42results about How to "Accurate Anomaly Detection" patented technology

A method for the detection of anomalies in a networked water distribution system is provided. It improves detection methods based on an iterative modification of control variables of the network, by determining a reduced set of entities of the water distribution network on which control variables should be iteratively modifier. The invention increases the computing costs, and the reliability of such methods of detecting anomalies in a water distribution system.

A device for supplying power to an inductive load includes a switching structure designed to control a current in the load, and elements for detecting anomalies designed to generate information on detection or information on non-detection of an anomaly of the short-circuit type able to occur in the cabling toward the load, in combination with information on validity of the information on non-detection of anomalies. The information on anomaly non-detection is delivered without setting the validity information if the measured current at the end of an appropriate time window is less than a given value of current.

The invention discloses an outdoor fire hydrant monitoring system and a detection method thereof. The system includes a measurement module, a control module, a communication module, a power supply module, a data management and data analysis module, and a terminal access module. The measurement module uses a number of high-precision sensors to measure together, especially the micro-turbine flowmeter used. When the error accuracy is less than 1%, the cost of 20 yuan can realize the function of a large-scale flowmeter. The valve opening and closing sensor used not only effectively provides water resource safety protection, but also quickly responds to water demand. The communication module effectively distributes the counting time of the IoT terminal equipment evenly within the period of the counting cycle to avoid possible network congestion. The data management and data analysis module is based on machine learning anomaly detection algorithm and cluster analysis algorithm to realize intuitive interpretation of fire hydrant data features, accurate detection of anomalies, and unified management of classification. Effectively realize the scientific management of fire hydrants, comprehensive and timely monitoring, and reduce the difficulty of management.

The present invention provides a method and device for abnormality detection. The method includes the following steps: acquiring a target detection image of a workpiece to be detected and a good product image of a good product; acquiring a first feature map according to the target detection image, and obtaining a second Feature map; obtain the cosine similarity according to the first feature map and the second feature map; perform the first abnormality detection on the workpiece to be detected according to the cosine similarity; if it is impossible to judge whether the workpiece to be detected is abnormal, then obtain the first Grayscale image, and obtain the second grayscale image according to the good product image; obtain the first segmentation image according to the first grayscale image, and obtain the second segmentation image according to the second grayscale image; obtain the second segmentation image according to the first segmentation image and the second segmentation image Obtain the index score; perform a second anomaly detection on the workpiece to be detected according to the index score. The invention can accurately detect the abnormality of the workpiece to be detected, has a wide application range, and does not need to consume a lot of manpower, material resources and time costs.

In order to overcome the problem in the prior art that the selection of cluster members is likely to cause a high rate of false alarms in detection results, the present invention provides a mobile user abnormal behavior detection method, which uses a sliding window to dynamically acquire data to improve the accuracy of user behavior acquisition. Accuracy, the concept of Duun_index is introduced after the initial clustering and incremental stages of the traditional FC algorithm, the cluster members generated after the increment are selected, and then the selected high-quality members are fused with the voting algorithm to obtain the final result, and then combined with When the user's normal behavior is compared with the similarity, the correlation matrix is ​​introduced, and the change of the average difference is used to judge whether the user's behavior is normal, so as to achieve the purpose of efficient and accurate anomaly detection.

The present invention relates to a semi-supervised time series anomaly detection method and system, and constructs an autoencoder model based on a long-term short-term memory network. The autoencoder model includes an encoder, a normal flow data decoder and an abnormal flow data decoder. Select normal labeled traffic dataset and unlabeled traffic dataset in the traffic time series data set, and use two training sets to train the autoencoder model. There is no need to predefine a threshold in advance. For unlabeled data, by comparing two The size of the reconstruction error of each decoder can be used to judge whether it is abnormal. The present invention avoids the difficulty of selecting the optimal threshold, and can accurately detect anomalies. It also adopts a sliding window to perform enrichment processing of abnormal flow data on unmarked flow data sets, which solves the problem of rare abnormal points and enriches abnormal data. , further improving the anomaly detection rate.

An anomaly detection apparatus 100 includes an image transformation unit 103 that calculates an image transformation parameter, based on an inspection image in which an inspection object appears, a reference image indicating a normal state of the inspection object and a parameter for image transformation parameter calculation, and performs image transformation on the inspection image using the image transformation parameter, an image change detection unit 104 that collates the reference image and the image-transformed inspection image using a change detection parameter, and calculates an anomaly certainty factor indicating whether there is a change in a specific region of the inspection image, a change detection parameter learning unit 106 that learns the change detection parameter, based on a difference between a training image indicating a correct answer value of the change and the anomaly certainty factor, and an image transformation parameter learning unit 108 that learns the parameter for image transformation parameter calculation, based on a collection amount derived from the difference between the training image and the anomaly certainty factor and to be applied to the inspection image that has undergone image transformation.