470 results about "System capacity" patented technology

The invention relates to an active and reactive collaborative optimization control method for a photovoltaic energy storage system based on a network-forming inverter, and the method achieves the effective control of active and reactive power through simulating the characteristics of a synchronous generator. And then collaborative optimization of active and reactive power control parameters is carried out. A system frequency response model is utilized to describe dynamic response of system frequency, and a dynamic equation of the system is converted into a discrete form in combination with the dynamic equation of the energy storage unit. Meanwhile, the optimization model is solved by considering system capacity limitation, so that transient frequency characteristics are improved, and frequency deviation after system disturbance is reduced. In addition, the minimum active power loss is taken as an optimization target, the voltage deviation is taken as a penalty term to be substituted into a target function, and a reactive power optimization target function is established. A PSO algorithm is adopted to solve a reactive power optimization function, and reactive power output of photovoltaic and energy storage is adjusted by adjusting VSG control parameters, so that the economical efficiency of a power grid is improved on the premise of ensuring qualified voltage.

The invention discloses a power coupling-considered capacity ratio estimation method for a network-constructing / network-following multi-machine hybrid system, and the method comprises the steps: building a GFL-GFM hybrid system model, combining VSG power coupling small signal modeling and a dynamic relative gain matrix (DRGA), and quantifying the coupling strength of active / reactive power to output voltage; the frequency domain impedance characteristic of the GFM converter is simplified, and an ideal voltage source series equivalent reactance equivalent model is constructed; deriving an analytical relationship between a generalized short-circuit ratio (gSCR) and a capacity ratio based on the model, and generating a ratio formula in combination with engineering boundary sub-scenes to form a ratio range suitable for multiple scenes; for coupling time-varying characteristics, a closed-loop mechanism of'mixed mode-coupling sensing-correction calculation-execution feedback 'is constructed, and the ratio is dynamically adjusted. According to the method, coupling quantization and multi-machine coordination control are deeply fused, the problem that the external characteristic model precision and the multi-working-condition adaptability are difficult to consider under complex environment coupling is effectively solved, and the engineering applicability of the hybrid fan system is enhanced.

The invention relates to the technical field of power system automation, in particular to a climbing-frequency modulation coordinated optimization method and system considering energy storage capacity uncertainty, and the method comprises the steps: building an opportunity constraint model of an energy storage system based on the capacity uncertainty, and enabling the opportunity constraint model to be used for describing a reliable feasible region under the deviation of energy storage electric quantity; establishing an energy storage system linearization model considering a climbing-frequency modulation power unified constraint, and realizing power coupling and sharing constraint description of two types of adjustment services; aiming at the nonlinear characteristic of the gradeability of the thermal power generating unit, establishing a piecewise linearization model of the gradeability by adopting a piecewise linearization method; and establishing a thermal power-energy storage combined climbing-frequency modulation coordinated optimization model, and solving the optimization model by adopting a quadratic programming method to obtain the climbing power and frequency modulation reserve capacity of the thermal power generating unit and energy storage. According to the method, the actual available range of the energy storage system capacity can be dynamically reflected, the electric quantity out-of-limit risk caused by the uncertainty of the energy storage capacity is avoided, and the reliability of the scheduling result is improved.

The invention discloses a data storage method and device and a computer storage medium, and the method comprises the steps: obtaining a system storage capacity, a new request resource capacity and a system used capacity, and obtaining a system capacity utilization rate based on the system storage capacity and the system used capacity; in response to the fact that the system capacity utilization rate reaches a threshold value and the current residual capacity does not meet the new request resource capacity, determining a storage capacity needing to be expanded based on the system storage capacity, the new request resource capacity, the used capacity and a preset target utilization rate, creating an expansion space corresponding to the storage capacity, and updating the system storage capacity; based on the updated system storage capacity, determining data to be stored of each node, and determining a data block to be migrated; migrating the to-be-migrated data block to the target node, and updating the data storage state of the current node; through the method, the change of the storage capacity can be dealt with more intelligently.

The invention discloses a quantum key distribution method, device and system, and belongs to the technical field of networks. The first quantum device generates a modulated optical signal carrying quantum information and classic information. The classic information and the quantum information are simultaneously modulated on at least one polarization of the modulated optical signal. The modulated optical signal includes a plurality of subcarriers at different frequency bands. The classic information and the quantum information are located within different ones of the plurality of subcarriers. The classical information comprises quantum key negotiation information. The first quantum device sends the modulated optical signal to the second quantum device over the optical fiber link. Quantum information and classical information are modulated on the same polarization of an optical signal, and the system capacity is high. The quantum information and the classical information are respectively modulated on subcarriers in different frequency bands in the optical signal, so that respective modulation precision requirements of the classical signal and the quantum signal can be met. And quantum information and classical information are transmitted simultaneously by multiplexing a set of devices, so that the hardware cost is saved.

The invention discloses a multi-mode underwater acoustic communication system and communication method based on a heterogeneous dual-core architecture, and belongs to the technical field of underwater acoustic communication. The multi-mode underwater acoustic communication system comprises a main control module, an acquisition module and a transmitting module; the main control module adopts a dual-core heterogeneous processor OMAPL138 formed by an ARM core processor and a DSP core processor; inter-core communication is carried out between the ARM core processor and the DSP core processor through a shared memory and interruption; the ARM core processor is responsible for protocol analysis, task scheduling and man-machine interaction, and the DSP core processor focuses on signal processing and real-time control and integrates four communication systems, a channel evaluation module and the like. The problems that application of a single communication mode is limited and underwater acoustic communication links in different complex sea areas are unstable are effectively solved, and the deep sea area underwater vehicle information collaboration and return and deep sea area observation and monitoring network system capacity are effectively improved and guaranteed.

The invention provides a communication method and device, which are applied to the fields of environmental Internet of Things and the like, and the method comprises the following steps: sending a first message; the first message is used for triggering a first random access opportunity, and the first message indicates a first identifier; receiving a first random access request message from the first terminal device at the first random access opportunity; the first terminal equipment is associated with the first identifier; sending a second message before responding to the first random access request message; the second message indicates a second identifier. Through the method, the message in one process can be located between two messages in the other process, for example, the second message in the second process is located between the first uplink message and the first response message in the first process, so that the network equipment can communicate with the terminal equipment in the process of communicating with the terminal equipment in one process. And communication with another terminal device can be carried out in another process, so that the resource utilization rate can be improved, and the system capacity can be improved.

The invention relates to the technical field of e-commerce, and discloses an e-commerce customer service processing system with intelligent quality inspection and high-concurrency simulation capabilities. According to the invention, millisecond index calculation and real-time alarm of full customer service dialogues are realized through a bypass monitoring technology, a traditional lagging mode of post sampling inspection is changed, and management response is converted from passive remedy to active intervention; a high-concurrency pressure simulation module reproduces a business scene in an e-commerce peak period by simulating a script of a real user behavior and a distributed cluster, and a real basis is provided for system capacity planning and stability guarantee; the integrated intelligent verbal skill evaluation module deeply analyzes the dialogue quality from multiple dimensions such as semantics, compliance and emotional change based on a natural language processing technology, and generates an evaluation report with guidance, so that the professional level of customer service is improved; and the service quality, the operation robustness and the management intelligence level of the e-commerce customer service system are improved.

A communication method and apparatus, applied to the field of environmental Internet of Things. The method comprises: sending a first message, the first message being used for triggering a first random access occasion, and the first message indicating a first identifier; receiving a first random access request message from a first terminal device at the first random access occasion, the first terminal device being associated with the first identifier; and before responding to the first random access request message, sending a second message, the second message indicating a second identifier. By means of the method, a message in one process can be located between two messages in another process, for example, a second message in a second process is located between a first uplink message and a first response message in a first process, so that while a network device is communicating with one terminal device in one process, the network device can communicate with another terminal device in another process, thereby increasing the resource utilization rate and increasing the system capacity.

The invention provides a district-based distributed power consumption capability assessment method and system and a medium, and belongs to the technical field of power system capability assessment. The method comprises the steps that pre-processed associated data parameters are classified and stored, and a structured feature database is generated; constructing a transformer area load vector model according to the load power flow direction of the power distribution transformer areas of the power grid line, screening and classifying a plurality of power distribution transformer areas with similar characteristics of the line, and constructing a line load quantification model based on the structured characteristic database and the transformer area load vector model; obtaining a power distribution area consumption capability range according to the structured feature database and the area load vector model, and obtaining a line consumption capability range according to the structured feature database and the line load quantification model; and on the basis of the principle of minimizing the two, combining the power distribution area consumption capability range and the line consumption capability range to generate an actual evaluation result of the power grid consumption capability. According to the method, the district-based distributed power consumption capability evaluation precision can be effectively improved.

The invention relates to the field of data management, and discloses a seat reservation processing method, platform and system and a storage medium, which are used for solving the problem that the system capacity cannot meet the requirements of all users when a large number of users enter the platform at the same time, and the method comprises the following steps: a user logs in or registers the platform by using an account password, collects real-time data of a seat reservation platform, and stores the real-time data of the seat reservation platform; real-time capacity monitoring is carried out according to real-time data of a seat reservation platform, early warning processing is carried out according to a detection result, after a user enters a reservation page, seat information is browsed, the user selects a needed seat according to the seat information and submits the seat to the platform for processing, the platform carries out conflict detection and time period locking operation on the seat, and if reservation succeeds, the user does not need to check the seat. And the platform generates the reservation record, stores the information in the database and sends a reservation success notification to the user, so that the user experience is effectively improved, and the seat reservation failure rate is reduced.

The invention provides a novel hybrid carrier multiple access scheme. According to the scheme provided by the invention, the energy propagation concept of the hybrid carrier is combined with a frequency division multiple access model, and a specific operation method and a processing flow of the hybrid carrier processing and the multiple access processing of the power signal in the sending and receiving processes are provided; and the requirements of signal anti-interference capability, transmission efficiency and multi-user and multi-time-space under an ad hoc network of a power wireless private network are met, and the method is called as hybrid carrier non-orthogonal multiple access (HC-OFDMA). Due to the fact that the energy dispersion capacity is enhanced, a plurality of users can share the same spectrum resource at the same time through non-orthogonal distribution of a power domain, the spectrum utilization rate can be remarkably improved, power distribution can be dynamically adjusted, transmission is optimized according to channel conditions of the users, more users are supported to access at the same time, the system capacity is improved, and the system performance is improved. The method is especially suitable for high-density users and large-scale Internet of Things scenes.

The invention discloses a user demand-driven distributed energy supply system capacity configuration method based on a large language model. The method comprises the following steps: collecting user demand information; performing semantic analysis and understanding on the acquired user demand information by utilizing a large language model, and outputting structured and quantitative user demand description; parameterizing the structured and quantized user demand description to obtain a mathematical model for capacity configuration optimization; and solving a mathematical model for capacity configuration optimization by adopting a numerical optimization algorithm, generating a plurality of optimization schemes, and performing performance evaluation and visual display to assist a user in selection. According to the method, the accuracy and the intelligent level of user demand understanding are improved through the large language model, automatic processing and personalized capacity configuration of unstructured demands are achieved, the capacity configuration efficiency and the user satisfaction degree of the distributed energy supply system are remarkably improved, and the system performance is effectively improved.

The invention provides a control method and device based on a network following proportion and a network construction proportion, equipment and a medium, and relates to the technical field of power systems, and the method comprises the steps: obtaining the system capacity and the device capacity of a power system; wherein the system capacity refers to the sum of rated capacities of various hub substations in the power system, and the device capacity refers to the installation capacity of equipment integrating a network following function and a network construction function in a preset range of the power system; determining a short-circuit ratio according to the system capacity and the device capacity; determining a network following proportion and a network construction proportion according to the short-circuit ratio; acquiring three-phase voltage, three-phase current, direct-current side voltage and direct-current side reference voltage of the power system; and performing pulse width control according to the three-phase voltage, the three-phase current, the direct current side voltage, the direct current side reference voltage, the network following proportion and the network construction proportion. According to the technical scheme, the pulse width of the power system is comprehensively controlled and adjusted based on the network following proportion and the network construction proportion by determining the network following proportion and the network construction proportion.

The invention discloses a terahertz wave signal transmission method and system based on a superheterodyne structure in the technical field of millimeter wave and terahertz wireless communication, and the method comprises a processing method of a transmitted signal and a processing method of a received signal. The transmitting signal processing method comprises the following steps: converting a baseband signal into a plurality of sub-intermediate-frequency signals through a plurality of intermediate-frequency sub-modules in an intermediate-frequency module; synthesizing the plurality of sub-intermediate-frequency signals into a broadband intermediate-frequency signal by using a power combiner, and feeding the broadband intermediate-frequency signal into a radio frequency module; the radio frequency module converts the broadband intermediate frequency signal into a terahertz wave radio frequency signal; the received signal processing method comprises the following steps: the radio frequency module converts a received terahertz wave radio frequency signal into a broadband intermediate frequency signal; dividing the broadband intermediate-frequency signal into a plurality of sub-intermediate-frequency signals by using a power divider; and respectively restoring the sub-intermediate-frequency signals into baseband signals through a plurality of intermediate-frequency sub-modules in the intermediate-frequency module. By optimizing the frequency synthesis and distribution of the signals, the transmission rate and the system capacity of the terahertz wave signals are improved.

The invention discloses a wireless communication network optimization method and system based on multi-intelligent-reflecting-surface grouping, and belongs to the field of intelligent-reflecting-surface-assisted wireless communication, and the method comprises the steps: building a multi-IRS-assisted wireless communication network according to the coverage characteristics of the equivalent cascade link information of a plurality of IRSs; according to the statistical property of the optimal beam domain channel state information, intelligent pre-grouping is carried out on reflection units of a plurality of IRSs, and each group of units share the same reflection coefficient; analyzing array performance gains of different pilot frequency overhead under a single-user single-antenna cascade channel; a system capacity of a multi-IRS intelligent packet assisted wireless communication network under a Rice channel is evaluated. According to the method provided by the invention, the number of pilots for estimating the channels related to the IRS can be greatly reduced while the multi-user communication service quality is improved.

Provided are a communication method and a communication apparatus, which can be applied to a non-terrestrial network (NTN) or a terrestrial network (TN). The communication method comprises: a communication device receiving first indication information, wherein the first indication information is used for indicating index value information of an OCC sequence; and on the basis of OCC-related configuration information and the first indication information, determining an OCC sequence corresponding to each resource group in transmission resources. In this way, by means of introducing an OCC multiplexing technique, the system capacity efficiency of a network (an NTN or a TN) is improved, and interference between different users can be reduced. Moreover, for the case where time-frequency resources of paired UEs are not aligned, a solution is further provided, such that OCC sequences of overlapping resources between the paired UEs remain orthogonal.

The invention discloses a capacity configuration optimization method for an energy storage system, relates to the field of energy storage planning, solves the problem of energy waste of the energy storage system, and comprises the following steps: collecting historical environmental data of a region where a power generation facility corresponding to the energy storage system is located and historical power consumption of a power grid corresponding to the energy storage system; analyzing the power generation performance of the power generation facility of the energy storage system and the power consumption performance of a power grid corresponding to the energy storage system to obtain a peak value operation time period, an average value operation time period and a valley value operation time period corresponding to the energy storage system; setting a system capacity configuration scheme of the energy storage system corresponding to the peak value operation time period and the valley value operation time period; analyzing the operation state of the energy storage system corresponding to the average operation time period, and setting a system capacity configuration scheme of the energy storage system in the average operation time period; the actual energy storage mode of the energy storage system is adjusted based on the system capacity configuration schemes corresponding to the peak value operation time period, the average value operation time period and the valley value operation time period, and stable operation of the energy storage system is guaranteed by improving the energy utilization rate.

The invention discloses a deep reinforcement learning method for combining unmanned aerial vehicle deployment and resource allocation, and the method comprises the steps: building a communication network model, and initializing a communication environment, the number of unmanned aerial vehicles and the number of users; determining an optimization target and a constraint condition according to the signal-to-noise ratio of the user and the system capacity; converting the optimization problem into a Markov decision process, and determining an intelligent agent, a state space, an action space and a reward function by using a deep reinforcement learning method; each agent continuously interacts with the environment through a D3QN algorithm to optimize and update the own network, so that the agent obtains an optimal resource allocation strategy; personalized federated deep reinforcement learning is utilized to maximize long-term network throughput, protect user privacy and adapt to a time-varying network state at the same time, and in a federated learning framework of PFDRL, a deep reinforcement learning algorithm is adopted, so that an unmanned aerial vehicle can decide resource allocation and position adjustment in real time based on local observation, and a global optimal solution is realized at the same time.

The invention relates to the field of spaceflight measurement and control, and discloses a large-scale constellation multiple access measurement and control spread spectrum signal de-spreading bit width quantization method. Signal samples are counted, a noise deterioration factor and a bit width quantification model are respectively established, the de-spreading bit width is accurately expanded, and the optimal balance of resources and performance is realized. The method solves the problem that the detection range of strong and weak signals is limited when a single receiver detects multi-satellite same-frequency signals at the same time, remarkably improves the system capacity, and has the characteristics of quantizability, large dynamic range and multi-constellation adaptation.

The invention discloses a capacity-expanding anti-interference measurement and control communication method, which comprises the following steps of: 1, generating M alternative spread spectrum sequences; step 2, the transmitting end performs parallel combined spread spectrum, including parallel transmission, spread spectrum superposition and carrier modulation; wherein during spectrum spreading, in I and Q, any one alternative spectrum spreading sequence in one path is selected as a synchronization sequence, and all the remaining alternative spectrum spreading sequences are subjected to combined spectrum spreading; step 3, blind interference signal separation at a receiving end: at the receiving end, firstly carrying out carrier demodulation on all interference signals and measurement and control signals received by the array antenna, and then carrying out blind interference signal separation on observation signals after carrier demodulation to obtain separated expansion anti-interference measurement and control signals; and finally, recovering the expanded anti-interference measurement and control signal to obtain recovered data. According to the invention, parallel combination spread spectrum and blind interference signal separation processing are organically combined, so that the anti-interference capability of the measurement and control system can be ensured while the system capacity is improved.

In the NB-IoT NTN deployment, the coverage area can be as large as hundreds of thousand square kilometers. In such huge geography area, the number of IoT devices that need to access the satellite will be massive. In another hand, the UL capability of NB-IoT is severely suppressed by large repetition number due to the limited link budget. To enlarge the system capacity, multiplexing of UEs by usage of orthogonal cover codes (OCC) for NPUSCH format 1 and NPRACH is proposed in 3GPP as a Release 19 IoT NTN work item. The Invention incudes various methods of overlaying the OCC code to the NPUSCH and NPRACH; various procedures on support OCC for NPRACH and NPUSCH such as random access and connection establishment; other related 3GPP specification impact.

The invention provides a communication method and a communication device, and relates to the technical field of communication. According to the invention, multiplexing of resources occupied by multiple channels such as PDSCH and PUSCH can be realized, so that the system capacity is improved. The method comprises the following steps: receiving first information from network equipment, wherein the first information is used for indicating a first orthogonal sequence; on the first time-frequency domain resource, using a first orthogonal sequence to receive first data sent by the network device through a first channel, and / or, sending second data to the network device through a second channel; wherein the first channel comprises a physical downlink shared channel (PDSCH) and / or a non-physical downlink shared channel (NPDSCH), and the second channel comprises one or more of the following: a physical uplink shared channel (PUSCH), a non-physical uplink shared channel (NPUSCH), a physical random access channel (PRACH) or a non-physical random access channel (NPRACH).

The invention relates to an asymmetric fault ride-through cooperative control method and system for a virtual synchronous generator, and belongs to the technical field of power electronics and new energy power generation. Aiming at the problems of three-phase current asymmetry, overcurrent, transient instability and the like faced by a virtual synchronous generator when a power grid has an asymmetric fault, the method comprises the following steps: firstly, setting a negative sequence current reference to be zero through positive and negative sequence separation control so as to keep output current symmetry, and freezing a reactive power ring; secondly, comprehensively considering current amplitude limiting, voltage supporting capacity and system capacity, and dynamically calculating optimal solutions of positive sequence excitation reference amplitudes under different fault depths; finally, positive and negative sequence power components and excitation dynamic changes are considered, an active reference instruction real-time dynamic adjustment method is provided, and a power angle is controlled to be stabilized at a target value by adjusting a power reference value in real time. The method can effectively improve the transient stability during the fault ride-through period, maximally supports the power grid voltage while ensuring that the current is not out of limit, and improves the system capacity utilization rate.

The invention belongs to the field of satellite communication application, and relates to a low-orbit constellation-based hopping beam dynamic scheduling method and a low-orbit constellation-based hopping beam dynamic scheduling system. The method comprises the steps of firstly constructing a ground wave position model; then formulating a satellite selection strategy, selecting an access satellite of the current hopping beam period for each ground beam position, and determining a link establishment relationship between the satellite and the beam position; and then establishing a beam scheduling optimization model, establishing a weighted graph for all ground beam positions, coloring nodes in the weighted graph according to a weighted DSATUR algorithm, and scheduling beams according to the determined graph. According to the method, the ground wave position satellite selection access strategy, the satellite load balancing, the wave position service distribution, the satellite-ground link state and the priorities of different services are comprehensively considered, the optimization of the system capacity in the dynamic environment is realized by adjusting the weight, and the overall service satisfaction rate of the ground wave position is effectively improved.

Embodiments of the present disclosure provide an information processing method, a terminal, a network device, a communication system, and a storage medium. The information processing method is executed by a terminal, and comprises: determining first information; determining a DMRS of the terminal on the basis of the first information; and on the basis of the DMRS, sending an OCC-based multi-user multiplexed NPUSCH to a network device, wherein the DMRS of the terminal is orthogonal to DMRSs corresponding to other terminals in a same OCC multiplexing user group, and the terminals comprised in the same user group use the same time / frequency-domain resource. In this way, the terminal can achieve system capacity expansion on the basis of the design of orthogonal DMRSs, so that more users can be supported to perform uplink transmission on the premise of limited time-frequency resources and the limited transmit power of the terminal.

This disclosure provides an apparatus and a method for wireless communication. Processing circuitry of the apparatus determines one or more antenna ports of the apparatus for data transmission. The processing circuitry generates, for each of the one or more antenna ports, a respective demodulation reference signal (DMRS) based on one of a first DMRS pattern, a second DMRS pattern, a third DMRS pattern, and a fourth DMRS pattern. The one or more antenna ports transmit the one or more DMRSs. The first DMRS pattern includes four code division multiplexing (CDM) groups each comprising a length-2 orthogonal cover code (OCC), the second DMRS pattern includes six CDM groups each comprising a length-2 OCC, the third DMRS pattern includes three CDM groups each comprising a length-4 OCC, and the fourth DMRS pattern includes two CDM groups each comprising a length-6 OCC.

Wireless communication assisted by a reconfigurable intelligent surface under near-field propagation faces the problem of array gain loss, and the increase of system capacity is limited. In order to further improve the communication system capacity of a large-scale intelligent reflecting surface (XL-RIS) in a near field and compensate for serious path loss, the invention provides a multi-user wireless communication system assisted by optimally deploying the large-scale intelligent reflecting surface, so that the near field range is further expanded, and the system capacity is improved. And the XL-RIS topological structure and the beam design are jointly optimized by taking the minimization of the beam gain error expected by the user as a target. In order to solve the non-convex problem, a self-adaptive tabu search algorithm, a Major-Minimization (MM) algorithm and the like are used for converting the sub-problems, and then alternate iterative optimization is carried out on the sub-problems. Particularly, on the premise that hardware cost is not increased, extra spatial freedom is obtained through selection of element positions, the user sum rate is effectively improved, and beam pointing is closer to the direction expected by the user. A simulation result verifies the effectiveness of the method, and a new solution thought is provided for the application of the XL-RIS in a near field communication system.

The invention discloses a communication method and device, relates to the technical field of communication, and can save air interface resources. The method comprises: a base station sending first information; a terminal receives first information, wherein the first information indicates a first parameter corresponding to an orthogonal cover code (OCC) category. The base station sends a second parameter, the terminal receives the second parameter, and the first parameter comprises the second parameter. And the terminal sends uplink information through the OCC category corresponding to the second parameter according to the second parameter and the first information. And the base station receives the uplink information through the OCC category corresponding to the second parameter. According to the scheme, the terminal receives the corresponding relation between the OCC category and the first parameter in advance. In this way, only the second parameter corresponding to the OCC category needs to be received and the OCC category does not need to be received again when the OCC is actually scheduled to be subjected to OCC coding. Therefore, the signaling overhead consumed by transmitting the OCC category is saved, and air interface resources are saved. In addition, data is transmitted in an OCC mode, the number of terminals sending uplink data can be increased, and the system capacity is improved.

The embodiment of the invention provides a neural network generation method and device for a multi-platform system, electronic equipment and a storage medium, the platform system comprises first equipment and at least one second equipment, and the method comprises the following steps: obtaining a training main network and a main network training data set, and pruning the training main network to generate a plurality of training sub-networks; based on the main network training data set, calculating to obtain a main network training gradient of the training main network and a sub-network training gradient of each training sub-network, and updating main network parameters of the training main network; taking the updated training main network as the main network of the first equipment; and performing neuron pruning on the main network based on the second system capacity of the second device and the neuron fraction of each main neuron, and generating the sub-network of the second device based on the pruned main network, thereby improving the reliability, the generation efficiency and the flexibility of network generation in the platform system.