47 results about "Particle swarm intelligence" patented technology

The invention discloses an intelligent power distribution network fault recovery method based on a multi-target discrete particle swarm. The method comprises the following steps: 1), initializing parameters; 2), initializing a position and a speed of a discrete particle swarm optimization algorithm, and according to intelligent power distribution network island dividing and load recovery algorithms, calculating each fitness value fitness k relative to a multi-target function for each particle; 3), based on a fitness control concept, performing classification of a control population and a non-control population; 4), according to a corresponding updating rule, updating a particle position and a particle speed in the control population; 5), performing dynamic exchange between the control population and the non-control population; 6), detecting whether a maximum iteration frequency is reached, and if the maximum iteration frequency is reached, skipping to step (4), and otherwise, entering step 7); and 7), outputting a final optimization result.

The invention discloses a charging scheduling method for an electric automobile battery swapping station. The charging scheduling method comprises the following steps that: S1, a scheduling department sends a next day load forecast to a scheduling system at the end of each day, and a management system calculates the upper limits and lower limits of total capacity and charging power of a battery within 24 time intervals in next day and submits the limits to the scheduling system; S2, a charging scheduling system of the electric automobile battery swapping station transmits received data to a charging scheduling model building module 2; S3, a charging power instruction calculation module calculates charging power Pij of the jth time interval of the ith battery swapping station in next day by a particle swarm intelligent optimization algorithm, and a charging power instruction output module transmits the charging power Pij of the Jth time interval of the ith battery swapping station in next day to the ith battery swapping station management system; and S4, an electric automobile power transformer management system executes charging operation according to a transmitted charging power instruction according to the battery condition in the station. By the charging scheduling method, the charging demand of the battery swapping station can be met, and the peak-load shifting effect on the power grid is realized; and the requirements of power grid running and electric automobile power transformation are coordinated.

The invention discloses a collaborative scheduling method of a power distribution network for the output uncertainty of a distributed power supply. The method comprises the following steps of (1) carrying out tracking analysis on a power flow direction of the distributed power supply in a control region of the power distribution network through a power flow tracking method and building interactive connection among the distributed power supply, the power distribution network and an electrical load in the power distribution network; (2) building a collaborative optimization scheduling model of a source network load for the output uncertainty of the distributed power supply under the given confidence level through building chance-constrained programming, and carrying out optimization employing the economic efficiency as the goal; and (3) solving an optimization model by combining a monte-carlo simulation method and a particle swarm intelligence algorithm and completing collaborative optimization. According to the collaborative scheduling method, uncertain factors, caused by the uncertainty of output fluctuation of the distributed power supply and power generation and load prediction errors, for optimization operation of the power distribution network are considered, and optimization operation and risk reduction are achieved through introducing a chance-constrained programming method of the uncertain factors.

The invention discloses a back analysis method of slope displacement based on safety monitoring. In the method, displacement monitoring data is fully utilized to perform optimized inversion and determine the rock mechanical parameter, the minimum error function between the displacement value of numerical calculation and the monitored displacement value is used as a target function, the modified particle swarm intelligent algorithm is utilized to optimize the target function, the numerical calculation is combined with the modified particle swarm intelligent optimization inversion method, and repeated iteration is carried out to fast converge the target function at a globally optimal solution, therefore, the optimal value of the specific parameter is determined, and a corresponding relationship between the displacement monitoring value and the rock mechanical parameter is established. In a practical project, the displacement value predicted by using the mechanical parameter of inversion perfectly matches with the actually measured displacement value and has little error, which indicates that the inversion method of the invention has relatively good practical application value in the back analysis of rock displacement.

The invention discloses an intelligent planning method for a three-dimensional global flight path of an unmanned aerial vehicle in an enemy threat uncertainty environment. The intelligent planning method comprises the steps: firstly building a three-dimensional environment model of an unmanned aerial vehicle according to a priori map; determining three objective functions for evaluating the advantages and disadvantages of the flight path, and establishing a three-objective optimization model of three-dimensional global flight path planning of the unmanned aerial vehicle in an enemy threat uncertain environment; secondly, performing particle swarm intelligent planning on the global path of the unmanned aerial vehicle by adopting an improved multi-target backbone particle swarm optimizationalgorithm; and finally, smoothing all paths in the obtained optimal path set by adopting a linear interpolation method, and displaying a plurality of solved feasible paths on a simulation map, so thata decision maker can select a final path according to actual conditions. The path selected by the intelligent planning method not only can avoid obstacles, but also can avoid threats of enemies, andthe length of the path is short, and a decision maker can select and obtain the optimal path according to actual requirements.

The invention discloses a dynamic parameter setting method for a superheater mechanism model by combining with field data. The method comprises the following steps that: according to a basic physicallaw, from the internal working process of a system, establishing a system mechanism simulation model; taking a superheater outlet parameter as a feature parameter, and carrying out modeling on the superheater in five sections in order to improve model accuracy; in addition, adding the dynamic parameter [Alpha] of the model; and according to practical operation features, carrying out optimal regulation on the simulation model. By use of the method, the superheater mechanism model is combined with a transfer function distinguishing model based on field data, a particle swarm intelligent algorithm is used to automatically set dynamic parameters in the mechanism model according to an error function on the basis of the time constant of the transfer function model distinguished by the field data, simplification and deficiency for the dynamic features of the practical process when the mechanism model is established can be effectively improved, and dynamic feature simulation accuracy and the development efficiency of a simulation system are improved.

The invention discloses an optimal dispatching method for a multi-energy complementary power generation system. According to the method, when a wind and light power generation system and a water suction energy storage power station are switched in a large power grid for operation, a coordinated dispatching method is applied, and the coordination problem between the power grid loss generated when the wind and light power generation system and the water suction energy storage power station are switched in the large power grid and the economic optimal object of a complementary system is solved. Moreover, the absorptive problem of wind and light power generation is further considered, a coordinated dispatching strategy is provided, and the method is used for carrying out system optimization onthe overall system containing the wind and light power generation system and the water suction energy storage power station. Therefore, a flexible, safe and reliable technology can be provided amongthe interior of the wind and light power generation system, the wind and light power generation system and the water suction energy storage power station and between the complementary system and the large power grid by using a particle swarm intelligence algorithm, flexible economic dispatching is realized, and the stability of the complementary system is guaranteed.

The invention provides a content center network caching method based on a software defined network. Under a fused architecture of the software defined network and a content center network, concentrated control and overall cache optimization are performed on caching nodes and content via perception of a controller on overall network topology and cached information. The controller periodically makes statistics on the cached information, and then performs cache optimization according to a data layer cache decision-making request. According to the method provided by the invention, an importance degree and an edge degree of the nodes and the popularity of content are brought into a cache decision-making strategy, mathematical modeling is performed according to the above information, and a particle swarm intelligent algorithm is used for optimizing the mathematical model. The method provided by the invention fully utilizes advantages of the controller on overall control and logic centralization, and thus the cache can perform decision optimization under multi-node coordination.

The invention discloses an intelligent optimization method for realizing multi-antenna position optimization configuration of a mobile terminal. System channel capacity ( ) is used as a system performance optimizing index, the antenna position of the mobile terminal is optimized through a particle swarm intelligent optimization algorithm, an optimal solution is obtained, and the obtained optimal solution is the optimum distribution position of mobile terminal antennas. According to the intelligent optimization method, the intelligent algorithm is applied to the DMIMO transmit-receive system mobile terminal, the antenna position of the mobile terminal is optimized, the performance of a current communication system can be well improved, the capacity performance of the current system can be well improved through the obtained mobile terminal optimal antenna distribution, requirements of communication services are met, and a reference is provided for the design of antenna distribution of the mobile terminal in the future.

The invention discloses a control method of a power station house cold source system based on machine learning and particle swarm optimization. The control method comprises the following steps that S1, modeling is carried out on the power station house cold source system according to an air conditioning refrigeration process mechanism, a refrigerator model, the actual refrigerating capacity of a refrigerator, the chilled water outlet temperature of the refrigerator, the wet bulb temperature and the cooling water supply and return temperature difference are input, and the cooling water inlet temperature of the single refrigerator is output; S2, modeling prediction is carried out on energy consumption of the power station house cold source system based on historical data; and S3, for predicted load demand data, the control parameters of the air conditioner cold source system are optimized in combination with the particle swarm optimization. According to the control method of the power station house cold source system based on machine learning and the particle swarm optimization, a mechanism model and a data driving model of the cold source system are combined, and the PSO intelligentcontrol algorithm is applied to optimize the air conditioner cold source system, so that the total energy consumption of the cold source system is reduced, and therefore the COP index is improved.

The invention relates to a steam turbine exhaust enthalpy predicating method based on a PSO-SVR soft measurement model. The method comprises the following steps of acquiring a sample data set; introducing a particle swarm intelligent algorithm, constructing a fusion type regression model based on a support vector machine for predicating the exhaust enthalpy, namely a PSO-SVR exhaust enthalpy softmeasurement model; training the PSO-SVR exhaust enthalpy soft measurement model based on the sample data set, performing solving for obtaining a best predication model, and establishing a corresponding exhaust enthalpy regression function; and performing steam turbine exhaust enthalpy predication based on the exhaust enthalpy regression function. Compared with the prior art, the steam turbine exhaust enthalpy predicating method has advantages of high predication capability, high predication precision, etc.

The invention belongs to the field of power system operation and control, and especially relates to an active safety verification method for a power transmission network line based on an extreme learning machine. The method provided by the invention gives consideration to a problem that a conventional heuristic-type active safety correction method based on sensitivity cannot be good in quickness and precision of strategy formulating adjustment at the same time, and the method provided by the invention balances the quickness and precision. The method comprises the steps: carrying out the analysis and learning of operation data of a power grid in different states based on the extreme learning machine, and obtaining the sensitivity of injection power of each node to a power transmission line; building an active safety correction model for the sensitivity of the power transmission line based on the injection power of the nodes; finally solving the active safety correction model through a particle swarm intelligent algorithm, and obtaining a generator and a load adjustment scheme.

The invention discloses a non-Gaussian fluctuating wind speed prediction method based on a hybrid intelligent algorithm. The method comprises the following steps: generating a large-span spherical roof structure spatial point non-Gaussian fluctuating wind speed sample through JT transformation and AR model simulation, and dividing the non-Gaussian fluctuating wind speed of each spatial point into a training set and a test set, and performing normalization processing before each simulation test; respectively deducing kernel parameter combinations consisting of kernel parameters and penalty factors searched based on cuckoo and particle swarm intelligent algorithms; by using a CS+PSO-LSSVM learning machine, transforming a non-Gaussian fluctuating wind speed training set sample into a kernel function matrix, mapping the kernel function matrix to a high-latitude feature space, then inputting the sample data and mapping it to a high-dimensional feature space through a nonlinear function; testing multiple linear algorithms on the kernel function matrix optimized by a hybrid intelligent algorithm, and acquiring a non-linear model of the non-Gaussian fluctuating wind speed training sample; and predicting the non-Gaussian fluctuating wind speed test set sample by using the non-linear model.

The invention relates to a BD-WPT system power coordination control method based on PI controller optimization, and the method specifically comprises the following steps: 1) controlling a primary sideinverter circuit to generate a constant voltage, and keeping a phase angle and a phase shift angle of the primary side inverter circuit unchanged; 2) taking the output power of the secondary side asan output variable, comparing the output power with a corresponding reference value, providing a comparison error for the optimized PI controller, and obtaining PI controller parameters in combinationwith a classical ZN setting method and a particle swarm intelligence algorithm; 3) generating a phase shift amount which should be generated by the secondary side output voltage of the system throughan amplitude limiting link, a gain link and the like, and adjusting the phase shift amount of the secondary side inverter output voltage of the system to be equal to the value; 4) comparing the output power of the system with the reference value again, and adjusting the actual phase shift amount to be equal to the value; and (5) repeating the steps (3) and (4) on the premise of ensuring that thephase difference between the output voltages of the inverters on the secondary side and the primary side of the system is fixed to be + / -90 degrees, and after a certain response time, ensuring that the system output tends to be stable.

The invention, which relates to the technical field of the monitoring system of the boiler explosion, discloses a monitoring system for a steam explosion in a furnace. The system comprises a real-time data acquisition module in a furnace. According to a boiler explosion early-warning system included by the real-time data acquisition module, a pressure sensor and a gas sensor are arranged in a boiler and collect relevant data in the furnace; and the relevant data are operated by using a relevant model in the software design. The system has the following beneficial effects: at a hardware design stage, the pressure sensor and the gas sensor are used for collecting and controlling data; and at a software design stage, a temperature element is added. With a particle swarm intelligent mining searching algorithm, a possible dangerous situation in a furnace is detected in artificial intelligence; and the explosion early-warning accuracy rate of the system is high on the condition of different boiler laying situations.

The invention discloses an optimum design method of a distributed antenna system suitable for highway and high-speed railway environments. N fan-shaped or linear areas are arranged in round communities or linear communities, and under the condition that particles are randomly distributed in the round communities or the linear communities, optimal positions of ports of base station antennas in different communities are obtained through a particle swarm intelligent optimization algorithm. The particle swarm intelligent optimization algorithm comprises the steps that a particle swarm is initialized, and the initial adaptive degrees of M particles are calculated; the position information and the speed information of the particles are updated, and whether the particles are still in a search area is judged; the individual optimal solution pbesti of the ith particle and the global optimal solution Gbest of the whole group are updated; the two earlier steps are executed until the convergence criterion of the algorithm is met, and the global optimal solution Gbest is output. According to the method, the layout positions of the antenna ports in the communities are subjected to theoretical optimization, the optimal coverage of the port antennas can be achieved, energy conservation and environment friendliness of the distributed antenna system can be conveniently achieved, and cost of network distribution is reduced.

The invention relates to the technical field of electric vehicle charging and renewable energy source integration. The invention discloses an electric vehicle photovoltaic charging station multi-dimensional target collaborative optimization scheduling method fused with an improved particle swarm intelligent optimization mechanism. The method is characterized by comprising the following steps: 1) constructing a multi-objective optimization model: taking electricity purchase cost minimization and storage battery cycle electric quantity minimization as objective functions, and taking a charging time constraint, an energy storage constraint, a power balance constraint and an equipment capacity constraint as constraint conditions; the method comprises the steps of (1) selecting a Pareto optimal solution set, (2) improving the design of a particle swarm algorithm, (3) outputting a result: outputting a Pareto optimal solution set after iteration is ended, and (4) selecting a decision: selecting an optimal scheme from the solution set through a fuzzy membership function or a weight preference method (for example, the optimal scheme focuses on reducing the electricity purchase cost when the optimal scheme is equal to 0.6). The method achieves the minimization of the electricity purchase cost and the minimization of the cycle electric quantity of the storage battery, prolongs the energy storage life and reduces the operation cost.

The invention relates to the technical field of intelligent loading, and discloses a particle swarm optimization-based three-dimensional loading optimization method and device and a medium, and the method comprises the steps: defining the size and volume of a carriage, generating tray data, and extracting the arrangement sequence and grouping information; calculating the plot ratio of the trays after loading, and analyzing the position vector to obtain a tray arrangement sequence and a grouping scheme; obtaining a single group fitness and an overall fitness; and constructing a particle swarm, initializing particle positions and speeds, then defining position vectors, updating an individual optimal solution and a global optimal solution, and outputting an optimal loading scheme. According to the method, multi-dimensional constraints such as spatial layout, weight distribution and loading and unloading sequence of cargo loading can be effectively fused, and global optimization of a loading scheme is realized. Through a particle swarm intelligent search mechanism, the arrangement and grouping strategy of trays is continuously iteratively optimized, the loading efficiency and the space utilization rate are remarkably improved, meanwhile, the stability and safety in the cargo transportation process are guaranteed, and the good practical application value is achieved.

The invention discloses a near-field underwater fixed multi-element linear array three-dimensional correction method, which comprises the following steps of: arranging a single signal correction signal source system at a plurality of positions and a plurality of depths around a seabed fixed multi-element underwater linear array in a time-sharing placement mode, transmitting single-frequency high-power correction signals, and recording azimuth information and time information of the signal source; recording depth information of each position of the array by using a depth sensor; carrying out synchronous time service on signal data sent by a fixed multi-element linear array receiving signal source by utilizing a navigation satellite signal, and carrying out time marking and conditioning on the data; after signals are received, constructing a near-field signal model containing amplitude-phase errors and three-dimensional position errors, then constructing a target function by taking a spatial spectrum estimation method as a theoretical basis, setting related parameters of a particle swarm intelligent algorithm, optimizing a target, and obtaining estimated values of the amplitude-phase errors and the three-dimensional position errors of an array; and substituting the error estimation value into an original array manifold, completing near-field array error correction, and obtaining absolute latitude and longitude coordinates of each array element of the whole array at the same time.

The invention discloses a flight recovery method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining the disturbance information of flights, and screening disturbed flights according to the disturbance information; carrying out heterogeneity classification on the disturbed flights to obtain priorities, obtaining runway information of the disturbed flights, and carrying out availability and suitability analysis on the runway information; a flight recovery optimization model is established by taking the priority, runway information and flight information of disturbed flights as constraint conditions, recovery parameters for recovering flight of the disturbed flights are output by utilizing the flight recovery optimization model, and the recovery parameters are solved by adopting a novel convergence-diverse particle swarm intelligent optimization algorithm CDPSO. The flight heterogeneity and the multi-runway cooperation characteristic are introduced, airplane-unit linkage modeling is combined, the integrated recovery decision model is constructed, the scientificity, operability and execution efficiency of recovery scheduling can be improved, the adopted novel CDPSO algorithm integrates population layering, infeasible solution repairing and feasible solution detection mechanisms, and the solving performance is greatly improved.

The invention relates to the technical field of power grid disaster prediction, and discloses a power grid rainstorm disaster prediction correction method based on a probability matching series-parallel connection coupling multi-model, so as to improve the accuracy of power grid rainstorm disaster prediction. The method comprises the following steps: selecting a forecast area, and obtaining actually measured precipitation and power grid rainstorm disaster information of the area in a historical period so as to establish at least two power grid rainstorm forecast models for forecasting the precipitation of the area; correspondingly correcting the precipitation of each power grid rainstorm forecasting model by using a probability matching correction method; predicting an error sequence et ofeach power grid rainstorm prediction model at the moment t by adopting an error autoregression model, and obtaining a precipitation sequence through series correction of each model; s4, carrying outparallel correction on the plurality of models in the step S4 by utilizing a least square method; and respectively solving parameters of series correction and parallel correction by adopting a particle swarm intelligent optimization algorithm, establishing an integrated coupling power grid rainstorm forecast correction model, and calculating a corrected result.

A method for optimizing the matching of flow component of reverse osmosis seawater desalination high-pressure pump features that the geometrical parameters of impeller, guide vane and anti-guide vaneare optimized. The method includes the following steps: first, the Latin square test design is used to obtain the multi-scheme design of the high-pressure seawater desalination pump, and then the sensitivity analysis of the pump multi-parameters is carried out, and the rule that the optimization objective is affected by the variation of these factors is explained by changing the numerical value ofthe relevant design variables one by one. At the same time, the response surface test design method is used to design several groups of test plans, and the function of the quadratic term between theoptimization objective and the design variables is established, and the quadratic regression analysis is carried out, so as to obtain the design variables which significantly affect the pump performance and remove the design parameters which have little influence on the pump performance. Thirdly, a PSO intelligent optimization algorithm is used to realize the automatic optimization of the performance. After many iterative calculations, the optimal parameter combination of the over-current components can be obtained in the global scope, and the performance matching design of the over-current components can be realized.

The invention belongs to the technical field of virtual oscillator control of power electronic inverters, and particularly discloses an AHO virtual oscillation controller stabilization method and an adaptive inertia method. The stabilization method comprises the following steps: establishing a nonlinear model of a GFM inverter in a single-inverter grid-connected and multi-inverter interconnection mode; acquiring a balance point by using a numerical analysis method, linearizing the balance point, extracting a linearized Jacobian matrix, and further designing a multi-target fitness function taking eigenvalue real part minimization as a core; and in combination with a particle swarm intelligent search algorithm, parameters influencing the stability in the AHO virtual oscillation controller are optimized, and the dynamic stability of the power system under the wide working condition is ensured. The AHO-VOC controller using the particle swarm optimization method can ensure stability under switching of various working conditions; compared with inertia-free and fixed inertia methods, the adaptive inertia method shows better transient characteristics.

The application provides a method for determining development parameters of a natural gas hydrate reservoir and related equipment. In the process of particle swarm intelligent optimization, a robot learning algorithm is used for data mining and training to obtain a training model. The potential particles are determined based on the training model. The global optimal objective function is determined based on the combination of the potential particles and the updated development parameters. In the case where the convergence condition is reached based on the development parameter combination corresponding to the global optimal objective function value, the development parameter combination corresponding to the global optimal objective function value is determined as the target development parameter combination. The calculation process of the intelligent optimization algorithm can be improved by the machine learning algorithm, and the optimization time of the development parameters of the natural gas hydrate reservoir is shortened.

The invention relates to a micro-gird community distributed energy distribution method based on a demand side response. The method includes following steps: (1) establishing an Operator optimization model; (2) providing a distributed optimization algorithm based on the demand side response; and (3) introducing a peak-to-average ratio PAR to adjust a load side to satisfy the requirement of a largepower grid. According to the demand side response technology, a distributed control policy is employed, micro-grids perform parallel operation, the leakage of privacy information is effectively avoided, the problem of single-point fault is avoided, the load requirement of the whole micro-grid community can be controlled in a coordinated manner through the setting of the PAR, and through a particleswarm intelligent optimization algorithm, the optimal operation cost of the micro-grid is guaranteed, and the power consumption cost of users is reduced.

The invention discloses a super-large shield section frozen soil body temperature characteristic optimization method based on Gaussian process machine learning, and the method comprises the following steps: S1, carrying out the fitting of a temperature function of a reinforced soil body through the Gaussian process machine learning, and obtaining a function rule of the temperature; and S2, performing particle swarm intelligent optimization on the obtained implicit function, and obtaining the frozen surface of the frozen body. The invention also discloses a super-large shield section frozen soil body temperature characteristic optimization device based on Gaussian process machine learning, and the device comprises a temperature function fitting module which carries out the fitting of a temperature function of a reinforced soil body through the Gaussian process machine learning, and obtains a function rule of the temperature; and a frozen body function fitting module which is used for carrying out particle swarm intelligent optimization on the obtained implicit function and obtaining a frozen surface of the frozen body. The method can be widely applied to the technical field of tunnels and underground engineering by effectively controlling the time length of positive freezing, intelligently optimizing freezing, reducing resource waste and enhancing environmental protection.

The invention relates to the technical field of voice customer service testing, in particular to an automatic testing method for a voice customer service system, which comprises the following steps of: acquiring multi-source data, and performing standardized preprocessing to form a standardized data set; constructing a multi-dimensional quantitative model, determining a core dimension classification rule and a test index standard, and performing iterative optimization through a multi-dimensional test configuration particle swarm intelligent optimization model to generate an optimal test configuration scheme; through a full-link test, collecting test data and extracting a core index, generating a system performance level and an optimization direction through a comprehensive evaluation model, and forming decision support data; constructing an abnormal root cause positioning model to mine an index association link, and positioning a system abnormal core cause; and integrating whole-process data to construct a traceable evaluation system, and outputting a quantitative evaluation result and optimized data. According to the scheme, the testing accuracy and efficiency of the intelligent voice interaction system are effectively improved.

The invention belongs to the technical field of aviation electromechanics, and discloses an intelligent parameter optimization method for a thermal management type combined power device, which comprises the following steps of: determining input and output parameters of each component of the thermal management type combined power device, performing mathematical modeling on the input and output parameters, establishing a system common working equation according to each input and output parameter, and optimizing the parameters of each component; establishing an equation set and a simulation model according to a system balance relation, and solving to obtain an output parameter result of each component; determining a function target, a performance target, an input and output parameter, an optimization parameter and a system constraint condition of the device according to the system requirement of the current working mode; and optimizing the simulation model by applying a particle swarm intelligent algorithm, and finally obtaining system output parameters with optimal energy efficiency. According to the device, an optimization evaluation system with energy consumption as an evaluation index and system requirements as constraint conditions is established, and optimal input parameters meeting the optimization evaluation system are obtained through parameter optimization, so that the energy consumption is reduced, and the system efficiency is improved.

The invention relates to a modeling method of an internal combustion engine-lithium bromide combined cooling heating and power system. The modeling method comprises: determining model input and outputvariables, determining a system model structure, determining a model parameter identification variable,; and substituting the system operation data into the improved particle swarm intelligence algorithm by adopting a method of combining a mechanism and data-driven modeling to identify unknown parameters of the system model so as to obtain a final model. Compared with the prior art, the technicalproblems that the internal combustion engine-lithium bromide combined cooling heating and power system is difficult to model, optimize and control are effectively solved, and a premise is provided for operation optimization and control strategy research of the combined cooling heating and power system.

The application discloses a kind of based on deep reinforcement learning and important sampling electromechanical system reliability evaluation method, first establish complex electromechanical system reliability block diagram, then analyze the sensor data of each level operating state of system, build electromechanical system sensor data and multi-physics field performance simulation analysis fusion dataset, then establish the reliability evaluation model based on deep reinforcement learning, based on the important sampling strategy of adaptive MCMC, the reliability of electromechanical system is evaluated, finally, the reliability sensitivity of electromechanical system operating parameters is analyzed, the operating parameters are optimized based on particle swarm intelligence algorithm, and the effectiveness of the optimization strategy is verified through environmental information.The method of the application combines the intelligent decision-making capability of deep reinforcement learning and the adaptive sampling strategy of important sampling, reduces the computing resources and time required for evaluation, while improving the accuracy and reliability of the evaluation results, to better support the design, maintenance and optimization decisions of electromechanical systems.