21results about How to "Increase sensitivity" patented technology

The invention provides a power distribution network fault positioning method based on FSGATv2 and topological similarity, and belongs to the technical field of power distribution network fault positioning. The small sample problem of topological structure change in the process from simulation to practical application of the model is solved; firstly, a feature selection module is embedded into a GATv2 network, and the sensitivity degree of the model to fault features is improved. Secondly, classifying the topology of the power distribution network based on topological similarity so as to improve the topological generalization capability of the fault positioning model; meanwhile, a domain self-adaptive transfer learning method for optimizing a dynamic loss function is adopted, so that the distribution difference between data of a source domain and a target domain is reduced, and the problem of few-sample fault positioning is effectively solved; the method is applied to fault location under different operation conditions of the power distribution network.

The invention provides a broadband electric field sensor, an electric field sensor assembly and an electric field detection method.The broadband electric field sensor comprises a substrate, at least one dynamic induction electrode and at least one static induction electrode, the substrate comprises an entity area and a concave area, and the dynamic induction electrodes and the static induction electrodes are arranged at intervals; one part of the dynamic induction electrode is attached to the entity area, the other part of the dynamic induction electrode is located on the concave area, one part of the static induction electrode is attached to the entity area, the other part of the static induction electrode is located on the concave area, and the dynamic induction electrode obtains at least one direct current signal of a current electric field according to an excitation signal; the static induction electrode obtains at least one alternating current signal of the current electric field at least according to charges induced in the current electric field. The direct current signal and the alternating current signal can be obtained at the same time, good detection of a direct current electric field and an alternating current electric field is achieved, the sensitive degree of electric field detection is improved, and the detection frequency band of the electric field sensor is widened.

The invention relates to a large model RAG retrieval method and system oriented to the judicial field, belongs to the technical field of natural language processing, deep learning, large language model and retrieval enhancement generation in artificial intelligence, and is used for enhancing the ability of a large model to retrieve accurate legal texts. According to the method, a large language model is finely adjusted by using instruction adjustment data generated without supervision, and aiming at the specific requirement of a retrieval enhancement generation model (RAG) in the legal field, the attention capability of the RAG on documents with high correlation in legal document retrieval and content generation is enhanced by optimizing the processing step of the model, so that the legal document retrieval and content generation efficiency is improved. Therefore, the performance and the accuracy of the model in the professional field of laws are remarkably improved.

The invention provides a broadband electric field sensor assembly and an electric field detection method.The electric field sensor assembly comprises a substrate, dynamic induction electrodes, static induction electrodes and shielding electrodes, the substrate comprises an entity area and a concave area, and the dynamic induction electrodes and the static induction electrodes are arranged at intervals; the shielding electrode is located between the dynamic induction electrode and the static induction electrode, one part of the dynamic induction electrode and one part of the static induction electrode are attached to the entity area, the other part of the dynamic induction electrode and the other part of the static induction electrode are located on the sunken area, and direct current signals and alternating current signals of a current electric field are obtained respectively; the shielding electrode faces the top of the electric field sensor and is used for shielding interference signals; the data processing part is used for obtaining the direct-current voltage value and the alternating-current voltage value and determining the current electric field type according to the comparison result of the two values. According to the invention, interference signals can be shielded, and the defect that accurate detection cannot be carried out due to interference between the electric field frequency and the induction electrode frequency when an alternating current signal is induced is overcome.

The present invention relates to the technical field of human posture estimation, and solves the technical problem that it is difficult for traditional methods to improve the efficiency of key points on single-scale features, and in particular to a multi-level efficient human posture estimation system and method, the system comprising: extracting features from an input image in sequence, and generating a local feature extraction module, a spatial feature extraction module, and a global feature extraction module for low-level features, intermediate features, and high-level features, respectively. The method comprises: obtaining low-level features, intermediate features, and high-level features of the input image in sequence, and converting the spatial scale of the high-level features into two-dimensional spatial features, and using deconvolution to restore the scale of the two-dimensional spatial features to obtain the key point positions corresponding to the high-level features in the input image. The present invention overcomes the defect that it is difficult for traditional methods to improve the efficiency of key points on single-scale features, can effectively improve the accuracy of key points, and maintain a good running speed.

The invention provides an electric field sensor assembly and an electric field type detection method, and the electric field sensor assembly comprises an electric field sensor and a data processing part, the dynamic induction electrode comprises a first signal generation assembly for generating a direct current signal, a first signal output assembly for outputting the direct current signal, and a first charge induction assembly for inducing charges; the static induction electrode at least comprises a second charge induction assembly for inducing charges and a second signal output assembly for outputting alternating current signals; the data processing part receives the direct current signal and the alternating current signal, obtains a direct current voltage value and an alternating current voltage value respectively, and determines the type of the current electric field according to the comparison result of the direct current voltage value and the alternating current voltage value. The direct current signal and the alternating current signal can be obtained at the same time, good detection of a direct current electric field and an alternating current electric field is achieved, the sensitivity of electric field detection is improved, and the detection frequency band of the electric field sensor is widened.

The application provides an image sensor and a manufacturing method thereof, and relates to the technical field of semiconductors. The image sensor provided by the application constructs a light absorption structure through a groove array and is arranged upstream of a light sensing structure, so that the light absorption structure increases the absorption of an incident light beam through the light absorption characteristics of the groove and guides the incident light beam to the light sensing structure, thereby improving the absorption of the incident light beam by the light sensing structure. The groove array is constructed based on a light filter upstream of the light sensing structure, so that the absorption of the light beam filtered by the light filter is specialized, the absorption capacity of the light sensing structure for the light beam filtered by the light filter is further improved, and the sensitivity of the image sensor to light is improved.

The invention provides an electric field sensor assembly and an electric field detection method, the electric field sensor assembly comprises an electric field sensor and a data processing part, the electric field sensor comprises a substrate, a first induction electrode and a second induction electrode, the first induction electrode and the second induction electrode are identical in structure and can be switched, and the substrate comprises an entity area and a charge induction area; the first parts of the first induction electrode and the second induction electrode are both attached to the entity area, and the second parts of the first induction electrode and the second induction electrode are both located on the charge induction area; the first induction electrode can generate and output a direct current signal or generate and output an alternating current signal according to the excitation signal, and the first induction electrode can generate and output an alternating current signal or generate and output a direct current signal according to the excitation signal. According to the invention, a direct current signal and an alternating current signal can be obtained at the same time, and the first induction electrode and the second induction electrode with the same structure can be switched according to needs, so that the adaptability is improved.

The invention discloses a vehicle-mounted hydrogen system fault diagnosis method and device based on multi-vehicle data cooperation and small sample learning, and belongs to the technical field of fault diagnosis. The method comprises the following steps: based on normal data of a normal vehicle-mounted hydrogen system and abnormal data of a fault vehicle-mounted hydrogen system, determining a target dimension strongly related to a fault in the vehicle-mounted hydrogen system; mixing the normal data and the abnormal data according to a set proportion to obtain a small sample data set; each sample in the small sample data set comprises sample data and a sample mark; the sample data is data corresponding to the target dimension, and the sample mark is a mark for judging whether a fault occurs or not; and training a neural network by using the small sample data set, and predicting data which is uploaded by the to-be-diagnosed vehicle and corresponds to the target dimension by using the trained neural network to obtain a fault diagnosis result. According to the method, the sensitivity degree of the neural network to fault features can be improved under the condition of small sample size, and the accuracy of fault diagnosis is improved.

The invention relates to the technical field of electric power operation, in particular to an electric power high-altitude operation construction falling protector which is used for locking a worker on a guide rail when the worker falls and comprises two sliding seats, and the two sliding seats are connected to the guide rail in a sleeving mode, are arranged in the extending direction of the guide rail at intervals and can elastically slide relative to the extending direction of the guide rail; a locking part is elastically hinged to each sliding seat, each locking part is used for being connected with a safety rope and has a moving state and a locking state, and when the locking parts are in the moving state, the locking parts are separated from the guide rails; and in the locking state, the locking part and the guide rail are locked in a friction mode, and each sliding seat is provided with a sliding locking part. During use, a worker is located between the two sliding seats; when a worker accidentally falls, the locking part located on the upper portion is located above the worker, so that the self-locking function of the falling protector can be triggered when the worker descends by an extremely short distance, and damage to the falling protector and the worker caused by too large kinetic energy when the worker falls can be avoided.

The application relates to the field of heat-sensitive ceramic materials, in particular to a high-stability rare earth-containing six-component low-resistance high-B-value NTC heat-sensitive ceramic material and a preparation method thereof. The raw material scheme of the NTC heat-sensitive ceramic material comprises trimanganese tetraoxide, nickel protoxide, reinforcing material, lanthanum trioxide, combined functional material and trimolybdenum tetraoxide. The final heat-sensitive ceramic material prepared by the application successfully breaks the technical bottleneck that the resistivity and the B value are restricted to each other in traditional NTC heat-sensitive ceramic materials, realizes good unification of low resistance and high sensitivity, significantly improves the accuracy and response speed of temperature detection; meanwhile, the long-term stability of the material under harsh service conditions such as high-temperature storage, humid heat environment and cold and hot impact is greatly improved, the resistance value drift is effectively inhibited, and the strict requirements of high-reliability scenes such as automobile electronics, industrial control and new energy battery management systems on long-term precise monitoring of temperature sensors can be met.

The present invention provides an air quality alarm method and system, which obtains the concentration values ​​of various pollutants collected by sensors in real time, and judges the standard range and concentration change amount of the concentration values ​​of various pollutants in turn, so as to preliminarily screen the health status of equipment and data, and then clusters the concentration values ​​of various pollutants in the pollutant concentration data set to determine the corresponding target cluster centroid, and then calculates the concentration values ​​collected subsequently with the target cluster centroid, and compares the obtained target Euclidean distance with the standard Euclidean distance to analyze the data more accurately. If the target Euclidean distance is greater than the standard Euclidean distance, it means that the data is abnormal, and then an alarm is issued to indicate that the air quality is abnormal. This method effectively improves the sensitivity to abnormal data, and can make alarm prompts in time according to subtle changes in the data.

The invention belongs to the field of fault diagnosis, and particularly relates to a steam turbine fault diagnosis method based on multi-channel time-frequency spectrum cloud model fusion, and the method comprises the steps: carrying out the time-frequency transformation through multi-channel sensor data, and obtaining a plurality of time-frequency spectrums; then, dividing the cloud model into a plurality of sub-regions along the frequency direction, overlapping the sub-regions, and calculating feature parameters of the cloud model of each sub-region, namely a mean value, entropy and hyper-entropy, so as to form three matrixes; three convolutional neural networks are established respectively, the matrixes serve as input of the convolutional neural networks respectively, decision layer weighted fusion is carried out on output, a known state data set is used for training a model, and the output after model fusion is the fault type of the steam turbine; three feature matrixes are calculated according to the above steps for the to-be-tested monitoring data, and the three feature matrixes are input into the trained neural network model to diagnose the fault type of the steam turbine. The fault diagnosis precision can be effectively improved, early warning is provided, and safe and stable operation of the steam turbine is ensured.

The application discloses a kind of ammonia nitrogen micro quick on-line detection device and detection method in environment water body, belong to ammonia nitrogen concentration detection technical field, the device with salicylate spectrophotometry as detection principle, integrated sequential injection system, ammonia nitrogen conversion reaction system, chromogenic reaction system and detection system, realize ammonia nitrogen on-line conversion, effectively improve the anti-matrix interference ability of salicylate spectrophotometry analysis ammonia nitrogen.Salicylate spectrophotometry is used as detection principle in the device, integrated sequential injection system, ammonia nitrogen conversion reaction system, chromogenic reaction system and detection system, realize ammonia nitrogen on-line conversion, effectively improve the anti-matrix interference ability of salicylate spectrophotometry analysis ammonia nitrogen.Ammonia gas absorption chromogenic process does not use gas-permeable membrane, avoid the pollution and replacement problem of gas-permeable membrane.The device quantitative analysis uses absorbance instantaneous change rate as quantitative basis, does not need to wait for chromogenic reaction to end, saves analysis time.The device has the advantages of simple operation, fast, miniaturization, reagent consumption is less, anti-matrix interference ability is strong, etc., is suitable for the on-line detection of ammonia nitrogen in surface water, domestic sewage and other water quality.

The invention provides a semi-supervised relation extraction method for aircraft fault data, which comprises the following steps of: 1, relation classification: extracting semantic information capable of representing a relation between entities from a text so as to realize relation extraction between the entities; the method comprises the following steps: firstly, mining context semantic information of a fault text by using a text encoder; utilizing a relation classifier to realize relation extraction of the unstructured fault data; step 2, pseudo label generation: a pseudo label generation module is adopted to generate pseudo labels for the unlabeled relational data, so that training samples of the relational data are effectively expanded, and the classification performance of a relational classification module is improved; in order to improve the quality of pseudo labels generated by the pseudo label generation module, a meta-learning mechanism is introduced, and the verification loss of the parameters of the relation classification module is used as a meta-learning target to update the parameters of the pseudo label generation module; 3, self-adaptive pseudo label selection is carried out, the step aims to select a pseudo label with higher quality from pseudo label data generated by a pseudo label generation module, and then the performance of relation extraction is improved. According to the method, semi-supervised relation extraction for the unstructured fault data of the airplane can be realized.

The present application discloses an image detection method and related devices and equipment, wherein the image detection method comprises: obtaining a medical image to be tested; performing feature extraction on the medical image to be tested to obtain a first feature map of several dimensions; using the first feature map of preset dimensions as a reference feature map, and using the reference feature map to generate a lesion probability map, wherein the lesion probability map is used to indicate the probability that different areas in the medical image to be tested belong to lesions; fusing the lesion probability map with the first feature maps of several dimensions to obtain a final fused feature map; performing detection processing on the final fused feature map to obtain a detection result about lesions in the medical image to be tested. The above scheme can improve the accuracy of image detection.

The application provides a cursor control method and device in cloud game, electronic equipment and storage medium; the method comprises the following steps: obtaining a cursor control message sent by a first terminal; wherein, in the first terminal, a virtual rocker for controlling a cursor in cloud game is displayed, and a cloud game picture is displayed on a second terminal; analyzing the cursor control message to obtain control parameters corresponding to the virtual rocker; selecting a target position determination mode which is matched with the control parameters from at least two position determination modes, and determining a target position of the cursor in the cloud game picture after moving based on the target position determination mode; rendering a target cloud game picture in which the cursor is in the target position, and sending the target cloud game picture, so that the second terminal displays the target cloud game picture. Through the application, the sensitivity of the virtual rocker in controlling the movement of the cursor can be effectively improved, and the movement precision of the cursor is effectively improved.

The application discloses a wind farm unit reconstruction and energy storage configuration collaborative planning method and system, and relates to the technical field of new energy planning. The method comprises the following steps: obtaining equipment parameters and cost parameters of wind farm unit reconstruction and energy storage configuration, and respectively constructing a technical model and a cost model of the unit reconstruction and the energy storage configuration based on an operation performance optimization target; a wind farm grid connection examination and auxiliary service income calculation model embedded with time sequence operation simulation is constructed, and a quantitative mapping of technical features to income of the wind farm grid connection examination and auxiliary service income calculation model is utilized; based on the quantitative mapping relationship of the technical features to the income, a double-layer optimization framework of flexibility improvement considering investment cost and income is constructed according to the technical model and the cost model of the unit reconstruction and the energy storage configuration. The application can systematically improve the flexibility level and market competitiveness of the wind farm while ensuring safe and stable operation of the wind farm.

The invention provides a deep learning-based household garbage rapid identification method and system. The method comprises the following steps of S1, acquiring a real-time image of a working area of a conveyor belt; s2, performing saliency enhancement processing according to the obtained real-time image to obtain a saliency enhancement feature map; s3, foreground recognition is carried out according to the saliency enhancement feature map, and a foreground recognition result is obtained; s4, segmenting garbage pictures according to the foreground recognition result, and performing classification recognition according to the obtained garbage pictures to obtain a garbage classification result; s5, mechanical arm grabbing position information is obtained according to the foreground recognition result; and S6, a control instruction is generated according to the grabbing position information of the mechanical arm and the corresponding garbage classification result so as to control the mechanical arm to grab the garbage and throw the garbage into the corresponding classification area. According to the invention, the accuracy of garbage classification identification and throwing is improved.

The invention designs an intelligent identification method for a non-meteorological target based on deep learning, and the method comprises the steps: carrying out the recognition of a non-meteorological target in millimeter wave cloud radar data; specifically, a millimeter wave cloud radar data labeling algorithm and a data iteration algorithm are designed to quickly obtain corresponding labels; designing a data expansion method for expanding millimeter wave cloud radar data to obtain more diversified training data, and designing feature extraction and feature reduction to improve the recognition performance of a non-meteorological target; and an excellent identification model is designed by using an encoder-decoder thought. Through the whole set of process, the accuracy of millimeter wave cloud radar non-meteorological target intelligent identification is improved. Furthermore, a prompt enhancement module is constructed to perform prompt enhancement on a specific label, so that the sensitive degree of the model to the specific label is improved, the calculated amount of the intelligent identification model is also reduced, and the generalization ability is improved.

The invention discloses an electret piezoelectric sensing device and a preparation method thereof. The electret piezoelectric sensing device comprises a polarized electricity storage material layer and an electrode layer; the electricity storage material layer comprises a plurality of PTFE compact layers and at least one porous layer which are stacked and tightly combined, the porous layer is arranged between the two compact layers, and the two end faces of the porous layer are compact layers. According to the piezoelectric device provided by the invention, optimized electric field distribution is formed near the hole, a pressure change layer is generated on a bonding interface through hot pressing bonding, the molecular orientation and conformation of the interface are changed, a field plate effect is formed, the sensitive degree of polarized charges to pressure deformation can be improved under the combined effect, and the residence capability of the polarized charges is improved; therefore, by combining the PTFE composite film technology with piezoelectricity, the functional material which is high in performance, durable and suitable for complex environments is manufactured, and the functional material can stably work in severe environments such as high temperature and high corrosivity and has wide application prospects.