118 results about "State-transition equation" patented technology

An intelligent control method for vertical recovery of carrier rockets based on deep reinforcement learning is disclosed. A method of autonomous intelligent control for carrier rockets is studied. Theinvention mainly studies how to realize attitude control and path planning for vertical recovery of carrier rockets by using intelligent control. For the aerospace industry, the autonomous intellectualization of spacecrafts is undoubtedly of great significance whether in the saving of labor cost or in the reduction of human errors. A carrier rocket vertical recovery simulation model is established, and a corresponding Markov decision-making process, including a state space, an action space, a state transition equation and a return function, is established. The mapping relationship between environment and agent behavior is fitted by using a neural network, and the neural network is trained so that a carrier rocket can be recovered autonomously and controllably by using the trained neural network. The project not only can provide technical support for the spacecraft orbit intelligent planning technology, but also can provide a simulation and verification platform for attack-defense confrontation between spacecrafts based on deep reinforcement learning.

The invention discloses a lithium ion battery service life prediction method based on traceless particle filtering, and the method can achieve more accurate estimation of the capacity state of a battery and improves the prediction accuracy of the service life of the battery. The method comprises the steps: enabling a double-index capacity attenuation model as a lithium ion battery capacity degradation model, and obtaining a state transfer equation and measurement equation of the lithium ion battery capacity; obtaining the distribution of the initial value of a state variable of the double-index capacity attenuation model according to the known service life attenuation data of other batteries; determining a corresponding prediction starting point for a to-be-measured battery, wherein the service life of the to-be-measured battery needs to be predicted; carrying out the state tracking of to-be-measured battery capacity data of charging and discharging times through employing a traceless particle filtering method, updating the state variable in the capacity attenuation model, and obtaining a corresponding state variable after the charging; predicting the corresponding state variable and battery capacity after the charging and discharging, plotting a capacity prediction curve, and determining the service life of the to-be-measured battery.

The invention relates to an energy storage optimal configuration method considering system gas and thermal power units. The method comprises the following steps: firstly, according to historical data,forecasting the output of a new energy generator set, constructing a typical scene set of the output of the new energy generator set, and combining the load fluctuation characteristics, constructinga load scene set of a power system; This paper analyzes the operation characteristics of gas-fired units and thermal power units in the start-up and stop stages, establishes the state transfer equations, defines the state transfer conditions, and establishes the state transfer models of gas-fired units and thermal power units in the start-up and stop stages, so as to realize the transfer and switching between different states of gas-fired units and thermal power units in the start-up and stop stages. The paper also analyzes the operation characteristics of gas-fired units and thermal power units in the start-up and stop stages. According to the system and operation parameters, considering the wind power consumption objective, the optimal energy storage allocation model considering the climbing ability and multi-stage state transition of gas turbine and thermal turbine is constructed, with the objective of minimizing the relevant investment and total operating cost. Solve the above power system energy storage optimal allocation problem, and obtain the power system energy storage optimal allocation scheme.

The invention discloses a hybrid measurement based power distribution network dynamic state estimation method, which comprises the steps of reading current network parameters and a network topology structure of a power distribution network, and forming a node admittance matrix and a branch-node association matrix; reading SCADA data and PMU data of each node of the power distribution network, configuring a mapping relation between the two kinds of data, and processing differences between the two kinds of data; building a state transition equation and a measurement equation which are adapt to PMU/SCADA hybrid measurement; adding a fault-tolerant mechanism to a traditional extended Kalman filter method, and performing state estimation on the power distribution network by adopting the improved extended Kalman filter method. Dynamic state estimation is applied to hybrid measurement, and the accuracy of state estimation is improved. Meanwhile, an adaptive data detection mechanism is introduced, and the accuracy of dynamic state estimation is improved.

The invention relates to an on-line sequence limit learning machine method possessing an autonomous learning capability and belongs to the intelligent robot technology field. Nine portions are included and comprise an external state set, an external motion set, a rewarding signal, a value function, a state transition equation, a limit learning machine network hidden layer output set, an intermediate parameter transfer equation, a limit learning machine output set and a limit learning machine output weight transfer equation. In the on-line sequence limit learning machine method possessing the autonomous learning capability, an on-line sequence limit learning machine is taken as a framework and Q learning is combined and reinforced. The on-line sequence limit learning machine method possessing the autonomous learning capability is provided and a model is used in mobile robot path program research so that a robot realize autonomous learning navigation according to an external environment state and rewarding. An autonomous learning capability of the robot in an unknown environment is increased.

The invention provides an acquiring method and system of queuing wait time, which are suitable for the technical field of control. The acquiring method of the queuing wait time comprises the following steps of: acquiring the average arrival rate of customers and the average service rate of a service desk, and respectively recording as lambda and mu; acquiring the state transition information of the customers according to the average arrival rate of the customers and the average service rate of the service desk, wherein the state transition information of the customers accords with state transition equations of the customers; analyzing and calculating the state transition equations of the customers to acquire the wait time of the customers, and outputting and displaying. Through the embodiment of the invention, the accurate queuing wait time for the customers is acquired by acquiring the average arrival rate of the customers and the average service rate of the service desk, acquiring the state transition information of the customers, which accords with the state transition equations of the customers, according to the average arrival rate of the customers and the average service rate of the service desk, analyzing and calculating the state transition equations of the customers to acquire the wait time of the customers and outputting and displaying, and therefore, the satisfaction degree of the customers is enhanced.

The present invention discloses a pedestrian detection method for the traffic environment based on the human body tree model, and belongs to the field of road traffic pedestrian detection. The method comprises: selecting a data set with annotation information of the human body joint as a training sample of the model, and expanding the joint into the required human body part; based on the relative position relation between each parent part and child part, using principles of the relative distance of the sample, the mean value of the sample correlation difference and the mean value of the total correlation difference of the sample set, optimizing the initial clustering center of the K-means algorithm to realize the clustering of the various parts of the human body, and obtaining hidden variables of the training samples; using a coordinate reduction method to solve the hidden structure SVM problem, and training, obtaining, detecting and determining the models; in the detection phase, according to the constructed human tree structure, the part state transition equation and the off-line training model, merging the dynamic planning idea to realize the traversal of the pyramid of the test sample, obtaining the whole human body detection result of the image, and using a non-maximal suppression algorithm to obtain the final detection bounding box.

The invention relates to a development automatic machine with a brain cognition mechanism and a learning method of the development automatic machine, and belongs to the technical field of intelligent robots. The development automatic machine comprises an inner state set, a system output set, an inner operation behavior set, a state transfer equation, a reward signal, a system evaluating function, a system action selecting probability and a Dopamine response differential signal. By means of the development automatic machine and the learning method, a mathematic model high in generalization ability and wide in application range is provided for the system autonomous development process with the learning automatic machine as the basic frame; by means of the method, a sensorimotor system and an intrinsic motivation mechanism are combined, the self-learning and self-adaption capacity of the system is improved, and the intelligence in the true sense is achieved.

The invention provides an error joint estimation method for a GM-EPHD filtering radar system based on ADS-B data. Firstly, the ADS-B observed values of a target is converted in a rectangular coordinate system with the position of a radar station as a center thereof to establish an ADS-B and radar observation equation of the target. A state transition equation of the target and the radar system error is established in the linear gaussian condition, and the radar system error is expanded to a target state to form an extended state. The gaussian hybrid-probability assumed density filtering on the target state is conducted by utilizing the ADS-B observed values after the coordinate transformation process, so that an accurate estimation on the target state is obtained. After that, the gaussian hybrid-probability assumed density filtering on the target state is conducted on the extended state after the updating of the target state by utilizing radar observation data. Furthermore, the joint estimation of the target state and the radar system error is conducted by a two-step kalman filter. Finally, the fusion estimation result of the radar system error is obtained through the weighted average algorithm. The method is high in estimation accuracy and good in estimation performance.

The invention provides a battery life prediction method based on prediction model fusion and a storage medium. According to the battery life prediction method, a long and short memory network model isnested in a particle filtering model; fusion model with simple structure, training a long and short memory network model by using existing historical data to obtain a degradation trend equation, anddetermining a state transition equation of a particle filtering model; the problem that a particle filtering model excessively depends on an empirical model is solved; the particle filtering model canobtain the uncertain expression of the residual life by using the weighted sum of the particles to approach the predicted value of the capacity, and in addition, a new sample obtained on line is added to the original training sample set to retrain the model, so that the model parameter is updated in time, the adaptability is better, and the prediction of the residual cycle life of the cadmium-nickel storage battery can be realized.

The invention discloses a train multi-interval operation curve rapid optimization method. The method comprises the steps that S1, data, including train parameters, line data and whole line operation time, of a train number model to be optimized are acquired; S2, setting speed curve optimization parameters; S3, preprocessing the train parameters and the line data in the step S1; S4, calculating thestate space, the state transition equation and the state transition cost in each interval; S5, connecting the state spaces of the intervals into a whole, and performing dynamic programming algorithmsolution in combination with a weight adaptive adjustment method to obtain a dynamic programming optimization result under the constraint of the whole line operation time; and S6, analyzing and outputting a dynamic planning optimization result to obtain a multi-interval energy-saving operation speed curve and the distributed operation time of each interval.

The invention relates to a lithium ion battery remaining life prediction method based on grey particle filter, which can use a small data amount for aging modeling, accurately estimate the battery capacity state and improve the battery life prediction accuracy. The method comprises steps: battery capacity data which are needed for prediction are firstly extracted as sample data to be inputted to agrey prediction model, the grey evolution coefficient a for capacity degradation of the lithium ion battery is calculated, and thus, a state transfer equation and an observation equation for the lithium ion battery capacity during the aging process are built; a particle filter algorithm is then used to track and update the battery capacity state changes; and finally, when prediction begins, according to the grey evolution coefficient, the change value of the capacity of each particle along with the cycle number is extrapolated, and according to the weight of each particle, the remaining lifeof the battery is predicted, and probability density distribution is given.

The invention discloses a load curve quantization-based pumped storage power station optimized dispatching method. The method comprises the following steps: arranging a set repairing plan; collecting load predetermination data of a power system; establishing a state transition equation of a pumped storage power station and a constraint condition thereof; establishing a load curve quantization index of a thermal power generating set; and using load curve quantization index minimization of the thermal power generating set as an objective function and carrying out optimized dispatching on the operation of the pumped storage power station by using a dynamic planning method. According to the invention, the fluctuation degree of a load curve carried by a thermal power generating set can be reduced; the power generation load rate can be improved; and the power supply coal consumption of the thermal power generating set can be reduced.

The invention discloses a three-pulse intersection approaching guidance method. The method comprises following steps: a relative orbit kinetic equation between a space robot and a GEO satellite is established by means of the C-W equation; a three-pulse C-W guidance status transition equation with revised pulses is established by means of the relative orbit kinetic equation; an optimized objective function for minimum fuel consumption and guidance precision is established according to the status transition equation, the pulse size, direction and time are determined as optimizing parameters; optimized results are obtained according to the optimized objective function. According to the three-pulse intersection approaching guidance method, a revised pulse is added between an inceptive pulse and a final pulse approaching the intersection for increasing guidance precision.

The invention relates to a 5G virtual access network mapping method in the time delay sensing condition and belongs to the technical field of mobile communication. According to the method, the difference between the arrivals of virtual network requests, the dynamic occupation and release of limited physical resources, and the difference between the service capabilities of network function virtualization servers are combined, and then a queue state transfer equation is established at a network functional virtualization choreographer and each network function virtualization server. Meanwhile, atwo-stage queue dynamic scheduling method is provided. Furthermore, the queue states of all queues in the system are sensed, so that the queue dynamic scheduling is carried out to meet the time delayrequirement. Moreover, the mapping is completed by taking a minimum mapping cost as a target. According to the 5G virtual access network mapping method, the mapping cost can be minimized, and the timedelay requirement of the system can be met.

The invention relates to a robot joint torque signal acquisition system and method, and belongs to the technical field of robots. The measurement interference noise is divided into harmonic mechanicaltransmission noise, power frequency electromagnetic interference noise and sensor measurement errors. An adaptive oscillator is designed to model the power frequency interference noise and harmonic transmission noise mathematically, a state transition equation and an observation equation are established, and the inherent noise of the system is estimated and separated. Meanwhile, aiming at the defects of a torque prediction method based on a dynamic model, a torque state prediction algorithm based on weighted quadratic regression is adopted, and the accuracy and the rapidity of state prediction are improved. A Kalman filter is used for filtering uncertain noises existing in the system transfer equation and the measurement equation, the measurement torque data and the prediction data are fused, the amplitude, the phase and the frequency of the inherent interference noise of the system and the real torque output value are subjected to unbiased optimal estimation, and the denoising filtering of the torque signal is realized.

The invention discloses an ECEF-GLS system error self-adaptive registration method based on real-time quality control, and belongs to the technical field of system error registration. A system error is one of main factors which restrict target tracking precision and improvement of different radar track association and fusion currently. To self-adaptively estimate and eliminate the system error which exists in the system at the time of target tracking, the system error to be estimated is regarded as a state vector, a state transition equation among system error estimation in a previous time and a later time is obtained, in this way, a system error self-adaptive estimation measurement equation is obtained from measurement data by utilizing the ECEF-GLS algorithm, and a system error self-adaptive estimation filter equation is established, quality control is conducted on the estimated system error through a filter technology, convergence speed of a system error self-adaptive estimation is expedited by utilizing an iterative loop, multi-target tracking precision and track association accuracy are improved through self-adaptive removal of the system error to improve system performance.

The invention discloses a fault rate evaluation method of an MMC (Modular Multilevel Converter). The fault rate evaluation method comprises the following steps: (1) acquiring a sub-module and a sub-module assembly of an MMC circuit; (2) inputting working parameters of the MMC circuit and the sub-module assembly thereof, and finishing setting of initial conditions; (3) calculating a fault rate of the sub-module assembly according to a fault model to obtain the fault rate of the sub-module; (4) establishing a Markov model of the MMC circuit and establishing a state transfer equation based on the Markov model; (5) solving the state transfer equation to obtain a variation function of the fault rate of the MMC circuit along with time and module quantity, a variation function of the reliability along with the time and the module quantity and average failure time. The calculation reliability of the fault rate of the sub-module assembly is greatly improved and the estimation reliability of the fault rate of the MMC circuit is improved; by applying the Markov model, the solved fault rate and average service life of the MMC circuit are relatively accurate and are varied along with time variation; the fault rate evaluation method has a dynamic property and high fault evaluation precision.

The invention provides a method for predicting remaining life of mechanical equipment based on multi-condition dynamic benchmarking, comprising the following steps of: Firstly, establishing a mechanical equipment degradation state space model comprising the state transition equation and the observation equation; Secondly, estimating the unknown parameters and signal transformation parameters of the model, that is, estimating the parameters of the state transition equation in the maximum likelihood estimation method based on the training sample failure time data and transforming the monitoringsignals under different operating conditions into the transformation parameters of the reference condition monitoring signal by linear interpolation estimation, and estimating the parameters of the observation equation by the transformed signal, then performing dynamically benchmarking on the monitoring signal of the test sample under different working conditions and estimating the state value ofthe test sample in the particle filter algorithm, finally, calculating the analytical solution of the probability density function of the remaining life of the test sample. The method for predicting remaining life of mechanical equipment based on multi-condition dynamic benchmarking can perform monitoring signal benchmarking under multiple conditions dynamically and dynamically in the process of remaining life prediction, which is beneficial for improving the accuracy of the prediction of the remaining life of mechanical equipment.

The invention provides a pose estimation method of a mobile robot and a computer readable storage medium. The method comprises the following steps: acquiring multi-source sensor data, the multi-source sensor data comprising: an image acquisition device acquires a ground scene image, an inertial measurement unit acquires ground data, and a wheel encoder acquires ground data; initializing pose estimation of the mobile robot according to scene image detection scene dynamics; collecting ground data by an inertial measurement unit, collecting ground data by a wheel type encoder, and calculating a prior estimator of a state vector of the mobile robot at the current moment and a corresponding covariance transfer matrix through a state transition equation; and iterating the prior estimator of the state vector at the current moment and the corresponding covariance matrix according to the scene image until convergence to complete filtering update. The data acquired by the image acquisition equipment, the data acquired by the inertial measurement unit and the data acquired by the wheel type encoder are used as the input data of the pose estimation and are combined, so that the advantages are complementary, and the tightly coupled pose estimation data can be obtained.

The invention discloses a time varying network link packet loss probability estimation method based on a Kalman filter. The time varying network link packet loss probability estimation method based on the Kalman filter mainly comprises a training phase and an estimation phase. In the training phase, a source node sends back-to-back detection packets to multiple destination nodes so as to obtain path data, then, prior information of time varying link packet loss probability is estimated according to the path data, and a state transition equation of the Kalman filter is established. In the estimation phase, under the condition that the detection packets do not need to be sent, recursive computation and estimation of the time varying link packet loss probability are completed through a feedback control method according to the state transition equation and the path data obtained from network background flow. Due to the fact that a Kalman filter module is introduced to estimate the link packet loss probability of a time varying network, an obtained link packet loss probability estimated result has a minimum mean square error as well as high estimated accuracy, and the time-variation characteristic of the time varying network link packet loss probability can be reflected in real time.

The invention discloses a time and fuel pulse optimal traversal method for observing a local scope of a spatial target during the on-orbit service of a spacecraft, and relates to spacecraft track control. The invention aims at the multi-azimuth on-orbit service observation of one local scope of the spatial target, and provides the method. The method comprises the steps:1, obtaining a state transition equation of the relative position and speed of a tracking spacecraft; 2, enabling an M-degree scope to be divided into 1*1 fine grids; 3, carrying out the traversal of all fine grids in a spiral mode, and determining the traversal sequence; 4, converting the solving of an optimal traversal scheme into a non-linear planning problem; 5, enabling all speed pulses to be converted into the corresponding speed pulses applied to the tracking spacecraft according to the real-time condition of an executing mechanism. The method is used in the field of time and fuel pulse optimal traversal.

The invention provides a gray target tracking method based on a self-adaptive window and mainly aims at solving the problem of real-time size change of a self-adaptive target. The method comprises the steps of (1) determining a target template, (2) extracting characteristics of a mixed histogram, (3) initializing a particle point set, namely setting the sampled particle number N and initializing particle weights, (4) updating the state of the particle point set, (5) observing the particle point set, (6) estimating the state of the target, (7) tracking a window scaling, and (8) resampling. Under a particle filter tracking framework, the movement target template is determined, the state transition equation of the movement target is established, the particle set target window is initialized, the particle set of next moment is predicted, the weight of a likelihood function is calculated, a tie distance between high-weight particles and the center of the particle set is calculated, and the window size is adjusted linearly according to d. The self-adaptive adjustment of the tracking window is realized; compared with similar methods, the method has the advantages of simple calculation, high speed and good effect.

The invention relates to a method applicable to the optimal power flow solution of an active power distribution network, and the method comprises the following steps that (S1) an optimal power flow model based on the active power distribution network is presented; (S2-1) the active distribution network optimal power flow scheduling cycle is classified into a plurality of sections which are connected back and forth; (S2-2) the state transition equation of an nth time section is as shown as follows: x (n+1)=Tn (x(n), u(n)) (3) and, the optimal power flow u(n) based on the nth time section is equal to opf(x(n)) (4); (S2-3) the target function of the section is shown in the specifications (5); and the optimal decision-making combinations of all sections form the optimal power flow scheduling strategy of the whole scheduling cycle. According to the method applicable to the optimal power flow solution of the active power distribution network, the operation cost is minimized within a complete scheduling cycle on the premise of meeting various network operation constraints and ensuring the utilization of renewable energy to the largest extent.

The invention relates to a non-linear dynamic system signal processing method based on sampling rejecting particle filter algorithm, which comprises the following steps that: the state space of a particle description state system is adopted, the state space model of a non-linear dynamic system is set to be: xk=f(xk-1)+vk-lzk=h(xk)+nk, wherein xk and zk respectively refer to the state and the observation value of the system at k, f(xk-l) and h(xk) respectively refer to the state transition equation and the observation equation of the system, and vk-l and nk respectively refer to the system noise and the observation noise; and for each newly sampled particle, firstly the probability that the particle is accepted is calculated and then whether the particle is accepted is judged. The non-linear dynamic system signal processing method based on sampling rejecting particle filter algorithm can effectively reduce particle weight degradation, has good particle diversity, and effectively process non-linear and non-gaussian signals.