374 results about "Normal state" patented technology

The invention relates to an anti-ant large-core-number fiber ribbon cable comprising a cable core and an outer sheath. The cable core includes loose tubes; and fiber ribbons are sleeved inside the loose tubes. The outer sheath is an outer nylon sheath; and a non-metallic reinforcement layer is arranged between the outer sheath and the cable core. According to the anti-ant large-core-number fiber ribbon cable provided by the invention, with the outer nylon sheath, ants can not bite the sheath, so that the damages on the optical cable by ant biting in a pipe at an area with a serious anti problem are prevented, the normal state of the cable is ensured, and the application region of the cable is extended. The cable based on a simple and reasonable structure has advantages of small cable diameter, high fiber density, large fiber capacity, and low duty ratio and can be laid in a pipe conveniently; and the bending performance is good. During branched connection, 4 to 24 fibers can be melted once, so that the construction time is saved and the manpower cost is lowered, and the construction efficiency is improved. On the basis of the full drying type dielectric structure, branched connection is realized conveniently and usage becomes convenient.

The invention provides a production control method and system based on SPC alarm linkage process locking, and relates to the technical field of quality management and control, and the method comprises the steps: deploying an SPC sensing node used for collecting technological parameters at a cutting edge position where a machining tool of production equipment is in contact with a workpiece, or at a contact point where a probe of online measurement equipment is in contact with the surface of the workpiece; the method comprises the following steps: acquiring process parameter data of a key production process in real time through an SPC sensing node deployed at a cutting edge position or a contact point; inputting the process parameter data as real-time data or small sample expansion data into an SPC judgment engine; and performing statistical process control analysis on the real-time data or the small sample expansion data through an SPC judgment engine so as to model data clusters which are densely distributed and have stable parameters in the data into a virtual reference curved surface representing a normal state domain. According to the invention, the accuracy of abnormity determination can be improved, and timely management and control of quality abnormity and minimization of production loss can be realized.

The invention provides an equipment fault early warning method and system, and belongs to the technical field of fault prediction, and the method comprises the steps: obtaining the real-time working condition parameters of equipment; the real-time working condition parameters comprise real-time operation parameters and environment parameters; extracting a first feature vector based on the real-time working condition parameter; respectively calculating the similarity between the first feature vector and a plurality of second feature vectors in a model library; determining the corresponding normal state model with the maximum similarity as a target model; and inputting the real-time working condition parameters of the equipment into the target model to obtain the deviation between the real-time working condition parameters and the normal parameter range, and carrying out fault early warning on the equipment based on the deviation. According to the equipment fault early warning method and system provided by the invention, the accuracy of equipment fault early warning can be improved.

The invention relates to the technical field of mechanical equipment fault diagnosis, and particularly discloses an attention mechanism CNN-based electric actuator fault feature extraction method and system. The method comprises the following steps: through a mode of combining simulation modeling and field acquisition, obtaining a fault feature of an electric actuator; obtaining multi-source data of the electric actuator in a normal state, a known fault state and an unknown fault state, and preprocessing and enhancing the multi-source data; and constructing a lightweight convolutional neural network model containing a cross-modal attention mechanism. According to the method, multi-source signals such as current, voltage and displacement can be effectively fused, weights of different modes and characteristics are automatically distributed, and the extraction capability of key fault characteristics is enhanced; accurate distinguishing between known faults and unknown faults is realized through multi-scale feature complexity measurement and a dynamic discrimination mechanism; in combination with domain self-adaption and an incremental learning mechanism, the model can migrate between simulation and real working conditions, and has self-expansion capability for a novel fault mode.

To provide a game machine that improves management of game media number display according to the game state.SOLUTION: A game machine of the present invention has three game states: a first normal state, a second normal state, and a specific game state. Control can be performed so that the degrees of advantage for auxiliary games differ between the first and second game states. A game machine information notification command containing information indicating the game state can be generated and transmitted to virtual game media number control means. Standby processing can be executed when a situation in which a response command from the virtual game media number control means cannot be received occurs a predetermined number of consecutive times. When it is determined that a power interruption has occurred, after storing a checksum in a used area of the memory of main control means and a back flag to be referenced when recovering from the power interruption, the standby processing can be executed. A power-on operation accompanied by a RAM clear clears the backup information in the main control means, but does not clear the virtual game media number which can be played.SELECTED DRAWING: Figure 40

The invention discloses a silicon-based new material compressor fault detection method based on multi-source heterogeneous data, and the method comprises the steps: firstly obtaining normal historical operation data of vibration, pressure and temperature through a compressor state detection platform, carrying out the data preprocessing, and constructing a training set; then, a star-graffiti optimization algorithm is adopted to carry out joint optimization on variational modal decomposition (VMD) parameters and decompose signals, and modal features are extracted; secondly, realizing cross-channel adaptive fusion through convolutional neural network (CNN) multi-scale coding and introduction of channel and time attention; performing time sequence modeling on the fusion feature sequence by adopting a bidirectional long short-term memory network (BiLSTM), and training an anomaly detection model through training set data; and finally, inputting the test set data into the initial anomaly detection model through the same process, and realizing fault judgment by calculating the deviation from a normal state. And when the evaluation index of the test set reaches the standard, solidifying the model into an optimal anomaly detection model for on-line reasoning application of on-site real-time data.

The invention provides a vehicle control method and device, electronic equipment and a storage medium, and the vehicle control method comprises the following steps: when it is detected that a user triggers a crawling mode activation instruction for a target vehicle, determining state detection information of the target vehicle; when the state detection information of the target vehicle represents that the vehicle is in a normal state, responding to the crawling mode activation instruction to enable the working mode of the target vehicle to be switched into a crawling mode; and when the state detection information of the target vehicle represents that the vehicle is in an abnormal state, refusing to respond to the crawling mode activation instruction. On one hand, when the vehicle is in the normal state, the crawling mode activation instruction can be normally responded, so that the vehicle can achieve the crawling mode, on the other hand, when the vehicle is in the abnormal state, the crawling mode activation instruction can be refused to be responded, a user is prevented from mistakenly touching the crawling mode, and therefore the driving safety of the vehicle is improved.

Disclosed are an intelligent monitoring method and system for a condenser vacuum system of a thermal power unit. The method comprises: collecting operating data of a condenser vacuum system, and performing preprocessing; extracting features from the preprocessed data, and performing anomaly recognition; and on the basis of an anomaly detection output result, adjusting operating parameters of the vacuum system. In the monitoring method and system provided in the present invention, temperature features reflect the heat exchange effect of a system, quality features reflect the operating load and cooling effect of the system, and vacuum system features reflect the overall operating state of the vacuum system; when an anomaly score S exceeds 0.3, the automatic adjustment of the system is realized; and an adjustment amount is calculated on the basis of the current anomaly score and the features, thereby ensuring that the system quickly restores to a normal state.

The invention provides a magnetic superconducting memory and a preparation method and a read-write method thereof. According to the magnetic superconducting memory, the superconducting layer maps the magnetization state by using the resistance difference between the superconducting state and the normal state; the ferromagnetic layer serves as a free layer capable of adjusting magnetization, and the magnitude of superconductive critical current of the superconductive layer is influenced through the magnetization direction. The heavy metal layer generates spin orbit torque through a spin Hall effect to drive the ferromagnetic layer to be magnetized and overturned; an infinite switching ratio between a superconducting state and a normal state is realized through a superconducting diode magnetoresistance effect, a high-efficiency and high-reliability nonvolatile storage scheme is realized in combination with a current-driven spin-orbit torque magnetization overturning technology, inherent defects of a traditional TMR are thoroughly overcome, and the signal contrast of a storage unit is remarkably improved; the heterostructure designed by the invention replaces a traditional magnetic tunnel junction, reduces material stacking and process complexity, and adopts a material compatible with a CMOS (Complementary Metal-Oxide-Semiconductor Transistor) process to promote large-scale integration of the low-temperature memory.

The invention discloses a phase deviation testing method based on a rotation test of a self-aligning rolling bearing, relates to the technical field of bearing detection, and solves the technical problems that potential risks in a normal state are lack of dynamic monitoring, a real-time operation state and a risk level are not combined, and the deviation between a prediction result and the reality is large. According to the method, by analyzing the stability of the periodic offset characteristic and the association rule of the periodic offset characteristic, the rotating speed and the load, an abnormal reason can be directly positioned, a basis is provided for targeted maintenance, blind shutdown and maintenance cost are reduced, and for a bearing in a normal state, the change rate of a key index is calculated in real time by establishing a phase-deviation-free reference database, so that the reliability of the bearing is improved. And comparing with a preset comparison value, so that a potential risk can be found before a fault occurs, precautions are taken in the bud, and a residual life calculation formula is dynamically corrected based on a risk early warning level in combination with a basic rated dynamic load, an actual equivalent dynamic load and a working condition correction coefficient, so that a prediction result is more fit with an actual operation state of equipment.

To provide a diagnostic device or the like for improving accuracy of gas leakage determination of a steam trap.SOLUTION: A diagnostic device includes: measurement means that measures vibration of a trap to be diagnosed and calculates first vibration data from measurement data; estimation means that estimates a drain amount of the trap to be diagnosed on the basis of the first vibration data; acquisition means that acquires a pressure value of the trap to be diagnosed; storage means that stores first relationship information in which a pressure value of the trap in a normal state in which there is no leakage of gas is associated with a determination threshold value, and second relationship information in which a pressure value of the trap in a state in which there is leakage of gas, and a drain amount is equal to or greater than a predetermined amount is associated with the determination threshold value; first determination means that specifies a determination threshold value based on the acquired pressure value on the basis of the first relationship information and performs leakage determination of the gas to be diagnosed on the basis of the first vibration data and the determination threshold value; and display means that displays a determination result by the first determination means and the drain amount. The first determination means executes the leakage determination again on the basis of the second relationship information in response to the satisfaction of a predetermined condition.SELECTED DRAWING: Figure 10

The invention relates to the technical field of sensors, and discloses a signal optimization method based on a diagnosability evaluation result, and the method comprises the following steps: S1, obtaining monitoring signal data collected by a plurality of candidate sensors in a normal state and a plurality of fault states of equipment, and S2, carrying out the optimization of the signals based on the monitoring signal data, according to the method, a fault detectability model is constructed, the fault detectability model is used for quantifying the difference between a fault state and a normal state on sensor signals, and when sensor signal optimization is carried out, the fault detectability model and the fault isolability model are constructed in a unified manner; and a detectability evaluation index and an isolability evaluation index are calculated based on a joint mean-variance difference value, so that comprehensive quantitative evaluation of the signal fault detection capability and the fault isolation capability is realized, and it is ensured that performance requirements of different dimensions of fault diagnosis in a signal optimization process are balanced. And the comprehensiveness and accuracy of fault diagnosis are improved.

The invention relates to a wirelessly usable sensor system, wherein the sensor system is suitable for being attached to the skin of a person and comprises the following components: (a) a measuring sensor which is suitable for measuring the oxygen saturation or the tissue perfusion of the skin covered by the sensor system and / or of the underlying tissue and outputting said oxygen saturation or tissue perfusion as a measured value; (b) an evaluation unit which is connected to the measuring sensor, is suitable for evaluating the measured value output by the measuring sensor, for comparing said measured value with a reference value and for outputting a signal if the measured value deviates from the reference value by a predetermined amount, and is suitable for learning the normal state of the tissue of a patient, for identifying patterns in the measured values measured by the measuring sensor by searching for repeating regularities in the sensor data and analyzing deviations from the determined normal state over a certain period of time, and for outputting a signal if the pattern deviates from a regularity; (c) a rechargeable energy store which supplies energy to the measuring sensor and to the evaluation unit and to further optional electronic components, and (d) a housing in which the components (a) to (c) and further optional electronic components are contained, wherein the housing is closed and impermeable to liquid such that the components (a) to (c) and further optional electronic components are protected from liquid. The sensor system is preferably used in medicine in decubitus prophylaxis.

A control unit includes plural control devices configured to control drive of an electric motor. The plural control devices have general functions used for operating the electric motor both during a normal state and during a fault state, the normal state being where the plural control devices is working properly, the fault state being where any one of the plural control devices is having a fault. The plural control devices have normal-time functions distributed among the plural control devices, the normal-time functions being for use during the normal state but unnecessary during the fault state.

The invention discloses an intelligent defect detection system and method applied to a PCBA (Printed Circuit Board Assembly), and relates to the technical field of big data analysis. By dividing mutually independent sub-detection areas, feature signal interference among different bonding pad areas is eliminated, and a detection range is accurately anchored; the first weight value is determined by counting the proportion of each characteristic exceeding threshold value of the abnormal bonding pad data, the second weight value is calculated by combining with the local time sequence sample data intercepted by the sliding window, and the dynamic weight value is obtained, so that the global historical data universality and the local time sequence data timeliness are balanced; the characteristic reference drift caused by technological parameter fluctuation and consumable batch difference is inhibited; an analysis threshold value is generated by calculating a feature floating value, a fine system of defect judgment and approaching abnormity early warning is constructed, the limitation of single-dimension judgment of a traditional fixed threshold value is broken through, samples which do not meet defect standards but deviate from a normal state are recognized, the problems of detection misjudgment and missed judgment are solved, and the precise quality control requirement of the dynamic production working condition of a production line is met.

The invention provides a node control method which can be applied to the fields of big data and distribution. The node control method comprises the following steps: acquiring state data uploaded by a first actuator and at least one second actuator respectively, the state data being generated by independently monitoring front nodes in one-to-one correspondence based on each actuator, the front nodes corresponding to the first actuator and the second actuator and the target system are in a connection activated state and a connection disabled state respectively; when it is detected that the state data uploaded by the first actuator is abnormal, a third actuator is determined from the second actuators with normal state data based on a preset performance evaluation rule or an operation instruction of a user; and controlling the first actuator to set the connection between the corresponding front node and the target system to be in a forbidden state, and controlling the third actuator to set the connection between the corresponding front node and the target system to be in an activated state. The invention further provides a node control method and device, equipment, a medium and a program product.

The invention discloses a surface-mounted permanent magnet synchronous motor short-circuit fault diagnosis method under a servo working condition, belongs to the field of servo motor fault diagnosis, and aims to solve the problems that harmonic-wave-based surface-mounted permanent magnet synchronous motor short-circuit fault characteristics under the servo working condition are influenced by working conditions, the diagnosis speed is low, and the accuracy is not high. The method comprises the following steps: 1, constructing a motor mathematical model of a servo working condition, deducing a q-axis current analytic expression, and calculating and generating a decoupled theoretical current characteristic signal based on a theoretical coordinate transformation relation; 2, collecting the three-phase current and the rotor position of the motor in real time, and extracting an actual current characteristic signal through a coordinate transformation chain including Clarke transformation, Park transformation, coordinate stretching transformation and Servo-Park transformation; and 3, calibrating the average value and the standard deviation of the actual characteristic signals in the normal state of the motor as a health reference. And 4, during online diagnosis, comparing an actual characteristic signal extracted in real time with a health reference, and carrying out fault judgment according to a preset statistical threshold value.

The control section 152 determines that the starting state is the normal state when the time TA is within a reference range decided based on a reference time decided in advance. The control section 152 determines that the starting state is the misalignment state when the time TA is outside the reference range. The time TA is the time from when the LB signal changes from the OFF state to the ON state until when the DZ signal changes from the ON state to the OFF state.

The embodiment of the invention provides a wind power master control PLC safe operation method and system based on fault diagnosis, and the method comprises the steps: building a multi-dimensional digital model which can represent the overall health state of a wind turbine generator under variable working conditions through deep analysis of massive historical operation data; specifically, the model evaluates the consistency between the actual running state vector of the current unit and the dynamic health domain defined by the model in real time. An anomaly score is generated by quantifying the degree of deviation between the actual state and the health baseline, any statistically significant deviation will be identified as a potential early fault signal. According to the method, normal state drifting and real fault germination under the extreme working condition can be intelligently distinguished from the system level, so that the problems of fault false report and fault missing report of a static threshold strategy are solved, and accurate early-stage fault early warning and grading are realized.

According to the technical scheme, the garbage can with the pressing ring comprises a can body and the pressing ring buckled on the top of the can body, and is characterized in that the top of the can body extends outwards to form an annular positioning table, the pressing ring comprises an inner side wall and an outer side wall, and a containing cavity allowing the top of the can body to be inserted is formed between the inner side wall and the outer side wall. The outer side wall of the pressing ring extends downwards to the position below the annular positioning table, the left outer side and the right outer side of the pressing ring extend out of mounting bases, transverse through holes are correspondingly formed in the positions, corresponding to the mounting bases, of the outer side wall of the pressing ring, and sliding blocks and compression springs are mounted in the mounting bases. Under the normal state, the front end of the lower sliding block penetrates through the transverse through hole under extrusion of the compression spring and then is clamped on the bottom surface of the annular positioning table for limiting; after the sliding blocks are pulled outwards, the sliding blocks overcome spring force to move outwards, the front ends of the sliding blocks retreat from the bottom face of the annular positioning table, and the sliding blocks and the pressing ring are lifted upwards to separate the pressing ring from the barrel body. When being fixed, the pressing ring can be locked on the garbage can body so as to press and fix the garbage can, and when being taken out, the pressing ring can be easily separated from the can body, so that a garbage bag can be conveniently taken out.

The utility model relates to the technical field of water meters, in particular to a rotary wing type water meter convenient to observe, which comprises a character wheel box, the character wheel box comprises a box body and a character wheel metering mechanism, the box body is provided with an inner cavity, the character wheel metering mechanism is arranged in the inner cavity, the character wheel metering mechanism comprises a rotatable character wheel set, the side wall of the character wheel set is provided with numbers, and the number of the character wheel set is larger than the number of the character wheel set. The character wheel box further comprises a cover body, a glass sealing cover and a turning cover, the cover body is provided with a reading face, a set of numbers on the side wall of the character wheel set are exposed on the reading face, the reading face is arranged in an inclined mode, the glass sealing cover seals a cavity opening of the inner cavity, the turning cover is installed on the box body in a rotating mode, and after the turning cover rotates by a preset angle, the turning cover covers or exposes the glass sealing cover. According to the utility model, the reading surface is arranged in an inclined manner, and the flip cover covers the glass sealing cover in a normal state, so that the advantages of reading in a side-looking manner and effectively preventing dust and stains from blocking the reading area are realized.

The invention discloses an online fault diagnosis method for a thermal resistor of a temperature measurement element, and the method comprises the following steps: S1, data collection and preprocessing: collecting a measurement resistance value (Rt) of the thermal resistor and a corresponding environment temperature (Tenv) in real time, and obtaining a stable resistance value (Rs) through digital filtering processing; s2, initial health state diagnosis: comparing the stable resistance value (Rs) with a theoretical resistance value (Rc) calculated based on the current environment temperature (Tenv), and if the deviation exceeds a first preset threshold value (delta R1), triggering an initial diagnosis process; and S3, preliminary fault diagnosis: starting a parallel diagnosis process comprising at least two different diagnosis sub-modules after triggering the preliminary diagnosis, not only detecting the occurrence of the fault, but also distinguishing possible types of the fault (such as slow response, continuous drifting, instantaneous interference and the like), and providing fault confidence, so that the fault diagnosis accuracy is improved, and the fault diagnosis accuracy is improved. And more accurate guidance is provided for maintenance decision making.

To provide a gaming machine that improves the management of the game state. [Solution] The gaming machine of the present invention has a first normal state, a second normal state, and an auxiliary game advantageous state with a high degree of advantage regarding auxiliary games as game states, advantageous auxiliary games are more likely to be executed when in the auxiliary game advantageous state, the first normal state is maintained when a special losing symbol stops in the first normal state, the first normal state is maintained when a special losing symbol stops in the second normal state, it is possible to transition to the first normal state or the auxiliary game advantageous state, if predetermined termination conditions are met in the auxiliary game advantageous state, it is possible to transition to the second normal state, information during consecutive jackpots can be set when in the auxiliary game advantageous state, the first type jackpot state, and the second type jackpot state, information during consecutive jackpots can be set even when the auxiliary game advantageous state ends, while an advantageous auxiliary game is being executed, and the game presentation means can execute a specific game mode presentation while an advantageous auxiliary game is being executed.

The invention discloses a roller passing state monitoring method and device for electrode section equipment, and relates to the technical field of equipment state monitoring. The method specifically comprises the following steps that current time sequence data output by a servo driver corresponding to a main driving roller is collected in real time to serve as an original monitoring signal reflecting the dynamic characteristics of a roller passing system; based on the acquired current time sequence data, performing signal processing on the current time sequence data to extract a real-time feature vector capable of quantitatively representing a current system state; according to the extracted real-time feature vector, the extracted real-time feature vector is compared with a preset reference model representing the normal state of the roller passing system so as to determine the deviation degree of the current state compared with the normal state; after the deviation degree is determined, whether the deviation degree exceeds a preset threshold value or not is judged, and when the deviation degree exceeds the threshold value, it is finally determined that the state of the roller passing system is abnormal, and an early warning signal is output. The invention aims to carry out real-time and effective on-line monitoring on the overall health state of a large-scale roller passing system in a low-cost mode.

The invention provides a plastic-lined pipe processing method and device and a medium, and the method comprises the steps: carrying out the prediction processing of the temperature data of data collection points on an outer pipe and an inner pipe of a to-be-processed plastic-lined pipe in a target time period, so as to obtain a plurality of first prediction data and a plurality of second prediction data corresponding to the current moment; determining a first matching weight and a second matching weight according to fluctuation characteristics of a plurality of temperature data of each data acquisition point of the inner pipe of the to-be-processed plastic-lined pipe at the current moment, so as to obtain a processing characteristic value corresponding to the to-be-processed plastic-lined pipe at the current moment; and if the processing characteristic value is within a preset processing normal characteristic value range, determining that the thermal processing state of the to-be-processed plastic-lined pipe at the current moment is a normal state, so as to determine the thermal processing state of the to-be-processed plastic-lined pipe at the current moment by respectively performing data prediction on an outer pipe and an inner pipe of the to-be-processed plastic-lined pipe. And a worker can control the quality in the processing process of the plastic-lined pipe.

The present disclosure provides a rolling-bearing monitoring device and method enabling more proper peak detection. The present disclosure includes determining a frequency spectrum of vibration data indicating vibration of a rolling bearing for every sampling period, storing the frequency spectrum in association with the sampling period into a storage unit, and detecting, from the frequency spectrum, a specific peak not appearing during a normal state within a first frequency range including a theoretical frequency at which a peak occurs during an abnormality. When a specific peak is detected, the present disclosure includes detecting a specific peak as a re-detection specific peak within a second frequency range, which includes the specific peak and is smaller than the first frequency range, from at least one frequency spectrum stored in the storage unit prior to this sampling period, and causing a display unit to display the specific peak and the re-detection specific peak.

The application discloses a bearing life prediction method based on an attention mechanism and a residual neural network, comprising the following steps: firstly, using time domain features and a k-means clustering algorithm to adaptively identify the degradation state of the bearing; using an empirical mode method to decompose the bearing data signal and remove noise; and training and using a residual neural network with an introduced attention mechanism to complete the life prediction of the bearing. The application has the following advantages: the degradation state of the bearing is adaptively identified, avoiding the interference of data generated in the normal state on the life prediction; the empirical mode method is used to decompose the bearing data signal and remove noise, preventing the mutual interference and cancellation of degradation and fault features at different frequencies, and thus making the features learned by the model more effective; and the residual neural network with the introduced attention mechanism is used, so that the neural network model is always affected by the original signal and the attention is focused on the channel where the fault features are located, preventing the model from overfitting.

The application discloses a kind of based on BIM's concrete construction temperature field dynamic visualization method and system, the method includes: obtaining the measured temperature data of multiple measuring points in concrete structure, and the measured temperature data is format standardization, time stamp matching and abnormal value preprocessing;According to the unique identification of each component in the BIM model of concrete structure, the measured temperature data after preprocessing is automatically bound to the corresponding component;For each component that has bound temperature data, calculate the relative position of its current measured temperature in the preset temperature interval, and linearly map the transparency parameter of the component according to the relative position;For each component, based on its historical measured temperature time series, the rate of change of temperature with time is calculated using linear regression method, and the predicted temperature value of future short-term time is predicted accordingly;According to the relationship between the current measured temperature and the predicted temperature value relative to the early warning threshold, the outline color of the component is determined;Through the graphic overlay interface of Revit API, the calculated transparency parameter and outline color parameter are applied to the corresponding component, only cover the transparency and outline color of component surface, do not modify the original material, texture and surface properties of component;The above method performs transparency mapping in normal state, and preferentially performs color intensification display in abnormal state, to realize the coordination and unity of continuous trend expression and abnormal highlight display.

The invention discloses an industrial edge rolling technology unknown fault early warning method and system based on unsupervised learning. The method comprises the steps that multi-dimensional time sequence track data in the operation process of an edge rolling robot is acquired; dividing the data into independent track segments; extracting a group of statistical features covering a centralized trend, a dispersion degree and a distribution pattern from each track segment to construct a high-dimensional feature vector; a sorting method based on feature importance is adopted, a key feature subset with the most discriminative power is screened out from all the features, and therefore optimal feature representation is constructed; training a deep auto-encoder model only by using sample features in a normal state, so that the deep auto-encoder model learns the accurate reconstruction capability of normal feature vectors; during state judgment, a reconstruction error between a new sample feature vector and a model reconstruction result is calculated to serve as an abnormal score; and a decision threshold is scientifically determined through receiver operation characteristic curve analysis and a Youden index maximization criterion, and the abnormal score is compared with the threshold, so that real-time and automatic judgment of the process state is realized.

A motor control device includes a plurality of control units corresponding to a plurality of windings (14, 15) provided in a motor (12). The plurality of control units includes a master control unit (20), a slave control unit (30), and at least one abnormality monitoring unit (53, 62, 64). The slave control unit (30) is configured to, after the abnormality monitoring unit (53, 62, 64) detects a communication abnormality between the control units, before determining the communication abnormality, generate a power supply control command value using information acquired through a path other than communication between the control units and maintained in a normal state in which the communication abnormality between the control units is not detected for controlling power supply to the windings (14, 15), and perform backup control for maintaining the power supply control command value.