43 results about "Non steady state" patented technology

The invention discloses a control system for an efficient energy-saving machine room of a data center, and belongs to the technical field of machine room control systems, and the control system specifically comprises the steps: collecting the air flow track of the machine room and the surface temperature distribution of equipment, and forming an airflow thermal field state set containing a velocity field and a temperature field; the calculation instruction type, the data access mode and the floating point calculation intensity of the server are collected to form a calculation feature set. And carrying out space-time alignment on an airflow track and temperature distribution, establishing an unsteady airflow-thermal field coupling equation set, converting floating point operation intensity into heat source item input, and carrying out iterative solution to obtain a future three-dimensional thermal field evolution sequence. Analyzing the sequence temperature gradient, identifying a heat retention area which is continuously lower than a threshold value, generating a guide plate angle control parameter and a calculation task migration parameter according to the heat retention area, converting the parameters into a driving signal and a task redistribution instruction, and transmitting the driving signal and the task redistribution instruction to execution equipment; the heat dissipation efficiency and the equipment stability are improved by accurately regulating and controlling the equipment parameters of the machine room.

The invention provides a liquid heat conductivity coefficient measurement method without absolute thickness information, and belongs to the field of liquid heat conductivity coefficient measurement. The problem that the universality and the accuracy are seriously restricted due to the fact that an unsteady state method must depend on the absolute thickness of a sample to be detected is solved. The method comprises the following steps: placing a to-be-detected liquid sample in a crucible, and adjusting an optical system to enable an infrared detector to be aligned with a to-be-detected area; changing the thickness of the liquid sample through an electric displacement table to obtain a plurality of thickness values; corresponding to each thickness, using a pulse laser to heat the sample, detecting a signal of the overheat residual temperature of the upper surface of the crucible along with time change, and obtaining a normalized temperature rise curve; repeatedly measuring under a plurality of different liquid thicknesses, and recording the half-peak time of the normalized heating curve corresponding to each thickness value; and calculating the thermal diffusion coefficient of the to-be-measured liquid according to the proportional relation between the half-peak time and the thickness square of the liquid, thereby determining the heat conductivity coefficient of the liquid. The method is used in practical industrial environments such as thermoelectricity and lithium ion batteries.

The objective is to provide a bearing cooling control system for a rotating machine that can appropriately estimate the inner ring temperature even if the rotating machine does not have special equipment for measuring the inner ring temperature, and even if the inner ring temperature and outer ring temperature are in a non-steady state, and that can appropriately control the cooling state of the bearing by the cooling device based on the estimation result. [Solution] The inner ring temperature is measured in relation to the load of the motor 9, the rotational speed of the spindle 7, and the elapsed time t after the change in the rotational speed of the spindle 7. N And the elapsed time t after the change in the operating state of the cooling device 14. C And the outer ring temperature T measured by the outer ring temperature measuring unit 6 out Based on this, one or more inner ring temperature estimation models corresponding to the operating state of the cooling device 14 are used to estimate the inner ring temperature T in The estimated inner ring temperature T in The outer ring temperature T is measured as out Based on this, the operating state of the cooling device 14 or the operating state of the spindle 7 is controlled.

The present application relates to the technical field of data processing and remote maintenance of oil logging instrument, and particularly relates to a remote data calibration method and system for a neutron measurement system. The method comprises the following steps: collecting multi-dimensional time sequence data of downhole instrument, and identifying the current temperature change state; based on the principle of non-steady-state heat conduction, the thermal hysteresis index is calculated by using the temperature change rate and the material thermal saturation limit constant; the real temperature of the crystal is reconstructed by using the thermal hysteresis index to inversely correct the measured temperature of the sensor; and based on the equivalent lattice temperature, the original energy spectrum characteristics are dynamically gain calibrated by using a preset quadratic polynomial multiplication compensation model. By reconstructing the real temperature of the crystal, the present application eliminates the gain hysteresis loop in the up-measuring and down-measuring processes, and significantly improves the interpretation accuracy and consistency of the logging data in a complex variable temperature environment.

The invention discloses a method for fusing OpenFOAM unsteady state simulation result boundary surface Euler field domain data, which comprises the following steps of: firstly, setting a current OpenFOAM working root directory, reading a grid file and creating an example grid; creating global blank file storage boundary surface Euler field domain data; writing a variable name in the first line of the global blank file; specifying and searching names of concerned boundary surfaces, and sequentially reading Euler field domain data of a plurality of boundary surfaces in the time folder; and sequentially writing each boundary surface Euler field domain data corresponding to the variable name into the global blank file according to the boundary surface grid number, and after data reading, fusion and output are completed, printing information on a screen and prompting that the operation is completed until the operation of the program is finished. According to the method for fusing the OpenFOAM unsteady state simulation result boundary surface Euler field domain data, boundary surface physical parameter spatio-temporal evolution data can be efficiently obtained, the processing efficiency is remarkably improved, and a one-stop solution is provided for post-processing of the OpenFOAM unsteady state simulation boundary surface data.

The present application relates to the technical field of industrial automation instrument and intelligent sensor detection, in particular to a pressure transmitter debugging and checking method; comprising parameter acquisition, thermal hysteresis evolution, error calculation and pressure reconstruction module; the method collects temperature and pressure data in non-steady state environment; the core is to calculate the internal temperature by using the filling liquid thermal hysteresis model, combine the transient pseudo pressure with the viscoelastic hysteresis error model, calculate the error caused by thermal imbalance and material hysteresis, and reconstruct the real steady state pressure; the present application realizes the change from static waiting to dynamic prediction, eliminates the thermal balance waiting time through prediction, and greatly improves the production efficiency.

The biosensor system is configured to establish a steady-state condition and alternate between the steady-state condition and a non-steady-state condition to determine an analyte concentration. The biosensor system includes an electrode system having at least one working electrode and one counter electrode. The working electrode is coated with an analyte catalyst layer for converting the analyte to a measurable species. A membrane system surrounds the electrode system and includes an analyte-permeable membrane. The membrane has an analyte permeability with an analyte solubility lower than the analyte solubility outside the membrane. The membrane is configured to capture the measurable species within the membrane such that a steady state of the measurable species generated from the analyte is established near the electrode surface. The bias circuit is configured to apply a potential modulation sequence to the working electrode to induce alternating steady-state and non-steady-state conditions within the electrode system for determining the analyte concentration.

The application relates to the technical field of fault diagnosis, and discloses a reciprocating compressor fault diagnosis method, device, equipment and storage medium, leakage correlation parameters in a cylinder of a target reciprocating compressor are acquired, then a non-steady-state fluctuation baseline formed by non-steady-state fluctuation is fitted in a local weighted regression mode, non-steady-state fluctuation noise is stripped through feature decoupling, pure fault features are extracted, the features are divided according to working condition stages and are adapted to a load sensing model, time sequence feature capturing is strengthened, finally, key features are focused through double-stage attention, a working condition compensation dynamic adjustment classification threshold value is combined, and a diagnosis result is output. Therefore, the application realizes the collaborative optimization of fluctuation noise separation, weak feature extraction and accurate classification of the start-stop transition working condition of the reciprocating compressor based on the combination of feature decoupling stripping of non-steady-state fluctuation noise and working condition stage division and load sensing BiGRU, and solves the problem that a traditional diagnosis method cannot adapt to the non-steady-state characteristics of the start-stop transition working condition.

The invention discloses a microwave amplifier output power detection method and device and a medium, and relates to the technical field of microwave detection, and the method comprises the steps: initializing a microwave amplifier and locking working conditions, generating a detection disturbance control sequence and a switching trigger pulse sequence, generating a dual-state mark and executing direction self-checking, and generating a dimension calibration relation; converting the drain current sequence and the temperature sequence through a dimension calibration relation, generating a dual-state mean value sequence, and obtaining a current steady-state threshold value and a temperature steady-state threshold value; constructing a current time window, calculating a dual-state mean value of the current time window, generating a steady-state judgment value, and obtaining a mismatch strength estimator; branch processing is executed, output power is calculated in a steady state, the output power is kept in an unsteady state, and dimension calibration relation updating is executed when a dimension calibration relation updating trigger condition is met, so that double-state response is clear, disturbance can be identified, working point drift is restrained, and transient disturbance of an amplitude stabilization purpose interface and a protection purpose interface is reduced.

The application relates to the computer technical field and discloses a memory dynamic adjustment method and system, an electronic device and a storage medium, which comprise the following steps: collecting resource performance indexes of a virtual machine and a host computer, forming historical standardized time sequence data, performing normalization processing to obtain standardized time sequence data, filtering abnormal data points exceeding a preset mutation threshold, if it is identified that the virtual machine is in a non-steady-state operation, suspending a prediction process, inputting a double-scale time sequence convolution network model, outputting a future short-term memory usage rate sequence through a short-term prediction branch, outputting a memory change trend label through a long-term trend branch, calculating a memory pressure entropy value based on the future short-term memory usage rate sequence, presetting a multistage constraint rule, generating a memory adjustment instruction, correspondingly triggering a hot plug operation according to the memory adjustment instruction, and executing a rollback mechanism when an adjustment exception is detected. The application can meet the actual needs of efficient, intelligent and self-adaptive dynamic adjustment of virtual machine memory resources.

The invention discloses an initial opening degree determination method and device, electronic equipment and a storage medium. The method comprises the steps that a temperature deviation set is determined according to historical temperature measured data corresponding to continuous sampling points of a stub bar in a non-steady-state time period in a historical batch; determining a target division reference for dividing each historical temperature deviation value in the temperature deviation set, and dividing each historical temperature deviation value in the temperature deviation set according to the target division reference to obtain at least two historical temperature parameter clusters; and for each historical temperature parameter cluster, actual adjusting valve opening values corresponding to the historical temperature deviation values in the historical temperature parameter clusters are obtained, the fitting relation between the temperature and the opening is determined according to all the historical temperature deviation values and the corresponding actual adjusting valve opening values, and the initial opening is determined based on the fitting relation between the temperature and the opening. The valve is controlled at the initial moments of different sub-stages according to the initial opening degree, and the moisture regaining and loosening effects of the craft product are guaranteed.

The application discloses a kind of regularized constraint heat transfer coefficient dynamic determination method and system.The application introduces sine oscillation in jacket temperature, data is divided into sliding window;Linear model containing low-frequency term and simple harmonic term of the same frequency is constructed in window, and the energy balance equation is established by extracting the same frequency complex amplitude;Further, the least square target function with regularization term is constructed with the estimation value of the previous window as prior, and the non-negative constraint based on the interpolation of the boundary before and after reaction is applied, and the UA dynamic evaluation is realized by cyclic solution.The application effectively decouples non-steady-state drift and noise, suppresses numerical mutation, obtains smoother and more physical law heat transfer coefficient and heat release rate result, and significantly improves the robustness of complex reaction process calorimetry.

The invention provides an isolation evaluation method and device for a distributed storage system, equipment and a medium. In the technical scheme provided by the invention, a centralized isolation evaluation center is introduced and a cluster steady-state sensing mechanism is combined; isolation evaluation logics originally dispersed on OSDs are collected to an isolation evaluation center elected by a state monitoring assembly to be processed in a unified mode, all isolation requests can be executed after being subjected to serialization evaluation through the isolation evaluation center, and when the isolation evaluation center judges that the distributed storage system is in an unstable state according to the state of a global PG fragment, all the isolation requests can be executed; according to the method, whether the redundancy of each isolated PG still meets the lowest availability requirement of the redundancy strategy adopted by the PG to which the PG fragment belongs is strictly evaluated, the situation that part of PGs cannot normally provide read-write response due to cross-fault-domain and multi-point concurrent isolation caused by multi-node decentralized decision is avoided, and the data security and service continuity of the system are improved.

The invention relates to a power battery state-of-charge estimation method and device and a storage medium, and the method comprises the steps: S1, obtaining the predicted values of an open-circuit voltage and a terminal voltage at a t moment through a second-order RC equivalent circuit model; s2, obtaining a predicted value of the heat production quantity at the moment t by utilizing a heat production equation; s3, obtaining a predicted value of the battery temperature at the t + 1 moment by using the single-state thermal model; s4, obtaining an SOC predicted value at the t + 1 moment by using the first prediction model; and S5, calculating the terminal voltage at the same moment and the difference between the measured value and the predicted value as an observation error, and inputting the observation error from the t-k moment to the t + 1 moment, the battery parameters, the SOC predicted value and the SOC correction value from the t-k moment to the t moment into the time sequence prediction model to obtain the SOC correction value at the t + 1 moment. Compared with the prior art, the SOC prediction precision under the unsteady state working condition can be improved.

The invention discloses a traction rectifier open-circuit fault diagnosis method under an unstable working condition, and belongs to the technical field of circuit fault diagnosis. The method comprises the following steps: establishing a control model of a two-level rectifier, and introducing a switching function to describe the conduction state of a power device; calculating a bridge arm upper side current estimation value based on the AC side current and the switch state, comparing the estimation value with an actual value to obtain a residual error, and constructing a first fault detection function; meanwhile, a second fault detection function is constructed by analyzing positive and negative half cycles of the alternating current side current; the two are combined to form an open-circuit fault diagnosis function to realize accurate positioning of a fault device; and whether the load changes can be detected by analyzing the current peak change. The method does not need an accurate system model and a large amount of training data, is suitable for an unsteady state working condition, has the advantages of high diagnosis speed, high accuracy and strong anti-interference capability, and is suitable for real-time diagnosis of the open-circuit fault of the two-level rectifier in a traction transmission system.

The application discloses a kind of electric power system non-steady-state carbon emission perception calculation method, device, equipment and medium, and the technical field belongs to low-carbon electric power technical field.The electric power system non-steady-state carbon emission perception calculation method includes: obtaining system data;Wherein, system data includes line data, unit data, load data and unit dynamic carbon emission coefficient under non-steady state in unit start-stop process;The corresponding unit combination problem model of electric power system is established and Dantzig-Wolfe decomposition is carried out, and the convergent restricted main problem model is obtained;The affine parameter solution of convergent restricted main problem model is calculated, and the extended marginal carbon perception value is calculated according to affine parameter solution;The modified value is obtained by multiplying the extended marginal carbon perception value and the correction factor, and the extended positioning average carbon perception value is obtained by Riemann numerical integration on the modified value.The application can realize the accurate measurement of electric power system carbon perception value under non-steady state considering unit start-stop, deep peak shaving and the like.

The application discloses a kind of data center efficient energy-saving machine room control systems, belong to machine room control system technical field, specifically include: the air flow trajectory of machine room and equipment surface temperature distribution are collected, form the air flow thermal field state set containing velocity field, temperature field;The computing instruction type of server, data access mode and floating-point operation intensity are collected, and computing feature set is formed.Non-steady-state air flow-thermal field coupling equation set is established by aligning time and space of air flow trajectory and temperature distribution, and the heat source term input is converted to floating-point operation intensity, and the future three-dimensional thermal field evolution sequence is obtained by iterative solution.Analysis sequence temperature gradient, identify the heat retention area that continues to be below threshold value, according to this generate angle control parameter and computing task migration parameter of guide vane, convert into drive signal and task reassignment instruction transmission to execution equipment;The application can improve the heat dissipation efficiency and equipment stability by accurately regulating and controlling the machine room equipment parameters.

The invention provides a self-adaptive dynamic setting method for cold continuous rolling additional tension under an unsteady rolling condition, and relates to the technical field of mechanical automation control. The method comprises the following steps: firstly, based on obtained strip steel hot rolling-cold rolling process information, constructing a cross-process cold continuous rolling unsteady state rolling force prediction model; and then an additional tension setting function in the cold continuous rolling unsteady-state speed increasing and decreasing rolling process is constructed with the minimum rolling force fluctuation as the target, finally, the optimal additional tension setting value of each rack is obtained through an artificial intelligence algorithm and is fitted into an additional tension curve, and collaborative setting of the tension of each rack is achieved. Compared with a traditional additional tension setting method, additional tension setting is dynamically adjusted according to the set and actually measured rolling force deviation, speed disturbance in the unstable rolling process can be responded in real time, the rolling stability of multiple low-speed and variable-speed stages is improved, the strip steel head thickness deviation and strip shape defects are effectively reduced, and the yield of the strip steel is improved. And the head quality consistency and the yield of the strip steel are guaranteed.

The invention relates to the technical field of industrial automation instrument and intelligent sensor detection, in particular to a pressure transmitter debugging and checking method. A parameter acquisition module, a thermal lag evolution module, an error calculation module and a pressure reconstruction module are included; the method comprises the following steps: acquiring temperature and pressure data in an unsteady state environment; the core of the method is to calculate the internal temperature by using a filling liquid thermal hysteresis model, solve errors caused by thermal imbalance and material hysteresis in combination with a transient pseudo pressure and viscoelastic hysteresis error model, and reconstruct real steady-state pressure; according to the method, the conversion from static waiting to dynamic prediction is realized, the heat balance waiting time is eliminated through prediction, namely verification, and the production efficiency is greatly improved.

The application discloses a method for constructing a relative permeability curve of a power-law non-Newtonian fluid, which comprises the following steps: first, a non-steady-state relative permeability curve displacement experiment device is built; then, rock sample pretreatment, non-Newtonian fluid rheological parameter testing and displacement experiment data acquisition are completed; the outlet water breakthrough time, the cumulative liquid production and the displacement pressure difference of the non-Newtonian fluid when the constant-speed displacement of the non-Newtonian fluid is recorded; after the experimental data is preprocessed, a relative permeability curve model is preset, the water cut is implicitly solved and the derivative is calculated, and the injection pore multiple of the non-Newtonian fluid and the displacement pressure difference are derived; based on the experimental data, the water saturation-water cut and the displacement pressure difference-water saturation relationship are simultaneously fitted through an iterative optimization algorithm, and finally, the coefficients in the relative permeability curve formula are determined, and the relative permeability curve is constructed. The method improves the smoothness and physical rationality of the relative permeability curve of the non-Newtonian fluid, improves the calculation accuracy, and provides a reliable means for analyzing the seepage characteristics of complex fluids.

The application discloses a synchronous reluctance motor MTPA control method based on an optimized online extremum search method and relates to the field of motor control. The method comprises the following steps: applying the extremum search method to the maximum torque / current ratio (MTPA) of a synchronous reluctance motor (SynRM); under a steady-state condition, constructing a lookup table by using an optimized current vector angle obtained by online extremum search; under a subsequent similar working condition, directly calling data in the lookup table to avoid a repeated search process and improve control efficiency; when the motor operation does not reach a steady state, performing interpolation calculation by using the existing data in the lookup table to estimate the MTPA current vector angle under the current working condition, thereby extending the MTPA control strategy to a non-steady-state condition. The method aims at the challenges in the traditional MTPA control strategy due to the nonlinearity of parameters and the dependence on a steady-state condition. By combining the online extremum search method and the real-time lookup table construction, the method improves the dynamic response speed, extends the application to a non-steady-state condition, and significantly improves the efficiency and performance of the SynRM.

The present invention provides a system and method for performing a single-scan rest-stress cardiac measurement. In one aspect, the system includes a positron emission tomography (PET) imaging system, a source of a first PET radiotracer for administration to a subject, a source of a second PET radiotracer for administration to a subject, and a processor. The processor has non-transient computer readable media programmed with instructions to obtain PET images of the subject administered with the radiotracer. Furthermore, the computer readable media is programmed with instructions to process the PET images with a non-steady-state, multi-compartment parametric model. An output of the non-steady-state, multi-compartment parametric model is a measure of myocardial blood flow for both a rest state and a stress state of the subject.

The invention relates to the technical field of working condition monitoring of long oil pipelines, and provides an identification method and device for an unsteady operation working condition of an oil pipeline, and the corresponding method comprises the steps: obtaining a time sequence of production operation parameters of the oil pipeline; according to the characteristics of the time sequence and a pre-generated unsteady-state operation condition identification model, the unsteady-state operation condition of the oil pipeline is identified; wherein the unsteady-state operation condition identification model is generated based on a dynamic graph neural network model, nodes of the unsteady-state operation condition identification model are determined by the number of time sequences, and whether edges exist in different nodes are determined by mutual information of the time sequences; the features of the time sequence are extracted by the long short-term memory network according to sub-time sequences of the time sequence under multiple dimensions. The method can effectively capture the delay correlation between the pump stations of the oil pipeline, solves the problem of space-time asynchronization in the prior art, and has high pipeline operation condition state recognition capability.

The invention provides a method and a system for rapidly predicting the gas-water two-phase yield of a compact water-containing gas reservoir fractured well, and the method comprehensively considers the influence of the slippage effect, the stress sensitive effect, the starting pressure gradient and the gas high-pressure physical property of gas phase seepage in a compact reservoir on a seepage mechanism. Respectively establishing gas phase correction pseudo-pressure and pseudo-time and water phase correction pseudo-pressure and pseudo-time to realize linear processing of nonlinear factors, and determining a two-phase unsteady-state seepage model; processing model parameters by adopting implicit pressure and explicit saturation, and deducing a corresponding analytical solution; determining a tight reservoir utilization range, and establishing a corresponding gas-water two-phase material balance equation for resolving average formation pressure and average water saturation to update seepage model parameters; therefore, the two-phase yield prediction of the tight reservoir in the full life cycle is realized. According to the scheme, the problems of insufficient calculation precision and low efficiency in the prior art can be solved, accurate and rapid prediction of the two-phase yield of the tight gas fractured well is achieved, and technical support is provided for efficient development of a tight water-containing gas reservoir.

The application provides an optimization method for an arc compression section of an energy recovery linear accelerator, comprising the following steps: determining the bunch length and the initial projection emittance of the bunch at the entrance and exit of the arc compression section, and obtaining an overall transmission matrix; dividing the arc compression section into a pre-compression section and a second compression section, taking the slice centrifugal deviation caused by the coherent synchrotron radiation effect of the second compression section as an optimization target, establishing a reverse optimization beam line as a means to optimize the reverse beam line of the second compression section; establishing a constraint condition of the pre-compression section, and optimizing the pre-compression section; and adding a transition matching section between the pre-compression section and the compression section to obtain the arc compression section. The method can make the bunch deflect a large angle while compressing the bunch, and can suppress the emittance growth caused by the steady-state CSR effect and part of the non-steady-state CSR effect, so that the emittance growth is reduced while the non-ideal bunch is compressed.

The invention relates to the technical field of fault diagnosis, and discloses a reciprocating compressor fault diagnosis method, device and equipment and a storage medium, and the method comprises the steps: obtaining a leakage related parameter in a cylinder of a target reciprocating compressor, and then fitting an unsteady fluctuation baseline formed by unsteady fluctuation in a local weighted regression mode, the method comprises the steps of stripping unsteady fluctuation noise through feature decoupling, extracting pure fault features, then dividing features according to working condition stages, adapting a load sensing model, strengthening time sequence feature capture, finally focusing key features through two-stage attention, dynamically adjusting a classification threshold in combination with working condition compensation, and outputting a diagnosis result. Therefore, the cooperative optimization of fluctuation noise separation, weak feature extraction and accurate classification under the start-stop transition working condition of the reciprocating compressor is realized on the basis of the combination of feature decoupling stripping unsteady fluctuation noise, working condition stage division and load sensing BiGRU; the problem that a traditional diagnosis method cannot adapt to the unsteady-state characteristic of the start-stop transition working condition is solved.

The application relates to a steady-state-based multi-threshold on-demand equipment frequency modulation method, which comprises the following steps: acquiring the current actual frequency and actual load of equipment in a steady-state judgment period; updating a steady-state counter and a state identifier based on the actual frequency, the actual load, a steady-state tolerance coefficient and a steady-state period judgment threshold value to obtain equipment state; determining a target frequency increase threshold value and a target frequency decrease differential threshold value based on the equipment state; determining a target frequency based on the size relationship between the current actual load of the equipment and the target frequency increase threshold value and the target frequency decrease differential threshold value; and controlling the equipment based on the target frequency. The application can reduce the energy consumption of the equipment in a steady state and improve the response speed of the equipment in a non-steady state.

The invention belongs to the technical field of power control, and particularly relates to a flexible direct current power transmission system cooperative control method and system considering distributed power supply access, computer equipment and a storage medium. Obtaining a DC bus voltage deviation value, a distributed power supply cluster system power fluctuation quantity and a system frequency change rate; acquiring a system operation state according to a comparison relationship between the system frequency change rate and a preset threshold range; and when the frequency change rate of the system exceeds a preset threshold range, judging that the system is in an unstable running state, and starting a virtual inertia compensation mode. The method can solve the problems of asynchronous instruction execution and phase deviation of a dynamic process, is more convenient for cooperative control, can quickly calculate the optimal power distribution scheme of each converter station when the system is in unsteady-state operation, achieves synchronous execution of each converter station, effectively inhibits system fluctuation, and improves the system reliability. The problems of insufficient inertia support, unreasonable power distribution and weak fault ride-through capability are solved, so that the requirement of high-proportion distributed power supply grid connection is met.

The invention discloses a soft reduction method for improving the internal quality of an unsteady-state casting blank. During unsteady-state casting, the action response speed of a roller is 0.2-0.3 mm / s during soft reduction of a fan-shaped section. According to the method, by improving the action response speed of the fan-shaped section and reasonably arranging the soft reduction roll gap in the unstable casting transition interval, continuous and stable transition of the actual roll gap in the unstable area is achieved, the situation that the strain of a casting blank is too large due to instantaneous change of the roll gap in the unstable casting process is avoided, and the soft reduction effect is guaranteed; according to the method, equipment changes stably, and the impact of roll gap changes on the equipment is obviously reduced; the macroscopic quality of the obtained casting blank is obviously improved, middle cracks are basically eliminated, and the proportion of center segregation class C is increased to 98.6% from 49.3%; the flaw detection qualification rate in the unsteady-state stage is increased to 95.2% from 57.4%, meanwhile, the casting blank in the unsteady-state stage achieves direct rolling of a steel plate within 50 mm according to the original specification, and the judgment change amount of the casting blank is reduced by 90%.

The application belongs to the technical field of fault diagnosis, and provides a complex equipment non-steady state condition fault self-adaptive diagnosis method and system. In order to solve the problems that the existing transfer learning method faces in the non-steady state condition, such as coarse global distribution alignment granularity, difficulty in adapting to multi-stage dynamic drift and the like, a fine-grained domain adaptation mechanism based on local maximum mean difference is designed, a graph Laplacian spectral distance constraint is constructed to maintain the topological structure consistency of the cross-domain feature dependency relationship, and the collaborative optimization of the source domain and the target domain in the sub-domain level is realized, so as to realize high-precision and high-robust cross-domain fault diagnosis under the non-steady state conditions such as time-varying speed.