51 results about "Diagnosis Classification" patented technology

The invention discloses an automatic diagnosis method for electrocardiographic abnormality and relates to the technical field of automatic diagnosis of electrocardiographic abnormality. The learning capacity of an RNN neural network to a time sequence and the learning capability of CNN to spatial features are combined to learn features of the electrocardiographic biological signal, and abnormal electrocardiographs of different types are automatically characterized. Then, a deep neural network based classifier is constructed, which is trained using an electrocardiograph with type annotations to improve the accuracy of classification, and automatic classification of different arrhythmia types is achieved. According to the invention, the process of manually extracting features is avoided; then, time sequence features and spatial features of the electrocardiographs are learned using RNN and CNN to form the classifier, and the classifier is trained through supervised learning so that the classifier can automatically diagnose abnormal electrocardiographs. As a result, the accuracy of automatic diagnosis and classification of electrocardiographic abnormality is improved.

The invention discloses a mobile ECG (electrocardiogram) monitoring system and monitoring method. ECG signals are collected; in addition, IR-UWB (impulse radio-ultra wide band) radar signals are collected; after the collected ECG signals and the IR-UWB radar signals are synchronized and processed, a cascade connection CNN (convolutional neural network) is used for feature extraction, integration analysis and diagnosis classification to obtain monitoring results to be output. According to the embodiment of the invention, the IR-UWB radar signals are introduced to be used as the supplement of the monitored ECG signals during the monitoring; through the synchronization on the two signals, the correlation between the two signals is obtained; the monitoring accuracy is improved. The cascade connection CNN with high applicability and stable performance is used for feature extraction, integration analysis and diagnosis classification; the system performance stability is effectively ensured.

The invention discloses a high-voltage circuit breaker mechanical state monitoring and fault diagnosis method. The method comprises the following steps of 1) according to the historical fault data statistics result of high-voltage circuit breakers, carrying out an artificial fault simulation experiment on some common high-voltage circuit breakers to obtain fault data; 2) analyzing the fault current data, and extracting the current data characteristic quantity of each mechanism fault to be used as one of state diagnosis classification basis; 3) analyzing a fault vibration signal, adopting wavelet packet decomposition and sample entropy for processing high-frequency and low-frequency components to obtain corresponding wavelet packet relative energy and sample entropy respectively to be usedas vibration signal characteristic quantities; and 4) performing dimensionality reduction on the vibration signal characteristic quantities through principal component analysis to be used as one of state diagnosis classification basis, and carrying out state diagnosis on the circuit breaker by a support vector machine. By adoption of the method, the multi-dimensional information of the fault statecan be effectively utilized, so that the development of multi-parameter multi-dimensional mapping fault diagnosis of the circuit breaker can be promoted.

The invention discloses a method and system for determining a disease diagnosis standardized coding recommendation list. The method comprises steps that an international disease classification database, electronic records and disease original diagnosis descriptions are obtained, the disease original diagnosis descriptions are preprocessed, the pre-processed disease original diagnosis descriptionsare inputted into a disease diagnosis classification prediction model, and a set of probability values of the pre-processed disease original diagnosis descriptions in each chapter of an internationaldisease classification database are outputted; a first-level candidate disease standard name database is established based on the set of probability values; a second-level candidate disease standard name database is established based on the first-level candidate disease standard name database; semantic similarity of disease standard names in the second-level candidate disease standard name database and the disease original diagnosis descriptions is calculated; the disease standard name coding recommendation list corresponding to the disease original diagnosis descriptions is determined according to the semantic similarity and provided for a coding main body for reference. The method is advantaged in that work efficiency of the coding main body can be improved.

The invention relates to a self-adaption wavelet neural network abnormity detection and fault diagnosis classification system and method, which can be applied to the fields, such as economic management abnormity detection, image recognition analysis, video retrieval, audio retrieval, signal abnormity detection, safety detection, and the like. The system comprises the following seven parts: an acquisition device, a transmitter device, an A / D (Analog / Digital) conversion device, a self-adaption wavelet neural network abnormity detection and fault diagnosis classification processor, a display interaction device, an abnormity alarm device and an abnormity processing device. The abnormity detection and fault diagnosis classification object of the self-adaption wavelet neural network abnormity detection and fault diagnosis classification system is acquired from samples for which a self-adaption mechanism is automatically established by the self-adaption wavelet neural network of a system to be detected, the characteristic information of a signal can be effectively extracted through wavelet transform multi-scale analysis, and a more accurate abnormity detection and fault diagnosis locating result can be obtained. The device adopting the method has the advantages of generalization, high accuracy in the application field, capability of real-time monitoring and low cost.

The invention relates to a transformer fault detection method based on an ant colony algorithm optimization random forest. The method comprises a step of discretizing an initial training sample and calculating a random forest feature importance score, a step of taking the importance score as heuristic information, generating a heuristic distance, and then initializing ant colony algorithm parameters including the node and node feature of each ant, a step of calculating transition probabilities of ants between nodes and constructing a solution space of a feature subset, taking random forest classification accuracy as an evaluation standard, a step of updating a pheromone and simultaneously selecting features to obtain an optimal or approximate optimal feature subset, and a step of satisfying a stop condition, outputting an optimal feature solution and carrying out fault diagnosis classification. According to the method, the improvement of the classification accuracy of a decision tree random forest is facilitated.

The invention discloses a bearing fault diagnosis method combining an improved sparse filter and a KELM. The method comprises the following steps: embedding a Min-Max regular term into an original sparse filter to obtain an improved sparse filter. The Min-Max regular term can describe the internal structure information of the original data, and promotes the similar samples to be close to each other and promotes the samples of different classes to be separated from each other, thereby generating discriminative characteristics. Feature discrimination mainly lies in that class label information is used in construction of the Min-Max regular term, and a pseudo label is used for replacing a real label to guide the construction of the Min-Max regular term. Vibration signals of different operation conditions of a rolling bearing are collected to serve as a training set, the training set is used for training an improved sparse filter model and a kernel extreme learning machine model to obtainmodel parameters, and therefore establishment of a fault diagnosis classification model is completed, and the diagnosis model can accurately identify the rolling bearing fault.

The invention discloses a sparse self-encoding rolling bearing fault diagnosis method. The method specifically comprises the following steps: S1, acquiring original vibration data of a rolling bearingin each fault state, performing linear projection on each kind of vibration data through compressed sensing, and combining compressed signals after linear projection of each fault type into a multi-fault type low-dimensional compressed signal matrix; S2, determining wavelet packet energy entropy of the multi-fault type low-dimensional compressed signal matrix to form a feature vector matrix for bearing fault diagnosis; S3, inputting the feature vector matrix of the rolling bearing under multiple fault types into a sparse automatic encoder for training, and further extracting the weight from an input layer to a hidden layer as a feature matrix; and S4, classifying features extracted by a sparse automatic coding neural network through a neural network classifier, and finishing fault diagnosis classification of the rolling bearing. According to the method, the diagnosis complexity is reduced, the diagnosis time is shortened, and the high diagnosis precision is ensured.

The invention provides a cervical cancer lesion diagnosis method fusing multi-modal prior pathological depth features in the field of medical image processing. The method comprises the following steps: step S10, acquiring a cervical image, pathological definite diagnosis data and annotation information; s20, inputting the cervix uteri image and the annotation information into a deep neural networkmodel for training, and generating a first-stage training result; s30, based on the pathology definite diagnosis data and the first-stage training result, coding the cervix uteri image by adopting asmall network, performing second-stage fusion on the first-stage training result, inputting the first-stage training result into a deep neural network model for training, and generating a second-stagetraining result; s40, determining backbone network parameters, inputting the backbone network parameters into the deep neural network model to perform progressive migration training on the cervical image, and generating a three-stage training result; and S50, carrying out diagnosis classification on the cervical images by utilizing a three-stage training result. The method has the advantage thatthe accuracy and efficiency of cervical cancer lesion diagnosis are greatly improved.

The invention discloses a clinical examination method of an electrocardio intelligent analyzing system. The method comprises the following steps that electrocardio signals are input and preprocessed, wherein after the electrocardio signals are input into the electrocardio intelligent analyzing system, morphology filtering and a self-adaptive threshold value are utilized for removing interference noise to the electrocardio signals; waveform detection and feature extraction are carried out, wherein a morphological analysis method is utilized for detecting and recognizing a QRS wave group, an ST wave band and a P wave band, and electrocardio parameters are output; diagnosis classification and result output are carried out, wherein a clinical database carries out diagnosis classification, and the clinical detection result and the clinical database reference result are compared to obtain the diagnosis result. The clinical examination method of the electrocardio intelligent analyzing system can effectively remove noise interference, achieve automatic diagnosing, and help people to find lesion from electrocardiogram examination in time.

The invention provides an intelligent fault diagnosis method and system for gas pressure regulating equipment, a terminal and a storage medium. The method comprises the following steps: acquiring a gas pressure regulating equipment fault data feature vector of a known fault type, training an SVM algorithm by using a training sample as the training sample of the SVM, and establishing an SVM fault diagnosis classification model by using a Gaussian radial basis kernel function K; optimizing the parameters C and g of the SVM fault diagnosis classification model by adopting a genetic algorithm, andestablishing an optimized SVM fault diagnosis classification model; and utilizing the optimized SVM fault diagnosis classification model to carry out fault diagnosis on the gas pressure regulator toobtain a fault type. The classification model based on the support vector machine is constructed, the fault type is diagnosed, the problems that in the prior art, fault data are few, and high-precision recognition cannot be carried out on the fault type are solved, good learning ability is achieved, and the problems of data small sample, non-linearity, high-dimension classification and the like can be solved. Finally, an implementation scheme of the computer system is provided.

The invention relates to a medical image analysis method, a device, computer equipment and a storage medium. The medical image analysis method comprises the steps of acquiring a medical image video stream which comprises no less than two video frames based on pathological slices; extracting the video frame of the medical image video stream; extracting the single-frame image characteristic of the video frame, mapping the single-frame image characteristic to a single-frame diagnosis classification result; based on a video stream characteristic sequence, performing classified mapping according toa preset mapping rule, and obtaining a target diagnosis classification result; wherein the video stream characteristic sequence comprises the single-frame image characteristic and the single-frame diagnosis classification result of each video frame. Because characteristic extraction and classification mapping are performed based on the medical image video stream for obtaining the target diagnosisclassification result and the medical image video stream comprises no less than two video frames based on the pathological slices, accuracy of the diagnosis classification result can be improved, thereby supplying an accurate basis for medical diagnosis.

The invention discloses an equipment status diagnosis method and an equipment status diagnosis device. The method comprises the following steps: processing an acquired historical sample to obtain a typical sample fault classification and feature values thereof; judging the similarity of to-be-detected samples to obtain a primary fault classification; acquiring the sample feature values corresponding to the primary fault classification, performing weighed calculation to obtain similarity parameters, and determining the fault classification of the to-be-detected samples; when the fault classification is single, determining the fault classification as a fault diagnosis classification of the to-be-detected samples; if not, establishing a precise model, and inputting the primary fault classification into the precise model for judgment to obtain the fault diagnosis classification of the to-be-detected samples. Through the equipment status diagnosis method and device, the purposes of improving the accuracy and precision of diagnosing the equipment status can be realized.

The invention provides an online fan blade damaging real-time diagnosis system and method, and belongs to the technical field of damage diagnosis of fan blades. The system comprises a quad-rotor unmanned plane, a cloud database and a computer system; the quad-rotor unmanned plane captures the images of the fan blades in real time and transmits the images to the computer system in real time; the cloud database stores an image library used for a VGG-19 net image classification method, and images in the image library stored in the cloud database are dynamically captured from a network; the computer system is used for obtaining an improved VGG-19 net image classification method through training of the image library, and the received fan blade images from the quad-rotor unmanned aerial vehicle are classified by using the improved VGG-19 net image classification method to obtain fan blade damage diagnosis classification and damage grade classification results.

The invention belongs to the related technical field of intelligent fault diagnosis of pipe expanding equipment, and discloses an intelligent fault diagnosis method suitable for the pipe expanding equipment, which comprises the following steps: (1) collecting pressure data of the pipe expanding equipment in real time; (2) preprocessing the original pressure data, and dividing the processed data into a training set and a test set; (3) constructing a deep neural network fault diagnosis model based on a stack type denoising sparse automatic encoder improved by a Leacky linear rectification function, and adopting a Softmax function as an activation function of a BP classifier of the deep neural network fault diagnosis model; and training the deep neural network fault diagnosis model by using the training set, inputting the test set into the deep neural network fault diagnosis model, and carrying out diagnosis classification on the test set by using the deep neural network fault diagnosis model so as to predict the fault type, thereby completing the fault diagnosis of the pipe expanding equipment. The production efficiency is improved, and the cost is reduced.

The invention relates to a fault detection system comprising an unmanned plane and a ground server communicating with the unmanned plane; the unmanned plane comprises an obstacle avoidance module, a flight control module, a processor, an image transmission camera, an image transmission module, and a tracking camera; the obstacle avoidance module, the flight control module, the image transmission camera and the tracking camera respectively communicate with the processor; the image transmission camera and the ground server respectively communicate with the image transmission module. The fault detection method comprises the following steps: the unmanned plane flies to a position right in front of a detected object; the processor uses a full contour extraction algorithm to extract the detectedobject contour and determines a scanning direction; the image transmission camera scans a local part of the detected object according to the scanning direction and takes photos; the processor runs alocal contour extraction algorithm and outputs an azimuth between the local part and an imaging plane; the flight control module runs a tracking algorithm, the image transmission camera collects imageinformation of the local part and sends same to the ground server, and the ground server runs a deep nerve network algorithm to analyze the image information, thus finishing fault diagnosis classification.

The invention relates to a wireless sensor network fault diagnosis method based on a time weight K neighbor algorithm, comprising steps of establishing a K-neighbor algorithm training database, sampling WSN state characteristic value to form characteristic vector through timing discrete, wherein each characteristic vector represents the sampling state of the wireless sensor network, performing a pre-diagnosis on a WSN characteristic vector through the K-neighbor algorithm and starting up a time correlation mechanism, starting up a weight amendment rule if the condition is met, and outputting results. The invention can establish the characteristic value according to the system fault mechanism by targeting the wireless sensor network (WSN) system fault diagnosis problem, and can design the fault diagnosis classification rules and parameters based on the weight according to the WSN system fault time correlation, and can establish a system fault diagnosis model by combining with the K-neighbor algorithm to achieve the fact the current diagnosis result is amended according to the diagnosis history. The invention can achieve the fault self-diagnosis and self updating of the WSN, has distributed calculation characteristics and guarantees the accuracy and low power consumption.

The invention discloses a sub-health online recognition and diagnosis method based on performance monitoring data, and belongs to the technical field of fault diagnosis. The method includes the stepsof establishing an initial model of a probability neural network state classification and calculating the threshold standard deviation, carrying out on-line monitoring and diagnosis classification onmonitored equipment by utilizing the current model, and further identifying and extracting sub-health state data and putting the sub-health state data into a sub-health state data set; if the sub-health data set to be recognized reaches a storage tolerance or a known state appears, pausing the storage work, subjecting all elements in the set to K-means clustering analysis to obtain a classification result, and clearing the storage space of the sub-health data set, combining the sub-health state data set after clustering analysis with a previous training sample, and updating to the initial model to obtain a new classification model; repeating the previous steps to identify the sub-health state, and carrying out timely maintenance when the fault state occurs. According to the method, timelyand effective measures are adopted according to the state of the equipment, and loss caused by faults is reduced.

The invention relates to the field of artificial intelligence and medical treatment, in particular to a training method, a diagnosis system and related equipment for a turner syndrome diagnosis model,and aims to improve the accuracy. The training method of the turner syndrome diagnosis classification model comprises the following steps:classifyinginput image samples according to labeling information of data samples, andgeneratingcorresponding training samples and test samples according to a preset proportion; respectively constructing different initial medical classification models based on multiple neural network bases; inputting the training sample into each initial medical classification model for training and parameter adjustment to obtain a corresponding intelligent diagnosis medicalclassification model; and respectively inputting the test samples into each intelligent diagnosis medical classification model for classification, and selecting a turner syndrome diagnosis classification model according to a classification result. The diagnosis system provided by the invention comprises the diagnosis classification model obtained by training through the method, associates the same parts in the photos at different angles, extracts more potential features, and improves the accuracy.

The invention provides a VMD-SSAE-based rolling bearing fault diagnosis classification method. The method comprises the steps of collecting vibration signals of rolling bearings of different fault types; performing time domain, frequency domain and time-frequency domain feature extraction on the vibration signal based on variational mode decomposition; forming a data set by the features, and dividing the data set into a training set, a verification set and a test set; connecting a stack sparse auto-encoder with a Softmax classifier, constructing a VMD-SSAE classification model, and using the training set to train the VMD-SSAE classification model; adopting a grey wolf optimization algorithm and an error back propagation algorithm to optimize the VMD-SSAE classification model, and obtaining an ideal VMD-SSAE classification model; and inputting data in the test set into the ideal VMD-SSAE classification model to obtain a diagnosis classification result. According to the VMD-SSAE-based rolling bearing fault diagnosis classification method provided by the invention, the fault type of the rolling bearing can be diagnosed more accurately.

The invention provides a method for diagnosing bone metastases in nuclide bone imaging based on deep learning. The method relates to a bone scanning diagnosis classification model, a bone metastasis tumor region segmentation model and a bone metastasis tumor load evaluation and automatic report generation model. By the adoption of the method, the bone metastasis tumor can be judged, automatic region segmentation can be conducted, the recognition accuracy is high, and full-automatic analysis from original image input to report generation is preliminarily achieved.

The invention provides a PET / CT (positron emission tomography / computed tomography)-based lung adenocarcinoma and squamous carcinoma diagnosis model training method and device, and aims to assist in training a PET / CT image-based diagnosis classification neural network by using a multi-task learning method and extracting pathological features through a neural network obtained by diagnosis classification training based on pathological images. According to the method, the lung cancer diagnosis classification precision based on the PET / CT image is improved, and meanwhile, the pathological image is only used as priori knowledge in the training process and does not need to be used as network input in practical application. According to the method, through the concept of multi-scale fusion, the precision of the PET / CT image for lung cancer diagnosis classification is improved, the PET / CT image can be further popularized and applied as a means for early lung cancer diagnosis, and help is provided for diagnosis of a clinician on a patient and a subsequent treatment scheme; and meanwhile, the pathology image is used as a priori knowledge auxiliary scheme, the interpretability of the pathology section is further improved, and pathology doctors can further extract pathology features.

The invention provides a neurology clinical nursing potential safety hazard analysis method and system, and the method comprises the steps: obtaining a first pathological index and a second pathological index from a pathological index set of a first user according to a principal component analysis method, inputting the first pathological index and the second pathological index into a first disease determination model, and obtaining the diagnosis disease information of the first user; according to the disease diagnosis information of the first user to the N-th user, constructing a disease diagnosis data set, obtaining a disease motor ability feature, a mental state feature and a language ability feature, obtaining a first root node feature, and combining the disease diagnosis data set to construct a disease diagnosis classification decision tree; obtaining a first classification result according to the diagnosis disease classification decision tree; obtaining different types of neurology clinical nursing potential safety hazard standards, obtaining neurology clinical nursing potential safety hazard standards of the first user, and analyzing and processing the clinical nursing potential safety hazards of the first user. The technical problems that in the prior art, analysis results are not comprehensive enough, and efficiency is low are solved.

The invention discloses a method for mining biomarkers based on multi-map neuroimaging data, relates to a method for identifying graphs, and can consider a high-order complementary relationship amongmultiple maps and sample weight information at the same time. Feature analysis is carried out on the neural image data by adopting a multi-map feature selection method based on weight-induced low-ranklearning. According to the method, a first-order neighborhood aggregation mode is adopted, the sum of all connection strengths of each brain region serves as the characteristics of the brain region,the selected characteristics are more stable in a loop iteration mode, finally, the selected characteristics are fused and classified through a multinuclear support vector machine, and thus the Alzheimer's disease diagnosis precision is improved. The method overcomes the defects that in an existing Alzheimer's disease classification technology, sample weight information and multi-map information cannot be considered, and Alzheimer's disease diagnosis and classification errors are prone to occurring.

The invention provides an auxiliary diagnosis method based on electronic medical record texts, belongs to the field of medical care informatics, and adopts multiple text classification models to perform disease classification on multiple electronic medical record texts respectively. The electronic medical record texts comprise two types, namely a medical record text obtained by single inquiry activity and a medical record text obtained by multiple observations. The invention further provides an auxiliary diagnosis system based on electronic medical record texts. The system comprises a preprocessing unit group and a classification unit group. Through the mode that multiple models correspond to multiple texts, the data property difference of the texts as input data can be considered, so thatthe influence of the data property difference is smaller when the models are used for disease classification, the overall accuracy is higher, the upper limit of the accuracy is higher, and a better diagnosis and classification effect can be obtained more easily at a lower cost.

The invention discloses a deep convolutional neural network-based lung cancer preventing self-service health cloud service system. The system comprises a convolutional neural network used for deep learning and training identification, a segmentation module which segments out a lung region from a CT image based on a full convolutional neural network, a deep convolutional neural network used for lung cancer diagnosis classification, and a self-service health cloud service platform used for performing early prevention and treatment according to an identified suspected lung cancer type. According to the system, the automation and intelligentization level of mobile internet-based lung cancer screening can be effectively improved, more citizens can know and participate in self-service health detection, assessment and guidance, the sensitivity, specificity and accuracy of early lung cancer screening and clinical diagnosis are improved, the lung cancer can be early discovered, early diagnosed and early treated, and the self-health management capability is enhanced.

The invention relates to a method, a device and a system for diagnosing pulmonary embolism based on a plain-scan CT (Computed Tomography) image. Comprising the following steps: acquiring a lung plain-scan CT image of a to-be-diagnosed patient; performing three-dimensional reconstruction on the lung flat scanning CT image to obtain a whole lung image; performing feature extraction on the whole lung image to obtain a feature vector; and inputting the feature vector into a trained machine learning model to obtain a pulmonary embolism diagnosis classification result of the patient to be diagnosed. The invention provides a novel non-invasive acute pulmonary embolism detection method for patients who do not have CTPA or have CTPA detection taboo clinically, and has important clinical application value.

The invention discloses a mechanical equipment diagnosis classification method based on a probability confidence convolutional neural network, and relates to the field of mechanical equipment state monitoring and fault diagnosis. The method comprises the following steps: training a CNN-based diagnosis classification model by taking known state category data of mechanical equipment state monitoringas a training sample, and outputting the probability that the sample belongs to each state category; and calculating the probability confidence of each state category of the diagnosis classificationmodel, testing the diagnosis classification model by utilizing the real-time operation data of the mechanical equipment, and judging the state category of the real-time operation data of the equipmentaccording to the probability confidence of each state category. Self-learning updating of the diagnosis classification model is carried out when an unknown state category appears. Whether the to-be-detected data is in an unknown state or not is judged according to the probability that the CNN outputs each type of state. And when an unknown state occurs, the diagnosis classification model can perform self-learning updating by utilizing the state data, thereby realizing self-adaptive learning of a new state.

The invention discloses a reciprocating compressor fault diagnosis method based on improved ball vector machine closure ball solution acquisition. Data of a reciprocating compressor operated under different working conditions are acquired to serve as a training set, when a ball vector machine algorithm is used for solving a closure ball problem, the dot product between a dot and a sphere center is cached when the training set is searched for the farthest dot, and the dot product is used for calculating the distance between the same dot and the sphere center after the sphere center is updated certain times; when the training set is searched for the farthest dot, part of non-farthest dots are eliminated; the solution of the distance between the dot and the sphere center is not related to support vectors any more through the change of a dot product solution mode and the support vector weights are updated once every certain times; when the number of the support vectors is too large, the times of searches for the farthest dot in a support vector set are increased. Through the strategies, a fault diagnosis classification model can be established within short time, the diagnosis model is detected through the acquired test data, it can be known that the diagnosis model is high in accuracy, and fault diagnosis of the reciprocating compressor can be finished efficiently.