合,并基于所述撕裂图像数据特征和所述跑偏图像数据特征,训练支持向量机模型,获得与每组超参数组合对应的分类准确率;Preset several groups of support vector machine models including hyperparameter combinations of kernel functions, penalty parameters, and relaxation vectors, and train the support vector machine models based on the tearing image data features and the deviating image data features, and obtain the The classification accuracy corresponding to the group hyperparameter combination;
构造超参数组合与分类准确率之间的响应关系;Construct the response relationship between hyperparameter combinations and classification accuracy;
结合增益期望,获取分类准确率最高时的超参数组合。Combined with the gain expectation, the hyperparameter combination with the highest classification accuracy is obtained.
本申请在一较佳示例中可以进一步配置为:使所述图像数据库的数据输入深度神经网络模型中进行训练前,还包括以下步骤:In a preferred example, the present application can be further configured as: before the data of the image database is input into the deep neural network model for training, the following steps are further included:
对划分的所述图像数据库的异常类别图像数据的撕裂图像数据和跑偏图像数据进行数据增强处理;performing data enhancement processing on the tear image data and the deviated image data of the abnormal category image data of the divided image database;
使所述图像数据库的正常类别图像数据和经过数据增强处理的所述撕裂图像数据和所述跑偏图像数据输入深度神经网络模型中。The normal category image data of the image database and the torn image data and the deviated image data subjected to data enhancement processing are input into the deep neural network model.
本申请在一较佳示例中可以进一步配置为:使所述图像数据库的正常类别图像数据和经过数据增强处理的所述撕裂图像数据和所述跑偏图像数据输入深度神经网络模型中的步骤前,还包括以下步骤:In a preferred example, the present application may be further configured as: the steps of inputting the normal category image data of the image database, the tearing image data and the deviated image data after data enhancement processing into the deep neural network model Before, it also includes the following steps:
使经过数据增强处理的所述撕裂图像数据和所述跑偏图像数据,以及所述正常类别图像数据进行归一化处理后,再输入深度神经网络模型中。The tearing image data, the deviated image data, and the normal category image data that have undergone data enhancement processing are normalized, and then input into the deep neural network model.
第二方面,本申请提供一种输送带撕裂和跑偏检测装置,具有对输送带进行撕裂和跑偏精确检测的特点。In a second aspect, the present application provides a conveyor belt tearing and deviation detection device, which has the characteristics of accurately detecting the tearing and deviation of the conveyor belt.
本申请是通过以下技术方案得以实现的:This application is achieved through the following technical solutions:
一种输送带撕裂和跑偏检测装置,包括:A conveyor belt tearing and deviation detection device, comprising:
数据获取模块,用于获取输送带的视频数据,构建图像数据库;The data acquisition module is used to acquire the video data of the conveyor belt and construct an image database;
第一训练模块,用于使所述图像数据库的数据输入深度神经网络模型中进行训练,并输出异常图像数据特征;The first training module is used to input the data of the image database into a deep neural network model for training, and output abnormal image data features;
第二训练模块,用于基于支持向量机模型的使分类准确率最高的超参数组合,并使标注的所述异常图像数据特征的撕裂图像数据特征和跑偏图像数据特征输入支持向量机模型进行训练,获得该超参数组合下的支持向量机模型;The second training module is used to combine the hyperparameters with the highest classification accuracy based on the support vector machine model, and input the torn image data features and deviated image data features of the marked abnormal image data features into the support vector machine model Perform training to obtain the support vector machine model under the hyperparameter combination;
检测模块,用于使训练的所述深度神经网络模型和所述超参数组合下的所述支持向量机模型对输送带的视频数据进行检测;A detection module, configured to enable the trained deep neural network model and the support vector machine model under the hyperparameter combination to detect the video data of the conveyor belt;
异常感知模块,用于在满足预设条件时,使所述深度神经网络模型输出特征向量z,并将当前的输送带的视频数据判断为异常;an abnormality perception module, used to make the deep neural network model output a feature vector z when a preset condition is met, and judge the current video data of the conveyor belt as abnormal;
异常类别预警模块,用于使所述支持向量机模型基于所述特征向量z,输出输送带撕裂或输送带跑偏的异常结果。The abnormal category early warning module is configured to enable the support vector machine model to output abnormal results of conveyor belt tearing or conveyor belt deviation based on the feature vector z.
第三方面,本申请提供一种设备,具有对输送带进行撕裂和跑偏精确检测的特点。In a third aspect, the present application provides a device with the characteristics of accurately detecting the tearing and deviation of the conveyor belt.
本申请是通过以下技术方案得以实现的:This application is achieved through the following technical solutions:
一种设备,包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序,所述处理器执行所述计算机程序时实现上述一种输送带撕裂和跑偏检测方法的步骤。A device comprising a memory, a processor and a computer program stored in the memory and running on the processor, the processor implements the above-mentioned tearing and deviation of the conveyor belt when executing the computer program The steps of the detection method.
第四方面,本申请提供一种存储介质,具有对输送带进行撕裂和跑偏精确检测的特点。In a fourth aspect, the present application provides a storage medium, which has the characteristics of accurately detecting tearing and deviation of a conveyor belt.
本申请是通过以下技术方案得以实现的:This application is achieved through the following technical solutions:
一种存储介质,所述存储介质存储有计算机程序,所述计算机程序被处理器执行时实现上述一种输送带撕裂和跑偏检测方法的步骤。A storage medium, which stores a computer program, and when the computer program is executed by a processor, implements the steps of the above-mentioned method for detecting the tearing and deviation of a conveyor belt.
本申请包括以下至少一种有益技术效果:The present application includes at least one of the following beneficial technical effects:
1、一种输送带撕裂和跑偏检测方法基于神经网络的视觉检测方法对输送带进行故障检测,训练深度神经网络模型提取输送带的图像特征,对输送带的正常情况和异常情况进行分类,当判断为异常情况时则输出异常图像数据特征;再基于异常图像数据特征,标注撕裂图像数据特征和跑偏图像数据特征,并获取支持向量机模型的使分类准确率最高的超参数组合,训练支持向量机模型,以对输送带异常图像的撕裂情况和跑偏情况进行进一步细分;改善了现有的输送带撕裂或跑偏场景的识别准确率较低的情况,提高了输送带撕裂或跑偏场景的识别准确率,克服了传统检测方法的及时性、效率性、准确性、可靠性、通用性的缺陷,能够实现高效、精确的输送带检测,及时对设备进行停机维修,对保障能源工厂安全可靠运行具有重要意义;1. A conveyor belt tearing and deviation detection method. The visual detection method based on neural network detects the fault of the conveyor belt, trains the deep neural network model to extract the image features of the conveyor belt, and classifies the normal and abnormal conditions of the conveyor belt. , when it is judged to be abnormal, the abnormal image data features are output; then based on the abnormal image data features, the tearing image data features and the deviated image data features are marked, and the hyperparameter combination of the support vector machine model that maximizes the classification accuracy is obtained. , train the support vector machine model to further subdivide the tearing and deviation of abnormal images of the conveyor belt; improve the recognition accuracy of the existing conveyor belt tearing or deviation scenes, and improve the The recognition accuracy of the conveyor belt tearing or deviation scene overcomes the defects of timeliness, efficiency, accuracy, reliability, and versatility of traditional detection methods, and can realize efficient and accurate conveyor belt detection, and timely carry out equipment inspection. Shutdown for maintenance is of great significance to ensuring the safe and reliable operation of energy plants;
2、使对比损失函数对深度神经网络模型进行预训练,以提升深度神经网络模型对图像特征的提取能力,能对图像背景、噪声等干扰特征进行有效过滤,并且能够提升模型的训练效率;2. Use the contrast loss function to pre-train the deep neural network model to improve the ability of the deep neural network model to extract image features, effectively filter the image background, noise and other interfering features, and improve the training efficiency of the model;
3、使训练损失函数对被对比损失函数预训练的深度神经网络模型进行优化,使得输送带正常图像的输出特征接近中心c,而输送带异常图像的输出特征远离中心c,以对输送带的正常图像和异常图像进行精确分类;3. Make the training loss function optimize the deep neural network model pre-trained by the contrast loss function, so that the output features of the normal image of the conveyor belt are close to the center c, and the output features of the abnormal images of the conveyor belt are far away from the center c, so as to improve the accuracy of the conveyor belt. Accurate classification of normal and abnormal images;
4、获取支持向量机模型的使分类准确率最高的超参数组合训练SVM模型,以采用最优超参数组合下的SVM模型,能对输送带的异常图像数据特征类别进行进一步分类,有效避免了由于异常数据较少导致的严重过拟合问题,提高了模型的泛化性能,即避免了使用深度神经网络模型进行分类引发的偏向问题,提高了输送带异常图像检测的判断精度;同时,缩小了输入数据的维度,能提高对输送带正常图像数据这一比例较大的类别图像的检测效率,提高了SVM模型的响应速度;4. Obtain the hyperparameter combination of the support vector machine model to train the SVM model with the highest classification accuracy, and use the SVM model under the optimal hyperparameter combination to further classify the abnormal image data feature categories of the conveyor belt, effectively avoiding the need for Due to the serious over-fitting problem caused by less abnormal data, the generalization performance of the model is improved, that is, the bias problem caused by using the deep neural network model for classification is avoided, and the judgment accuracy of abnormal image detection of the conveyor belt is improved. The dimension of the input data can be improved, the detection efficiency of the category image with a large proportion of the normal image data of the conveyor belt can be improved, and the response speed of the SVM model can be improved;
5、对划分的撕裂图像数据和跑偏图像数据进行数据增强处理后再输入深度神经网络模型中,以增加输送带在跑偏和撕裂状态下的异常图像样本,减轻了输送带的图像样本数据不平衡的情况,利于提高输送带撕裂和跑偏场景的识别准确率;5. Perform data enhancement processing on the divided tear image data and deviation image data and then input them into the deep neural network model to increase the abnormal image samples of the conveyor belt in the deviation and tear state, and reduce the image of the conveyor belt. Unbalanced sample data is beneficial to improve the recognition accuracy of conveyor belt tearing and deviation scenarios;
6、使经过数据增强处理的撕裂图像数据和跑偏图像数据,以及正常类别图像数据进行归一化处理后,再输入深度神经网络模型中,以减少深度神经网络模型训练时的数据计算量,便于后续的深度神经网络模型训练。6. Normalize the tearing image data, deviated image data, and normal category image data after data enhancement processing, and then input it into the deep neural network model to reduce the amount of data calculation in the training of the deep neural network model. , which is convenient for subsequent deep neural network model training.
附图说明Description of drawings
图1是本申请其中一实施例一种输送带撕裂和跑偏检测方法的模型训练流程图。FIG. 1 is a model training flow chart of a method for detecting tearing and deviation of a conveyor belt according to an embodiment of the present application.
图2是深度神经网络模型和支持向量机模型的训练示意图。Figure 2 is a schematic diagram of the training of the deep neural network model and the support vector machine model.
图3是本申请其中一实施例一种输送带撕裂和跑偏检测方法的流程图。FIG. 3 is a flowchart of a method for detecting tearing and deviation of a conveyor belt according to an embodiment of the present application.
图4是本申请其中一实施例一种输送带撕裂和跑偏检测装置的结构框图。FIG. 4 is a structural block diagram of a conveyor belt tearing and deviation detection device according to an embodiment of the present application.
具体实施方式Detailed ways
本具体实施例仅仅是对本申请的解释,其并不是对本申请的限制,本领域技术人员在阅读完本说明书后可以根据需要对本实施例做出没有创造性贡献的修改,但只要在本申请的权利要求范围内都受到专利法的保护。This specific embodiment is only an explanation of the application, and it does not limit the application. Those skilled in the art can make modifications to the embodiment without creative contribution as needed after reading this specification, but as long as the rights of the application are All claims are protected by patent law.
为使本申请实施例的目的、技术方案和优点更加清楚,下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments It is a part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.
另外,本文中术语“和/或”,仅仅是一种描述关联对象的关联关系,表示可以存在三种关系,例如,A和/或B,可以表示:单独存在A,同时存在A和B,单独存在B这三种情况。另外,本文中字符“/”,如无特殊说明,一般表示前后关联对象是一种“或”的关系。In addition, the term "and/or" in this article is only an association relationship to describe the associated objects, indicating that there can be three kinds of relationships, for example, A and/or B, it can mean that A exists alone, A and B exist at the same time, There are three cases of B alone. In addition, the character "/" in this text, unless otherwise specified, generally indicates that the related objects before and after are an "or" relationship.
下面结合说明书附图对本申请实施例作进一步详细描述。The embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
参照图1,本申请实施例提供一种输送带撕裂和跑偏检测方法,所述方法的主要步骤描述如下。Referring to FIG. 1 , an embodiment of the present application provides a method for detecting tearing and deviation of a conveyor belt. The main steps of the method are described as follows.
获取输送带的视频数据,构建图像数据库;Obtain the video data of the conveyor belt and build an image database;
使图像数据库的数据输入深度神经网络模型中进行训练,并输出异常图像数据特征;The data of the image database is input into the deep neural network model for training, and the abnormal image data features are output;
基于支持向量机模型的使分类准确率最高的超参数组合,并使标注的异常图像数据特征的撕裂图像数据特征和跑偏图像数据特征输入支持向量机模型进行训练,获得该超参数组合下的支持向量机模型;Based on the support vector machine model, the hyperparameter combination with the highest classification accuracy, and the tearing image data features and deviated image data features of the marked abnormal image data features are input into the support vector machine model for training, and the hyperparameter combination is obtained. The support vector machine model of ;
使训练的深度神经网络模型和超参数组合下的支持向量机模型对输送带的视频数据进行检测;Use the trained deep neural network model and the support vector machine model under the combination of hyperparameters to detect the video data of the conveyor belt;
当满足预设条件时,深度神经网络模型输出特征向量z,并将当前的输送带的视频数据判断为异常;When the preset conditions are met, the deep neural network model outputs the feature vector z, and judges the current video data of the conveyor belt as abnormal;
支持向量机模型基于特征向量z,输出输送带撕裂或输送带跑偏的异常结果。Based on the feature vector z, the SVM model outputs abnormal results of belt tearing or belt misalignment.
具体地,收集电厂动力煤输送带视频数据,隔若干帧取一幅图像,构建图像数据库。本实施例中,在输送带视频中每隔5帧取一幅图像,聚集形成图像数据集合。Specifically, the video data of the thermal coal conveyor belt of the power plant is collected, and an image is taken every several frames to construct an image database. In this embodiment, an image is taken every 5 frames in the conveyor belt video, and is aggregated to form an image data set.
再将图像数据集合中的图像标注为正常和异常两个类别,得到正常类别图像数据和异常类别图像数据;其中,类别图像数据进一步细分,标注为撕裂图像数据和跑偏图像数据两类,构建图像数据库。Then, the images in the image data set are marked as normal and abnormal categories, and normal category image data and abnormal category image data are obtained; among them, the category image data is further subdivided and marked as two categories: tearing image data and deviating image data. , build the image database.
进一步地,通过分析图像数据库内的图像信息,发现输送带在正常状态下的图像样本很多,而在跑偏和撕裂状态下的异常样本很少,出现了图像样本数据不平衡的问题,进而影响了输送带撕裂和跑偏场景的识别准确率。Further, by analyzing the image information in the image database, it is found that there are many image samples in the normal state of the conveyor belt, but very few abnormal samples in the deviation and tear state, and the problem of image sample data imbalance occurs. It affects the recognition accuracy of conveyor belt tearing and deviation scenarios.
故对撕裂图像数据和跑偏图像数据进行数据增强处理,以扩充异常类别图像的训练数据集,提高模型的精度。本实施例中,可采用旋转、缩放、色彩空间调整、马赛克增强等增强处理方式对撕裂图像数据和跑偏图像数据进行增强处理,增加电厂动力煤输送带在跑偏和撕裂状态下的异常样本,以减轻输送带的图像样本数据不平衡的情况,利于提高输送带撕裂和跑偏场景的识别准确率。Therefore, data enhancement processing is performed on the tearing image data and the deviating image data to expand the training data set of abnormal images and improve the accuracy of the model. In this embodiment, enhancement processing methods such as rotation, scaling, color space adjustment, mosaic enhancement, etc. can be used to enhance the tearing image data and the deviation image data, so as to increase the power plant thermal coal conveyor belt in the deviation and tearing state. Abnormal samples to reduce the imbalance of the image sample data of the conveyor belt, which is beneficial to improve the recognition accuracy of the tearing and deviation of the conveyor belt.
进一步地,使经过数据增强处理的撕裂图像数据和跑偏图像数据,以及图像数据库的正常类别图像数据进行归一化处理后,再输入深度神经网络模型中训练。本实施例中,可采用最大最小值归一化方法,具体公式如下:Further, after normalizing the tearing image data and deviating image data after data enhancement processing, as well as the normal category image data of the image database, they are input into the deep neural network model for training. In this embodiment, the maximum and minimum normalization method can be adopted, and the specific formula is as follows:
其中,xi表示待处理的图像原始像素点,min(x)表示图像像素的最小值, max(x)表示图像像素的最大值。Wherein, x i represents the original pixel point of the image to be processed, min(x) represents the minimum value of the image pixel, and max(x) represents the maximum value of the image pixel.
归一化处理不会改变图像本身的信息存储,但是归一化处理后的图像像素值的取值范围从0~255转化为0~1,以减少深度神经网络模型训练时的数据计算量,便于后续的深度神经网络模型训练。The normalization process will not change the information storage of the image itself, but the value range of the normalized image pixel value is converted from 0 to 255 to 0 to 1, so as to reduce the amount of data calculation during the training of the deep neural network model. It is convenient for subsequent deep neural network model training.
本实施例中,深度神经网络模型采用ResNet18-SVDD深度神经网络模型,以提取输送带的图像数据特征。In this embodiment, the deep neural network model adopts the ResNet18-SVDD deep neural network model to extract the image data features of the conveyor belt.
具体地,参照图2,构建ResNet18深度神经网络模型,模型的主体部分采用ResNet18结构,其后接若干个全连接层FC。本实施例中,全连接层FC可以为3个。Specifically, referring to Fig. 2, a ResNet18 deep neural network model is constructed. The main part of the model adopts the ResNet18 structure, followed by several fully connected layers FC. In this embodiment, there may be three fully connected layers FC.
再通过预设对比损失函数,以使用对比学习方法对深度神经网络模型进行预训练。Then, a preset contrastive loss function is used to pre-train the deep neural network model using the contrastive learning method.
其中,对比损失函数可包括,Among them, the contrastive loss function can include,
其中,I1表示正常图像数据集合, I2表示异常图像数据集合,i表示正常图像数据集合I1或异常图像数据集合I2中的某一个图像,Zi表示图像i的特征向量,I1\i表示正常图像数据集合I1中的某一图像i以外的其他所有图像数据集合,j表示正常图像数据集合I1或异常图像数据集合I2中某一图像i以外的一个图像,Zj表示图像j的特征向量,I2\i表示异常图像数据集合I2中的某一图像i以外的其他所有图像数据集合,t为正常图像数据集合I1或异常图像数据集合I2中的一个图像,Zt为图像t的特征向量。in, I 1 represents the normal image data set, I 2 represents the abnormal image data set, i represents a certain image in the normal image data set I 1 or the abnormal image data set I 2 , Z i represents the feature vector of the image i, I 1 \i represents the normal image data set I 1 All other image data sets except a certain image i, j represents an image other than a certain image i in the normal image data set I 1 or abnormal image data set I 2 , Z j represents the feature vector of image j, I 2 \ i represents all image data sets other than a certain image i in the abnormal image data set I 2 , t is an image in the normal image data set I 1 or the abnormal image data set I 2 , Z t is the feature vector of the image t .
基于归一化处理的图像数据,使对比损失函数对深度神经网络模型进行预训练,以提升深度神经网络模型对图像特征的提取能力,能对图像背景、噪声等干扰特征进行有效过滤,并且能够提升模型的训练效率。此时,深度神经网络模型输出得到输出向量zi。Based on the normalized image data, the contrast loss function is used to pre-train the deep neural network model to improve the ability of the deep neural network model to extract image features, effectively filter the image background, noise and other interfering features, and can Improve the training efficiency of the model. At this point, the output vector zi is obtained from the output of the deep neural network model.
进一步地,预设训练损失函数,训练损失函数可包括,Further, the preset training loss function, the training loss function may include,
其中,I1表示正常图像数据集合,I2表示异常图像数据集合,zi为输入图像i在深度神经网络模型上的输出向量,c为通过对比学习得到的输出向量均值,R为预设训练集。Among them, I 1 represents the normal image data set, I 2 represents the abnormal image data set, z i is the output vector of the input image i on the deep neural network model, c is the mean value of the output vector obtained through comparative learning, and R is the preset training set.
基于正常图像数据集合I1、异常图像数据集合I2和输入图像i在深度神经网络模型上的输出向量zi,使训练损失函数对被对比损失函数预训练的深度神经网络模型进行优化,得到优化后的ResNet18-SVDD深度神经网络模型。Based on the normal image data set I 1 , the abnormal image data set I 2 and the output vector zi of the input image i on the deep neural network model, the training loss function is optimized to the deep neural network model pre-trained by the contrast loss function, and the obtained The optimized ResNet18-SVDD deep neural network model.
采用优化后的ResNet18-SVDD深度神经网络模型,因其支持向量数据描述(Support Vector Data Description,SVDD),即通过训练损失函数,找到一个最小半径的超球面使得内部为正样本,外部为负样本,则恰好位于该超球面上的样本就是构建该超球面的支持向量,进而使得输送带正常图像的输出特征接近中心c,而输送带异常图像的输出特征远离中心c,以对输送带的正常图像和异常图像进行精确分类。The optimized ResNet18-SVDD deep neural network model is used because of its Support Vector Data Description (SVDD), that is, through the training loss function, a hypersphere with a minimum radius is found so that the interior is a positive sample and the exterior is a negative sample , then the sample just located on the hypersphere is to construct the support vector of the hypersphere, so that the output feature of the normal image of the conveyor belt is close to the center c, and the output feature of the abnormal image of the conveyor belt is far away from the center c, so as to make the normal image of the conveyor belt close to the center c. Images and abnormal images are accurately classified.
又由于输送带异常图像的数据量较少,即使使用优化后的 ResNet18-SVDD深度神经网络模型对输送带的视频数据进行分类,仍然容易出现过拟合问题而影响对输送带撕裂或跑偏场景的识别准确率,而支持向量机模型(support vector machines,SVM)在少量数据的分类问题上表现更为优异,支持向量机是一种二分类模型,它的基本模型是定义在特征空间上的间隔最大的线性分类器。SVM的学习策略就是间隔最大化,可形式化为一个求解凸二次规划的问题,也等价于正则化的合页损失函数的最小化问题。SVM的学习算法就是求解凸二次规划的最优化算法。In addition, due to the small amount of data of abnormal images of the conveyor belt, even if the optimized ResNet18-SVDD deep neural network model is used to classify the video data of the conveyor belt, it is still prone to overfitting, which affects the tearing or deviation of the conveyor belt. The recognition accuracy of the scene, and the support vector machine model (support vector machines, SVM) is more excellent in the classification of a small amount of data. The support vector machine is a two-class model, and its basic model is defined in the feature space. A linear classifier with the largest margin. The learning strategy of SVM is interval maximization, which can be formalized as a problem of solving convex quadratic programming, which is also equivalent to the minimization problem of regularized hinge loss function. The learning algorithm of SVM is the optimization algorithm for solving convex quadratic programming.
因此,本方案将深度神经网络模型输出的异常图像数据特征,输入到支持向量机模型中进行训练,以对输送带异常图像的类别作进一步细分。Therefore, in this scheme, the abnormal image data features output by the deep neural network model are input into the support vector machine model for training, so as to further subdivide the categories of abnormal images of the conveyor belt.
参照图1和图2,具体地,获取优化后的ResNet18-SVDD深度神经网络模型输出的输出向量zi中的所有异常图像输出向量z2,以作为异常图像输出特征数据集,并将异常图像输出特征数据集标注为撕裂图像数据特征和跑偏图像数据特征两类。Referring to FIG. 1 and FIG. 2, specifically, obtain all abnormal image output vectors z 2 in the output vector zi output by the optimized ResNet18-SVDD deep neural network model, as the abnormal image output feature data set, and the abnormal image The output feature datasets are labeled as tearing image data features and deviating image data features.
对撕裂图像数据特征和跑偏图像数据特征进行归一化处理,并将数据集的80%划分为训练集,20%划分为测试集,以预设训练集和测试集。其中,训练集用于训练模型;测试集用于对经训练集训练的模型进行测试,以找出效果最佳的模型。Normalize the torn image data features and the deviated image data features, and divide 80% of the data set into training sets and 20% into test sets to preset training sets and test sets. Among them, the training set is used to train the model; the test set is used to test the model trained on the training set to find the model with the best effect.
使用归一化处理后的撕裂图像数据特征和跑偏图像数据特征的数据集训练SVM模型,并通过Kiring响应面方法,获取SVM的最优超参数组合,以使分类准确率最高,继而得到该组最优超参数组合下的SVM模型。The SVM model is trained using the normalized data sets of tearing image data features and deviating image data features, and the optimal hyperparameter combination of SVM is obtained through the Kiring response surface method, so that the classification accuracy is the highest, and then the The SVM model under this set of optimal hyperparameter combinations.
其中,获取SVM模型的使分类准确率最高的超参数组合的步骤包括:预设若干组支持向量机模型的包括核函数、惩罚参数、松弛向量的超参数组合,并基于撕裂图像数据特征和跑偏图像数据特征,训练支持向量机模型,获得与每组超参数组合对应的分类准确率;其中,分类准确率是评价分类器性能的指标,分类准确率指对于给定的测试数据集合,分类器正确分类的样本数和总样本数之比,即分类准确率P的公式为:Wherein, the step of obtaining the hyperparameter combination of the SVM model with the highest classification accuracy includes: presetting several sets of hyperparameter combinations of the support vector machine model including the kernel function, the penalty parameter, and the relaxation vector, and based on the tear image data features and Deviate the image data features, train the support vector machine model, and obtain the classification accuracy rate corresponding to each set of hyperparameter combinations; among them, the classification accuracy rate is an indicator for evaluating the performance of the classifier, and the classification accuracy rate refers to the given test data set, The ratio of the number of samples correctly classified by the classifier to the total number of samples, that is, the formula of the classification accuracy P is:
其中,TP指将正类样本预测为正类数,FP指将负类样本预测为正类数;Among them, TP refers to predicting positive class samples as positive class numbers, and FP refers to predicting negative class samples as positive class numbers;
构造超参数组合与分类准确率之间的响应关系,即Kriging响应面方法,通过克里金插值法得到超参数组合和分类准确率的函数表达式,以构造 SVM模型;Construct the response relationship between hyperparameter combination and classification accuracy, that is, Kriging response surface method, obtain the function expression of hyperparameter combination and classification accuracy through Kriging interpolation method to construct SVM model;
结合增益期望(expected improvement,EI),通过遗传算法遍寻分类准确率最高的超参数组合,以获取分类准确率最高时的超参数组合,作为 SVM的最优超参数组合,并训练获得该超参数组合下的SVM模型。Combined with the expected improvement (EI), the hyperparameter combination with the highest classification accuracy is searched through the genetic algorithm to obtain the hyperparameter combination with the highest classification accuracy as the optimal hyperparameter combination of the SVM, and the hyperparameter combination with the highest classification accuracy is obtained by training. SVM model under parameter combination.
进而最优超参数组合下的SVM模型能对输送带的异常图像数据特征类别进行进一步分类,有效避免了由于异常数据较少导致的严重过拟合问题,提高了模型的泛化性能,即避免了使用深度神经网络模型进行分类引发的偏向问题,提高了输送带异常图像检测的判断精度;同时,缩小了输入数据的维度,能提高对输送带正常图像数据这一比例较大的类别图像的检测效率,提高了SVM模型的响应速度。Furthermore, the SVM model under the optimal hyperparameter combination can further classify the abnormal image data feature categories of the conveyor belt, which effectively avoids the serious over-fitting problem caused by the lack of abnormal data, and improves the generalization performance of the model. The bias problem caused by the use of the deep neural network model for classification is improved, and the judgment accuracy of abnormal image detection of the conveyor belt is improved; at the same time, the dimension of the input data is reduced, which can improve the accuracy of the category images with a large proportion of the normal image data of the conveyor belt. The detection efficiency improves the response speed of the SVM model.
本实施例中,超参数组合可以随机设定。In this embodiment, the hyperparameter combination can be set randomly.
参照图3,实时获取输送带的视频数据,每隔若干帧提取一幅图像,上传到服务器,并对图像数据进行归一化处理后,输入训练好的深度神经网络模型中。Referring to FIG. 3 , the video data of the conveyor belt is acquired in real time, an image is extracted every several frames, uploaded to the server, and the image data is normalized and input into the trained deep neural network model.
使训练的深度神经网络模型对输入的图像数据进行实时检测。Make the trained deep neural network model perform real-time detection on the input image data.
当满足预设条件时,深度神经网络模型输出特征向量z,并将当前的输送带的视频数据判断为异常。本实施例中,预设条件包括深度神经网络模型输出的特征向量z与预设中心向量c之间的距离r大于预设阈值R,其中,r=||z-c||^2。When the preset conditions are met, the deep neural network model outputs the feature vector z, and judges the current video data of the conveyor belt as abnormal. In this embodiment, the preset condition includes that the distance r between the feature vector z output by the deep neural network model and the preset center vector c is greater than the preset threshold R, where r=||z-c||^2.
本实施例中,预设阈值R可以为训练ResNet18-SVDD深度神经网络模型得到的超球面的半径值。In this embodiment, the preset threshold R may be the radius value of the hypersphere obtained by training the ResNet18-SVDD deep neural network model.
当距离r大于预设阈值R时,深度神经网络模型输出特征向量z,并判断当前的输送带的视频数据异常;当距离r小于或等于预设阈值R时,深度神经网络模型则判断当前的输送带的视频数据正常,继续进行下一幅图像的检测。When the distance r is greater than the preset threshold R, the deep neural network model outputs the feature vector z, and judges that the video data of the current conveyor belt is abnormal; when the distance r is less than or equal to the preset threshold R, the deep neural network model judges the current The video data of the conveyor belt is normal, and the detection of the next image is continued.
同时,输出特征向量z输入最优超参数组合下的SVM模型中,SVM模型基于特征向量z,输出输送带撕裂或输送带跑偏的异常结果,以获得输送带异常情况的类型,并发出异常类型提醒。At the same time, the output feature vector z is input into the SVM model under the optimal hyperparameter combination. Based on the feature vector z, the SVM model outputs the abnormal results of conveyor belt tearing or conveyor belt deviation, so as to obtain the type of conveyor belt abnormality and send out Exception type alert.
进而一种输送带撕裂和跑偏检测方法基于神经网络的视觉检测方法对输送带进行故障检测,训练深度神经网络模型提取输送带的图像特征,对输送带的正常情况和异常情况进行分类,当判断为异常情况时则输出异常图像数据特征;再基于异常图像数据特征,标注撕裂图像数据特征和跑偏图像数据特征,并获取支持向量机模型的使分类准确率最高的超参数组合,训练支持向量机模型,以对输送带异常图像的撕裂情况和跑偏情况进行进一步细分;改善了现有的输送带撕裂或跑偏场景的识别准确率较低的情况,提高了输送带撕裂或跑偏场景的识别准确率,克服了传统检测方法的及时性、效率性、准确性、可靠性、通用性的缺陷;能够实现高效、精确的输送带检测,及时对设备进行停机维修,对保障能源工厂安全可靠运行具有重要意义。Further, a conveyor belt tearing and deviation detection method is based on a neural network visual detection method to detect the fault of the conveyor belt, train a deep neural network model to extract the image features of the conveyor belt, and classify the normal and abnormal conditions of the conveyor belt. When it is judged to be abnormal, the abnormal image data features are output; then based on the abnormal image data features, the tearing image data features and the deviation image data features are marked, and the hyperparameter combination of the support vector machine model with the highest classification accuracy is obtained. The support vector machine model is trained to further subdivide the tearing and deviation of abnormal images of the conveyor belt; the recognition accuracy of the existing conveyor belt tearing or deviation scene is improved, and the conveyor belt is improved. The recognition accuracy rate of belt tearing or deviation scene overcomes the defects of timeliness, efficiency, accuracy, reliability and versatility of traditional detection methods; it can realize efficient and accurate conveyor belt detection and stop equipment in time Maintenance is of great significance to ensuring the safe and reliable operation of energy plants.
应理解,上述实施例中各过程的执行顺序应以其功能和内在逻辑确定,而不应对本申请实施例的实施过程构成任何限定。It should be understood that, the execution order of each process in the foregoing embodiments should be determined by its functions and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
参照图4,本申请实施例还提供一种输送带撕裂和跑偏检测装置,该一种输送带撕裂和跑偏检测装置与上述实施例中一种输送带撕裂和跑偏检测方法及系统一一对应。该一种输送带撕裂和跑偏检测装置包括:Referring to FIG. 4 , an embodiment of the present application also provides a conveyor belt tear and deviation detection device, which is the same as the conveyor belt tear and deviation detection method in the above-mentioned embodiment. One-to-one correspondence with the system. The conveyor belt tearing and deviation detection device includes:
数据获取模块,用于获取输送带的视频数据,构建图像数据库;The data acquisition module is used to acquire the video data of the conveyor belt and construct an image database;
第一训练模块,用于使图像数据库的数据输入深度神经网络模型中进行训练,并输出异常图像数据特征;The first training module is used to input the data of the image database into the deep neural network model for training, and output abnormal image data features;
第二训练模块,用于基于支持向量机模型的使分类准确率最高的超参数组合,并使标注的异常图像数据特征的撕裂图像数据特征和跑偏图像数据特征输入支持向量机模型进行训练,获得该超参数组合下的支持向量机模型;The second training module is used to combine the hyperparameters with the highest classification accuracy based on the support vector machine model, and input the torn image data features and deviated image data features of the marked abnormal image data features into the support vector machine model for training , obtain the support vector machine model under the hyperparameter combination;
检测模块,用于使训练的深度神经网络模型和最优超参数组合下的支持向量机模型对输送带的视频数据进行检测;The detection module is used to detect the video data of the conveyor belt with the support vector machine model under the combination of the trained deep neural network model and the optimal hyperparameter;
异常感知模块,用于在满足预设条件时,使深度神经网络模型输出特征向量z,并将当前的输送带的视频数据判断为异常;The abnormality perception module is used to make the deep neural network model output the feature vector z when the preset conditions are met, and judge the current video data of the conveyor belt as abnormal;
异常类别预警模块,用于使支持向量机模型基于特征向量z,输出输送带撕裂或输送带跑偏的异常结果,并发出异常类型提醒。The abnormal type early warning module is used to make the support vector machine model output abnormal results of conveyor belt tearing or conveyor belt deviation based on the feature vector z, and send out abnormal type reminders.
进一步地,一种输送带撕裂和跑偏检测装置还包括:Further, a conveyor belt tearing and deviation detection device also includes:
数据增强处理模块,用于对划分的图像数据库的异常类别图像数据的撕裂图像数据和跑偏图像数据进行数据增强处理;The data enhancement processing module is used to perform data enhancement processing on the tear image data and the deviation image data of the abnormal category image data of the divided image database;
数据归一化处理模块,用于使经过数据增强处理的撕裂图像数据和跑偏图像数据,以及图像数据库的正常类别图像数据进行归一化处理。The data normalization processing module is used to normalize the torn image data and deviated image data after data enhancement processing, as well as the normal category image data of the image database.
关于一种输送带撕裂和跑偏检测装置的具体限定可以参见上文中对于一种输送带撕裂和跑偏检测方法的限定,在此不再赘述。上述一种输送带撕裂和跑偏检测装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中,也可以以软件形式存储于计算机设备中的存储器中,以便于处理器调用执行以上各个模块对应的操作。For the specific definition of a conveyor belt tear and deviation detection device, please refer to the above definition of a conveyor belt tear and deviation detection method, which will not be repeated here. Each module in the above-mentioned conveyor belt tearing and deviation detection device can be implemented in whole or in part by software, hardware and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
在一个实施例中,提供了一种设备,该设备包括计算机设备,该计算机设备可以是服务器。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中,该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种输送带撕裂和跑偏检测方法。In one embodiment, an apparatus is provided that includes a computer device, which may be a server. The computer device includes a processor, memory, a network interface, and a database connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The nonvolatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements a method for detecting a conveyor belt tear and deviation.
在一个实施例中,提供了一种存储介质,该存储介质包括计算机可读存储介质,该计算机可读存储介质包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序,处理器执行计算机程序时实现以下步骤:In one embodiment, there is provided a storage medium comprising a computer-readable storage medium comprising a memory, a processor and a computer program stored on the memory and executable on the processor, The processor performs the following steps when executing a computer program:
获取输送带的视频数据,构建图像数据库;Obtain the video data of the conveyor belt and build an image database;
使图像数据库的数据输入深度神经网络模型中进行训练,并输出异常图像数据特征;The data of the image database is input into the deep neural network model for training, and the abnormal image data features are output;
基于支持向量机模型的使分类准确率最高的超参数组合,并使标注的异常图像数据特征的撕裂图像数据特征和跑偏图像数据特征输入支持向量机模型进行训练,获得该超参数组合下的支持向量机模型;Based on the support vector machine model, the hyperparameter combination with the highest classification accuracy, and the tearing image data features and deviated image data features of the marked abnormal image data features are input into the support vector machine model for training, and the hyperparameter combination is obtained. The support vector machine model of ;
使训练的深度神经网络模型和超参数组合下的支持向量机模型对输送带的视频数据进行检测;Use the trained deep neural network model and the support vector machine model under the combination of hyperparameters to detect the video data of the conveyor belt;
当满足预设条件时,深度神经网络模型输出特征向量z,并将当前的输送带的视频数据判断为异常;When the preset conditions are met, the deep neural network model outputs the feature vector z, and judges the current video data of the conveyor belt as abnormal;
支持向量机模型基于特征向量z,输出输送带撕裂或输送带跑偏的异常结果。Based on the feature vector z, the SVM model outputs abnormal results of belt tearing or belt misalignment.
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机程序来指令相关的硬件来完成,所述的计算机程序可存储于一非易失性计算机可读取存储介质中,该计算机程序在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM) 或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other medium used in the various embodiments provided in this application may include non-volatile and/or volatile memory. Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
所属领域的技术人员可以清楚地了解到,为了描述的方便和简洁,仅以上述各功能单元、模块的划分进行举例说明,实际应用中,可以根据需要而将上述功能分配由不同的功能单元、模块完成,即将所述系统的内部结构划分成不同的功能单元或模块,以完成以上描述的全部或者部分功能。Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. Module completion, that is, dividing the internal structure of the system into different functional units or modules to complete all or part of the functions described above.
Claims (10)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111341853.8A CN114187237A (en) | 2021-11-12 | 2021-11-12 | A conveyor belt tearing and deviation detection method, device, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111341853.8A CN114187237A (en) | 2021-11-12 | 2021-11-12 | A conveyor belt tearing and deviation detection method, device, equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114187237A true CN114187237A (en) | 2022-03-15 |
Family
ID=80539952
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111341853.8A Pending CN114187237A (en) | 2021-11-12 | 2021-11-12 | A conveyor belt tearing and deviation detection method, device, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114187237A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114821305A (en) * | 2022-04-02 | 2022-07-29 | 华南理工大学 | Electric work site seat belt identification method, device, equipment and storage medium |
CN117585399A (en) * | 2024-01-19 | 2024-02-23 | 汶上义桥煤矿有限责任公司 | Coal mine conveyor belt tearing detection method based on image processing |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107563431A (en) * | 2017-08-28 | 2018-01-09 | 西南交通大学 | A kind of image abnormity detection method of combination CNN transfer learnings and SVDD |
WO2018032040A1 (en) * | 2016-08-18 | 2018-02-22 | Technological Resources Pty. Limited | A belt rip detection system |
CN108010029A (en) * | 2017-12-27 | 2018-05-08 | 江南大学 | Fabric defect detection method based on deep learning and support vector data description |
CN111517092A (en) * | 2020-06-03 | 2020-08-11 | 太原理工大学 | Transmission belt tearing detection method |
CN111709907A (en) * | 2020-04-26 | 2020-09-25 | 江南大学 | Cloth defect detection method based on one-class deep support vector description |
CN112541905A (en) * | 2020-12-16 | 2021-03-23 | 华中科技大学 | Product surface defect identification method based on lifelong learning convolutional neural network |
CN112699811A (en) * | 2020-12-31 | 2021-04-23 | 中国联合网络通信集团有限公司 | Living body detection method, apparatus, device, storage medium, and program product |
CN113110546A (en) * | 2021-04-20 | 2021-07-13 | 南京大学 | Unmanned aerial vehicle autonomous flight control method based on offline reinforcement learning |
CN113283344A (en) * | 2021-05-27 | 2021-08-20 | 中国矿业大学 | Mining conveying belt deviation detection method based on semantic segmentation network |
-
2021
- 2021-11-12 CN CN202111341853.8A patent/CN114187237A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018032040A1 (en) * | 2016-08-18 | 2018-02-22 | Technological Resources Pty. Limited | A belt rip detection system |
CN107563431A (en) * | 2017-08-28 | 2018-01-09 | 西南交通大学 | A kind of image abnormity detection method of combination CNN transfer learnings and SVDD |
CN108010029A (en) * | 2017-12-27 | 2018-05-08 | 江南大学 | Fabric defect detection method based on deep learning and support vector data description |
CN111709907A (en) * | 2020-04-26 | 2020-09-25 | 江南大学 | Cloth defect detection method based on one-class deep support vector description |
CN111517092A (en) * | 2020-06-03 | 2020-08-11 | 太原理工大学 | Transmission belt tearing detection method |
CN112541905A (en) * | 2020-12-16 | 2021-03-23 | 华中科技大学 | Product surface defect identification method based on lifelong learning convolutional neural network |
CN112699811A (en) * | 2020-12-31 | 2021-04-23 | 中国联合网络通信集团有限公司 | Living body detection method, apparatus, device, storage medium, and program product |
CN113110546A (en) * | 2021-04-20 | 2021-07-13 | 南京大学 | Unmanned aerial vehicle autonomous flight control method based on offline reinforcement learning |
CN113283344A (en) * | 2021-05-27 | 2021-08-20 | 中国矿业大学 | Mining conveying belt deviation detection method based on semantic segmentation network |
Non-Patent Citations (2)
Title |
---|
LI W 等: "Design of online monitoring and fault diagnosis system for belt conveyors based on wavelet packet decomposition and support vector machine", 《ADVANCES IN MECHANICAL ENGINEERING》, 31 December 2013 (2013-12-31), pages 1 - 10 * |
陈桂军: "基于支持向量机的异常检测关键问题研究及应用", 《中国优秀博士学位论文全文数据库-工程科技Ⅰ辑》, no. 2016, 15 August 2016 (2016-08-15), pages 18 - 24 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114821305A (en) * | 2022-04-02 | 2022-07-29 | 华南理工大学 | Electric work site seat belt identification method, device, equipment and storage medium |
CN114821305B (en) * | 2022-04-02 | 2024-06-11 | 华南理工大学 | Method, device, equipment and storage medium for identifying safety belt of electric power operation site |
CN117585399A (en) * | 2024-01-19 | 2024-02-23 | 汶上义桥煤矿有限责任公司 | Coal mine conveyor belt tearing detection method based on image processing |
CN117585399B (en) * | 2024-01-19 | 2024-04-09 | 汶上义桥煤矿有限责任公司 | Coal mine conveyor belt tearing detection method based on image processing |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Shang et al. | Defect-aware transformer network for intelligent visual surface defect detection | |
CN111027058B (en) | Method for detecting data attack of power system, computer equipment and storage medium | |
CN106649789B (en) | It is a kind of based on the industrial process Fault Classification for integrating semi-supervised Fei Sheer and differentiating | |
CN114187237A (en) | A conveyor belt tearing and deviation detection method, device, equipment and storage medium | |
CN106843195B (en) | The Fault Classification differentiated based on adaptive set at semi-supervised Fei Sheer | |
Li et al. | Deep learning-based unsupervised representation clustering methodology for automatic nuclear reactor operating transient identification | |
Wang et al. | A novel tool condition monitoring based on Gramian angular field and comparative learning | |
CN117726240B (en) | A quality evaluation classification method and system based on convolutional neural network | |
TWI649659B (en) | Automatic optical detection image classification method, system and computer readable medium containing the same | |
CN118467930B (en) | Abnormal data processing method applied to robot | |
CN113297885A (en) | Belt conveyor surface state detection method and device based on convolutional neural network | |
CN118710035A (en) | Real-time monitoring method, system and storage medium for electroplating production process | |
CN119004303A (en) | Power plant equipment fault diagnosis method and system based on artificial intelligence and automation | |
CN115861190A (en) | Comparison learning-based unsupervised defect detection method for photovoltaic module | |
Meiners et al. | Process curve analysis with machine learning on the example of screw fastening and press-in processes | |
CN116933170A (en) | Mechanical seal fault classification algorithm | |
Wu et al. | Condition-based monitoring and novel fault detection based on incremental learning applied to rotary systems | |
CN115965625A (en) | Instrument detection device based on visual identification and detection method thereof | |
CN112381110A (en) | Coal mine production multidimensional data abnormal probability prediction method based on fuzzy rule | |
Vijayaraj et al. | Simultaneous Simulation Study of Metal Product Design Based on Machine Learning Algorithm | |
Platzer et al. | On the suitability of different features for anomaly detection in wire ropes | |
Rebekah et al. | AI Based Data Augmentation Process Used to Find the Defect in Injection Molding | |
Santana et al. | Unsupervised fault detection in refrigeration showcase with single class data using autoencoders | |
Zheng et al. | Abnormal temperature detection of blower components based on infrared video images analysis | |
Wan et al. | Multiple recurrent neural networks based fault diagnosis model for multi-mode process |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |