Efficient and dynamic processing method, system and platform for park neural network data and storage medium

Abstract

The invention discloses an efficient and dynamic processing method, system and platform for park neural network data and a storage medium, and the method comprises the steps: obtaining to-be-processed neural network data corresponding to a park in real time, and generating at least five pieces of corresponding first parameter data based on the neural network data; based on the first parameter data, establishing a programmable program architecture corresponding to the park, and generating corresponding second parameter data according to the program architecture; dynamically processing and adjusting third parameter data corresponding to the park in real time according to the second parameter data; wherein the third parameter data is a neural network data operation strategy corresponding to the park, and the system, the platform and the storage medium corresponding to the method can realize efficient processing of the neural network data. Moreover, the operation strategy of the neural network data can be dynamically adjusted according to the parameter change of the park, and the coordination between the management planning and the actual situation is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of network data processing, and specifically relates to an efficient and dynamic processing method, system, platform and storage medium for park neural network data. Background Art

[0002] Neural networks (NN) include artificial neural networks (ANN) and spiking neural networks (SNN). ANN is widely used in information processing to simulate the behavioral characteristics of animal neural networks. In the planning and management of the park, it is necessary to establish a management system based on neural networks in order to coordinate the management of the entire park. The current park management method is relatively fixed and difficult to adapt to parameter changes, resulting in poor management results.

[0003] In other words, the existing management structure is difficult to cope with parameter changes in the park, resulting in the inconsistency between management planning and actual conditions, that is, the management structure is fixed. In addition, the management method is single: there is a lack of dynamic adjustment strategies, and it is impossible to flexibly adjust the management method to adapt to the changing environment and needs.

[0004] Therefore, in view of the technical defects that the above existing management architecture is unable to cope with parameter changes in the park, resulting in inconsistency between management planning and actual conditions, as well as a lack of dynamic adjustment strategies and inability to flexibly adjust management methods to adapt to changing environments and needs, it is urgent to design and develop an efficient and dynamic processing method, system, platform and storage medium for park neural network data. Summary of the invention

[0005] In order to overcome the shortcomings and difficulties of the above-mentioned prior art, the purpose of the present invention is to provide an efficient and dynamic processing method, system, platform and storage medium for park neural network data, so as to introduce a programmable hardware architecture: through the programmable hardware architecture, efficient processing of neural network data is achieved. And a dynamic adjustment strategy is implemented: according to the changes in park parameters, the operation strategy of neural network data is dynamically adjusted to ensure the coordination of management planning and actual conditions.

[0006] The first object of the present invention is to provide an efficient and dynamic processing method for campus neural network data; the second object of the present invention is to provide an efficient and dynamic processing system for campus neural network data; the third object of the present invention is to provide an efficient and dynamic processing platform for campus neural network data; the fourth object of the present invention is to provide a computer-readable storage medium.

[0007] The first object of the present invention is achieved in that the method comprises the following steps: Acquire in real time the neural network data corresponding to the park and to be processed, and generate corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is the variable parameter data of the neural network data; Based on the first parameter data, a programmable program framework corresponding to the park is established, and corresponding second parameter data is generated according to the program framework; wherein the second parameter data is the preferred optimal parameter data; According to the second parameter data, the third parameter data corresponding to the park is dynamically processed and adjusted in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0008] Furthermore, the real-time acquisition of neural network data corresponding to the park and to be processed, and the generation of corresponding at least five first parameter data based on the neural network data, also includes: Acquire application data corresponding to the park and in different scenarios, and interactively process the neural network data under each platform based on the application data; wherein the application data includes industry application data and basic application data.

[0009] Furthermore, the method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: Based on the program architecture and in combination with the first parameter data, corresponding fourth parameter data is generated; wherein the fourth parameter data is conditional data corresponding to and affecting each variable parameter; According to the fourth parameter data, the first parameter data is adjusted in real time, and corresponding fifth parameter data is generated; wherein the fifth parameter data is the minimum value data of the sum of all parameters.

[0010] Furthermore, the method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: The resource data corresponding to the park is virtualized, and resource scheduling and service issuance processing are performed on the resource data in real time.

[0011] Furthermore, the method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: Build a server-centric neural network data architecture, and use the server as the basis for data storage, network calls, and data security protection.

[0012] Furthermore, the real-time dynamic processing and adjustment of the third parameter data corresponding to the park according to the second parameter data also includes: According to the second parameter data and in combination with the first parameter data, sixth parameter data corresponding to the third parameter data is generated; wherein the sixth parameter data is data for specifying a neural network data operation strategy; Based on the sixth parameter data, first control data corresponding to the park is generated; wherein the first control data is control instruction data for obtaining the effect of the strategy.

[0013] The second object of the present invention is achieved in this way: the system is used to implement the efficient and dynamic processing method for park neural network data, and the system includes: A data acquisition and generation unit, used for acquiring in real time the neural network data corresponding to the park and to be processed, and generating corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is variable parameter data of the neural network data; A data establishment and generation unit, used to establish a programmable program architecture corresponding to the park based on the first parameter data, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; A data processing and adjustment unit is used to dynamically process and adjust the third parameter data corresponding to the park in real time according to the second parameter data; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0014] Furthermore, the data acquisition generation unit further includes: A first data processing module is used to obtain application data corresponding to the park and in different scenarios, and interactively process the neural network data under each platform based on the application data; wherein the application data includes industry application data and basic application data; And / or, the data establishment generation unit further includes: A first data generating module, configured to generate corresponding fourth parameter data based on the program architecture and in combination with the first parameter data; wherein the fourth parameter data is conditional data corresponding to and influencing each variable parameter; A second data generating module, used to adjust the first parameter data in real time according to the fourth parameter data, and generate corresponding fifth parameter data; wherein the fifth parameter data is the minimum value data of the sum of all parameters; The second data processing module is used for virtualizing the resource data corresponding to the park and performing resource scheduling and service issuance processing on the resource data in real time; The third data processing module is used to build a server-centric neural network data architecture and perform data storage, network calls and data security protection operations based on the server; And / or, the data processing adjustment unit further includes: A third data generating module, used to generate sixth parameter data corresponding to the third parameter data according to the second parameter data and in combination with the first parameter data; wherein the sixth parameter data is data for specifying a neural network data operation strategy; The fourth data generation module is used to generate first control data corresponding to the park based on the sixth parameter data; wherein the first control data is control instruction data for obtaining the effect of the strategy.

[0015] The third object of the present invention is achieved as follows: it includes a processor, a memory, and a platform control program for efficient and dynamic processing of park neural network data; wherein the platform control program for efficient and dynamic processing of park neural network data is executed on the processor, the platform control program for efficient and dynamic processing of park neural network data is stored in the memory, and the platform control program for efficient and dynamic processing of park neural network data implements the efficient and dynamic processing method for park neural network data.

[0016] The fourth object of the present invention is achieved in this way: the computer-readable storage medium stores an efficient and dynamic processing platform control program for campus neural network data, and the efficient and dynamic processing platform control program for campus neural network data implements the efficient and dynamic processing method for campus neural network data.

[0017] The present invention uses a method to obtain in real time the neural network data corresponding to the park and to be processed, and generates corresponding first parameter data of at least five in number based on the neural network data; wherein the first parameter data is variable parameter data of the neural network data; based on the first parameter data, a programmable program architecture corresponding to the park is established, and corresponding second parameter data is generated according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; according to the second parameter data, the third parameter data corresponding to the park is dynamically processed and adjusted in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park, and the system, platform and storage medium corresponding to the method can realize efficient processing of the neural network data; and the operation strategy of the neural network data can be dynamically adjusted according to the changes in the park parameters to ensure the coordination of the management plan with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of the system structure of a programmable hardware architecture for an efficient and dynamic processing method of campus neural network data according to the present invention; Figure 2 A schematic flow chart of a dynamic adjustment strategy of an embodiment of a method for efficiently and dynamically processing park neural network data according to the present invention; Figure 3 A schematic diagram of the process steps of an efficient and dynamic processing method for park neural network data according to the present invention; Figure 4 A schematic diagram of an efficient and dynamic processing system architecture for park neural network data according to the present invention; Figure 5 A schematic diagram of an efficient and dynamic processing platform architecture for park neural network data according to the present invention; Figure 6 A schematic diagram of a computer-readable storage medium architecture in an embodiment of the present invention.

[0020] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0021] In order to better understand the purpose, technical solutions and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification.

[0022] The present invention may also be implemented or applied through other different specific examples, and the details in this specification may also be modified and changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention.

[0023] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0024] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. Secondly, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0025] Preferably, the present invention is applied to one or more terminals or servers for efficient and dynamic processing of park neural network data. The terminal is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), digital signal processors (DSP), embedded devices, etc.

[0026] The terminal can be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal can interact with the client through a keyboard, a mouse, a remote control, a touch pad, or a voice control device.

[0027] The present invention is to realize an efficient and dynamic processing method, system, platform and storage medium for park neural network data.

[0028] like Figure 3 Shown is a flow chart of an efficient and dynamic processing method for campus neural network data provided by an embodiment of the present invention.

[0029] In this embodiment, the efficient and dynamic processing method for campus neural network data can be applied to terminals with display functions or fixed terminals. The terminals are not limited to personal computers, smart phones, tablet computers, desktop computers or all-in-one computers equipped with cameras, etc.

[0030] The efficient and dynamic processing method for campus neural network data can also be applied to a hardware environment consisting of a terminal and a server connected to the terminal via a network. The network includes but is not limited to: a wide area network, a metropolitan area network or a local area network. The efficient and dynamic processing method for campus neural network data of the embodiment of the present invention can be executed by a server, can be executed by a terminal, or can be executed by both a server and a terminal.

[0031] For example, for a terminal that needs to perform efficient and dynamic processing of campus neural network data, the efficient and dynamic processing function for campus neural network data provided by the method of the present invention can be directly integrated on the terminal, or a client for implementing the method of the present invention can be installed. For another example, the method provided by the present invention can also be run on a server or other device in the form of a software development kit (SDK), and an interface for the efficient and dynamic processing function for campus neural network data is provided in the form of SDK. The terminal or other device can implement the efficient and dynamic processing function for campus neural network data through the provided interface. The present invention is further described below in conjunction with the accompanying drawings.

[0032] like Figure 1-Figure 3 As shown, the present invention provides an efficient and dynamic processing method for park neural network data, wherein the method comprises the following steps: S1. Acquire the neural network data corresponding to the park and to be processed in real time, and generate corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is the variable parameter data of the neural network data; S2. Based on the first parameter data, establish a programmable program architecture corresponding to the park, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; S3. Based on the second parameter data, dynamically process and adjust the third parameter data corresponding to the park in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0033] The real-time acquisition of neural network data corresponding to the park and to be processed, and the generation of corresponding at least five first parameter data based on the neural network data, further includes: S11. Acquire application data corresponding to the park and in different scenarios, and interactively process the neural network data under each platform based on the application data; wherein the application data includes industry application data and basic application data.

[0034] The method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: S21, based on the program architecture and in combination with the first parameter data, generating corresponding fourth parameter data; wherein the fourth parameter data is conditional data corresponding to and influencing each variable parameter; S22. According to the fourth parameter data, adjust the first parameter data in real time and generate corresponding fifth parameter data; wherein the fifth parameter data is the minimum value data of the sum of all parameters.

[0035] The method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: S23, virtualizing the resource data corresponding to the park, and performing resource scheduling and service issuance processing on the resource data in real time.

[0036] The method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: S24. Build a server-centric neural network data architecture, and use the server as the basis for data storage, network calls, and data security protection.

[0037] The real-time dynamic processing and adjustment of the third parameter data corresponding to the park according to the second parameter data also includes: S31, generating sixth parameter data corresponding to the third parameter data according to the second parameter data and in combination with the first parameter data; wherein the sixth parameter data is data for specifying a neural network data operation strategy; S32. Generate first control data corresponding to the park based on the sixth parameter data; wherein the first control data is control instruction data for obtaining the effect of the strategy.

[0038] Specifically, in an embodiment of the present invention, a programmable hardware architecture for efficiently processing neural network data is provided, including a central architecture, wherein the central architecture includes a data application platform, a cloud platform, a hardware platform, and an IT infrastructure; The data application platform establishes data applications in different scenarios, including industry applications and basic applications, and performs neural network data interaction on various application platforms based on software operation services; The cloud platform forms resource virtualization, and based on the cloud platform, completes the operation objectives of service provisioning and resource scheduling; The hardware platform forms a server-centric neural network data architecture, and then performs data storage, network calls and data security protection operations based on the server; The IT infrastructure forms an operating architecture based on a physical computer room, which includes a container computer room and a modular computer room, and is used to provide a carrier for neural network data interaction and processing.

[0039] The operating basis of the industry applications and basic applications in the data application platform is the cloud computing model, and the service types of the cloud computing model include SaaS, PaaS and LaaS.

[0040] The industry applications in the data application platform include government cloud, medical cloud, education cloud and park cloud, and the government cloud, medical cloud, education cloud and park cloud can all be programmed for data; among them, when a single industry park is operating, the government cloud is used as the control center, the medical cloud and education cloud are used as logistical support, and the park cloud is used as the intermediate execution structure, which are combined to form the coordinated processing of the neural network data of a single industry park.

[0041] In the park cloud, a multi-tenant application management system is established and runs in SaaS mode. The basic applications in the data application platform include cloud host, cloud storage and cloud security, and certain parameters can be replaced in cloud host, cloud storage and cloud security; when a single data project is operated, data processing is performed with the cloud host as the center, and during data processing, cloud security ensures data security, and cloud storage performs data storage.

[0042] In the cloud host, a big data analysis platform and a computing framework are established, and the cloud host operates in a PaaS mode. Furthermore, the cloud platform is distributed in multiple servers of the hardware platform, and both the cloud platform and the hardware platform operate in a LaaS mode.

[0043] That is, a dynamic adjustment strategy for a programmable hardware architecture for efficiently processing neural network data is provided. The specific implementation steps of the dynamic adjustment strategy are as follows: Step 1: Count the neural network data of the park and set variable parameters of the neural network data; collect neural network data by real-time monitoring of various parameters in the park, and set multiple variable parameters according to environmental changes.

[0044] Step 2: Establish a programmable program architecture based on variable parameters; use the collected parameter data to build a programmable program architecture so that each parameter corresponds to an independent program module and can be flexibly adjusted.

[0045] Step 3: Combine the variable parameters and the corresponding program architecture to find the best parameters. If there are more than one variable parameters, the program will find the impact of each variable parameter. Then, when adjusting the parameters, the minimum value of the sum of each parameter is found to find the best parameters. In other words, use the algorithm to optimize each parameter and find the best parameter combination to ensure the optimality of the management strategy. Specifically, first, the initial parameter setting assumes that the system collects multiple parameters of the neural network operation in the park. ; These parameters may include computing load, network bandwidth, latency, etc. Each parameter will affect the overall performance of the system. We want to optimize these parameters so that the system's performance loss function minimize.

[0046] Construction of loss function: Setting a loss function To measure the performance of the system. Common loss functions can be based on factors such as error, delay, or power consumption. For example:

[0047] in, is the loss function, which represents the difference between the system performance and the target value; It is the first parameters; is the target ideal value of the parameter; is the weight associated with the parameter, indicating the importance of the parameter to the system performance.

[0048] Gradient descent method for parameter optimization: To optimize these parameters, gradient descent method is used to minimize the loss function The update rule for gradient descent is:

[0049] in: It is parameters, is the learning rate, which controls the step size of each update; is the loss function Parameters The partial derivative of is used to measure the contribution of this parameter to the overall loss.

[0050] Calculating the gradient requires the loss function Calculate each parameter The gradient is:

[0051] This formula represents the parameters Distance from target value The farther away, the greater the contribution to system losses.

[0052] Parameter update process: In each iteration, the above gradient is used to update the parameters , so that it gradually tends to the optimal value ; The updated parameters are:

[0053] in: Indicates The parameter value in iteration . Indicates the parameter value after the next iteration.

[0054] Termination condition: Set a termination condition, such as stopping the optimization process when the parameter change is less than a certain threshold or when the maximum number of iterations is reached. Common termination conditions are:

[0055] in: is a small threshold used to determine whether the change in the parameter is small enough to stop the iteration.

[0056] Assume that in the park management system, the three key parameters of concern are: System calculation load; Network latency; Energy consumption; The goal is to make these parameters close to the ideal target value =50; =10 and =100; and weight =0.5; =1.0; =2.0; the initial parameter value is =60; =20; =110; The loss function is:

[0057] Gradient calculation:

[0058] Assuming the learning rate: =0.1, then the parameters are updated as follows:

[0059] Through multiple iterations, the parameters gradually approach the ideal values ​​and the performance of the system is optimized.

[0060] : Loss function, which measures the deviation of system performance from the ideal value.

[0061] No. Neural network parameters; No. The target ideal value of each parameter. No. The weight of a parameter indicates its influence on the loss function. The learning rate controls the step size of each parameter update. Loss function vs. parameter The gradient of is used to update the parameters. This algorithm minimizes the loss function through the gradient descent method, thereby optimizing the various parameters in the operation of the neural network and achieving efficient and dynamic adjustment of the park management system.

[0062] Step 4: According to the optimal parameters, adjust the operation strategy of the park's neural network data. When obtaining the optimal parameters, the optimal parameters are the composite results of various variable parameters. In this way, when the optimal parameters are determined, the corresponding variable parameters will also be determined, and then the determined parameters will be added to the corresponding program to generate the corresponding neural network data operation strategy, that is, to adjust the park's neural network data operation strategy, that is, to apply the optimized parameters to the program architecture, and dynamically adjust the park management neural network data processing strategy to achieve coordination between the actual situation and the management plan.

[0063] Furthermore, in the step one, the number of variable parameters set is not less than five; in the step two, each parameter sets an independent program architecture, and each program architecture adopts the same programming system.

[0064] In the scheme of the present invention, the programmable program is an existing program. During the programming process, the variable parameters used to implement park management can be changed to complete the formulation of park management strategies, thereby realizing efficient processing of neural network data in park management, reducing data redundancy, and reducing the complexity of data caused by multiple attempts at data calculation; when the optimal parameters are determined, the variable parameters can be adjusted according to the optimal parameters to achieve dynamic adjustment strategies. In other words, it can effectively respond to parameter changes in the park, ensure the coordination of park management planning and actual conditions, and then ensure management effects, that is, according to the optimal parameters, determine the variable parameters, and then substitute the variable parameters to complete the strategy designation, and implement the designated strategy to obtain the corresponding effect.

[0065] That is to say, if the Figure 1 As shown, a programmable hardware architecture for efficiently processing neural network data includes a central architecture, which includes a data application platform, a cloud platform, a hardware platform, and an IT infrastructure; Establish data applications in different scenarios in the data application platform, including industry applications and basic applications, and perform neural network data interaction on various application platforms based on software operation services; The cloud platform forms resource virtualization, and based on the cloud platform, completes the task of business issuance and resource scheduling. The hardware platform forms a server-centric neural network data architecture, and then uses the server as the basis for data storage, network calls, and data security protection. An operating architecture based on a physical computer room is formed in the IT infrastructure. The physical computer room includes a container computer room and a modular computer room, which are used to provide a carrier for neural network data interaction and processing.

[0066] The central architecture system can effectively ensure the unified management of the park and form a neural network of park data, thereby ensuring the efficient processing of park management data. In the central architecture, the data application platform, cloud platform, hardware platform and IT infrastructure are used to coordinate different operating areas to ensure management effectiveness.

[0067] Specifically, the operating basis of industry applications and basic applications in the data application platform is the cloud computing model, and the service types of the cloud computing model include SaaS, PaaS and LaaS.

[0068] SaaS enables users to use cloud-based applications through an Internet connection. PaaS can provide development frameworks, block and analyze mining data. LaaS can provide services such as off-site servers, storage and network hardware.

[0069] Specifically, the industry applications in the data application platform include government cloud, medical cloud, education cloud and park cloud, and government cloud, medical cloud, education cloud and park cloud can all perform data programming; among them, when a single industry park operates, the government cloud is used as the control center, the medical cloud and education cloud are used as logistical support, and the park cloud is used as the intermediate execution structure, which are combined to form the coordinated processing of neural network data of a single industry park.

[0070] Specifically, a multi-tenant application management system is established in the campus cloud and runs in SaaS mode. Specifically, the basic applications in the data application platform include cloud host, cloud storage and cloud security, and certain parameters can be replaced in cloud host, cloud storage and cloud security; among them, when a single data project is operated, data processing is centered on the cloud host, and during data processing, cloud security ensures data security, and cloud storage performs data storage.

[0071] Cloud hosting integrates computing, storage and network resources into IT infrastructure rental services, providing on-demand and pay-as-you-go server rental services based on cloud computing models. Customers can deploy the required server environment through a self-service platform on the web interface.

[0072] Cloud storage uses cluster applications, grid technology or distributed file systems and other functions. A large number of different types of storage devices in the network are brought together through application software to work together to provide a system with data storage and business access functions to the outside world, ensuring data security and saving storage space.

[0073] Cloud security uses a large number of networked clients to monitor abnormal software behavior on the network, obtain the latest information on Trojans and malicious programs on the Internet, push it to the server for automatic analysis and processing, and then distribute virus and Trojan solutions to each client.

[0074] Specifically, a big data analysis platform and computing framework are established in the cloud host, and the cloud host operates in PaaS mode. Specifically, the cloud platform is distributed in multiple servers of the hardware platform, and both the cloud platform and the hardware platform operate in LaaS mode.

[0075] To sum up, when conducting park management and planning, the data application platform is used as the management center, the IT infrastructure is used as the management tool, and then the cloud platform and hardware platform work together to complete effective dynamic management of data.

[0076] Preferably, in the scheme of the present invention, the central architecture involved is: Data application platform: Establish data applications for different scenarios, including industry applications and basic applications. Utilize software operation services to realize the interaction of neural network data under each application platform. Cloud platform: Realize resource virtualization, complete business issuance and resource scheduling. Through virtualization technology, integrate various hardware resources and improve resource utilization. Hardware platform: Build a neural network data architecture with the server as the center to perform data storage, network calls and data security protection. IT infrastructure: Use physical computer rooms (such as container computer rooms and modular computer rooms) to provide a carrier for the interaction and processing of neural network data.

[0077] Programmable hardware architecture: Use programmable hardware to achieve high efficiency and flexibility in neural network data processing. By dynamically adjusting strategies, it can adapt to the changing needs of park management in real time to ensure efficient data processing capabilities and management effects.

[0078] Cloud platform: adopts SaaS, PaaS and LaaS models to achieve on-demand use and flexible scheduling of resources. SaaS: provides cloud-based applications that can be used through Internet connections. PaaS: provides development frameworks to block and analyze mining data. LaaS: provides off-site server, storage and network hardware services.

[0079] Hardware platform: Through the server cluster, we build an efficient neural network data architecture, and combine it with the cloud platform to achieve unified management and scheduling of resources. The cloud platform and the hardware platform work together to ensure maximum resource utilization and scheduling flexibility through virtualization technology and cloud computing models.

[0080] Cloud security technology: Through a large number of clients in a network, the abnormal behavior of software in the network is monitored, the latest information of Trojans and malicious programs on the Internet is obtained, and the information is pushed to the server for automatic analysis and processing, and then the solutions for viruses and Trojans are distributed to each client. In the process of data processing and storage, cloud security technology can monitor and protect in real time to ensure the integrity and security of data. Cloud storage: Using cluster applications, grid technology or distributed file systems and other functions, different types of storage devices in the network are integrated to ensure the efficiency and security of data storage.

[0081] Differences of existing technologies: Flexible response to parameter changes: In existing technologies, the park management method is relatively fixed and difficult to respond to changes in environmental parameters. However, this solution can adjust the management strategy in real time through dynamic adjustment strategy to ensure the coordination of management planning with actual conditions.

[0082] Efficient data processing: Existing technologies are inefficient in neural network data processing. This solution improves the efficiency and flexibility of data processing through a programmable hardware architecture combined with a central architecture.

[0083] Resource scheduling and utilization: In the existing technology, resource scheduling is relatively rigid and the utilization efficiency is low. This solution realizes flexible scheduling and efficient utilization of resources through the collaborative work of cloud platform and hardware platform.

[0084] Data security protection: Existing technologies have limited ability to protect data security. This solution uses cloud security technology to provide comprehensive data security monitoring and protection measures to ensure the security of data processing and storage.

[0085] Preferably, the solution of the present invention introduces multi-dimensional data fusion technology to achieve more sophisticated scene perception and optimization adjustment. On the basis of the invention, support for multi-dimensional data fusion is added to not only process single-type data (such as video or sensor data) in the neural network, but also process the fusion of multi-dimensional data sources, such as combining video, sensor, meteorological and user behavior data. Through multi-dimensional data fusion, the system can perceive environmental changes more accurately and optimize the results of neural network processing.

[0086] Specific implementation: Multi-source data collection: In addition to the existing parameter sets (such as video, sensor data), the system will dynamically collect multiple types of data according to scene changes (such as weather, traffic, time, etc.). Data fusion model: Use models such as convolutional neural networks (CNN) or long short-term memory networks (LSTM) to fuse and analyze data from different sources. Adaptive optimization mechanism: Based on the fused data, adjust the system's resource allocation strategy. For example, when the system detects that the weather affects traffic flow, it will prioritize allocating more computing resources to the traffic management module.

[0087] Introducing multi-dimensional data fusion into neural network data processing, intelligent optimization in more complex scenarios is achieved. Based on the adaptive optimization algorithm, the system can automatically detect environmental changes and adjust the operation strategy.

[0088] Resource allocation strategy based on reinforcement learning: Use reinforcement learning algorithms to allocate resources and achieve the optimal solution for dynamic strategies. Introducing reinforcement learning (RL) algorithms in programmable hardware architectures enables the system to optimize its resource allocation strategy through trial and error and feedback mechanisms. Reinforcement learning can continuously learn according to changes in the park environment and optimize the system's resource management and task scheduling.

[0089] Specific implementation: Definition of state, action, and reward: State (State, S): The current state of the campus system (such as traffic flow, network load, etc.). Action (A): The actions that the system can take (such as allocating more resources to traffic or reducing computing load). Reward (R): The effect of the system after performing a certain action (such as reduced latency, reduced energy consumption, etc.). Policy update: Using Q-Learning or Deep Q-Networks (DQN), the system learns how to optimally allocate resources through historical experience. After each resource allocation, the system adjusts its future resource scheduling strategy based on the actual effect (such as reduced latency). Dynamic learning and adjustment: When the system detects changes in resource utilization or task priority, it uses a reinforcement learning model to automatically adjust task scheduling to achieve global optimization. The introduction of reinforcement learning mechanism enables the system to autonomously learn and optimize resource allocation strategies through historical feedback, thereby improving the intelligence level of dynamic adjustment. The system can find the optimal task allocation strategy through automatic trial and error, thereby improving the long-term performance of the system.

[0090] Collaborative optimization of edge computing and cloud computing: Introduce edge computing nodes to collaborate with cloud computing to achieve a more efficient distributed computing architecture.

[0091] In order to further improve the efficiency and real-time performance of neural network data processing, the system can combine edge computing and cloud computing to achieve collaborative work between the two. This architecture can sink part of the data processing to the edge device, reduce data transmission delay, and improve real-time processing capabilities. Specific implementation: Edge node setting: Multiple edge computing nodes are distributed and deployed in the park. Each node is responsible for processing local data (such as monitoring data or sensor data in a specific area), and at the same time aggregating important information to the central cloud. Computing task division: Computing tasks are dynamically assigned to edge devices or the cloud according to the real-time and importance of the task. High real-time tasks (such as video processing) are assigned to edge devices, and more complex or low-priority tasks are assigned to the cloud. Edge-cloud collaborative optimization algorithm: Introduce a task offloading strategy, and use an optimization algorithm to determine which tasks are calculated at the edge and which tasks are processed in the cloud to minimize latency and resource consumption.

[0092] The collaborative processing of edge computing and cloud computing reduces the latency of data processing and improves data processing efficiency. By introducing intelligent task offloading strategies, optimal resource allocation and dynamic scheduling of computing tasks are achieved. Blockchain-based data security protection: Blockchain technology is introduced to ensure the security and integrity of neural network data. In order to ensure the security and reliability of neural network data during transmission, storage and processing, blockchain technology can be used for data protection. Blockchain can provide a tamper-proof mechanism for data transmission in a decentralized manner.

[0093] Specific implementation: Data on-chain: Store the hash value of each data packet on the blockchain to ensure the security of data during transmission between different nodes and prevent data from being tampered with or forged. Smart contracts: Use smart contracts to control data access permissions and achieve automation and transparency in the data processing process. Multi-node verification: Verify the integrity and authenticity of data through the blockchain multi-node consensus mechanism to prevent single point failures and data leakage.

[0094] The combination of blockchain technology ensures the security and reliability of neural network data during transmission and storage, and improves the security protection capabilities of the system. The decentralized and smart contract characteristics of blockchain are used to achieve highly secure dynamic data processing.

[0095] The core of the invention is to efficiently process neural network data through a programmable hardware architecture, combined with real-time parameter data input and dynamic adjustment strategies. This architecture can dynamically adjust computing resources according to different scenario requirements and adapt to changes in park management.

[0096] Example Overview: Assume that in a campus management system, the neural network data that needs to be processed involves three core areas: Security monitoring: Real-time target monitoring through video data and neural networks. Energy management: Optimize power load management through sensor data. Traffic management: Use camera and sensor data to optimize traffic flow in real time. The amount and complexity of data processing in these three areas will change according to time periods and needs (for example, traffic management during peak hours takes up more resources). This solution uses programmable hardware architecture and dynamic adjustment strategies to ensure that the neural network processing capabilities of each area are dynamically allocated under different loads.

[0097] Working mechanism of programmable hardware architecture: This hardware architecture can be dynamically adjusted in real time according to the requirements of different scenarios (such as peak traffic flow). The core of the architecture is flexible resource allocation and parameter optimization, which schedules and optimizes data processing in real time through programmable circuits (such as FPGA, ASIC, etc.). The programmable hardware architecture of the system optimizes according to the parameters collected in real time through dynamic adjustment strategies to ensure the flexibility and efficiency of resource allocation. The gradient descent method is used to dynamically adjust the system performance indicators to optimize the computational delay and resource utilization of each task. This allows the system to operate efficiently under high load conditions while maintaining the optimal neural network data processing effect.

[0098] To achieve the above objectives, the present invention also provides an efficient and dynamic processing system for park neural network data, such as Figure 4 As shown, the system specifically includes: A data acquisition and generation unit, used for acquiring in real time the neural network data corresponding to the park and to be processed, and generating corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is variable parameter data of the neural network data; A data establishment and generation unit, used to establish a programmable program architecture corresponding to the park based on the first parameter data, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; A data processing and adjustment unit is used to dynamically process and adjust the third parameter data corresponding to the park in real time according to the second parameter data; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0099] The data acquisition generation unit further includes: A first data processing module is used to obtain application data corresponding to the park and in different scenarios, and interactively process the neural network data under each platform based on the application data; wherein the application data includes industry application data and basic application data; And / or, the data establishment generation unit further includes: A first data generating module, configured to generate corresponding fourth parameter data based on the program architecture and in combination with the first parameter data; wherein the fourth parameter data is conditional data corresponding to and influencing each variable parameter; A second data generating module, used to adjust the first parameter data in real time according to the fourth parameter data, and generate corresponding fifth parameter data; wherein the fifth parameter data is the minimum value data of the sum of all parameters; The second data processing module is used for virtualizing the resource data corresponding to the park and performing resource scheduling and service issuance processing on the resource data in real time; The third data processing module is used to build a server-centric neural network data architecture and perform data storage, network calls and data security protection operations based on the server; And / or, the data processing adjustment unit further includes: A third data generating module, used to generate sixth parameter data corresponding to the third parameter data according to the second parameter data and in combination with the first parameter data; wherein the sixth parameter data is data for specifying a neural network data operation strategy; The fourth data generation module is used to generate first control data corresponding to the park based on the sixth parameter data; wherein the first control data is control instruction data for obtaining the effect of the strategy.

[0100] In the system solution embodiment of the present invention, the specific details of the method steps involved in the efficient and dynamic processing of campus neural network data have been described above, that is, the functional modules in the system are used to implement the steps or sub-steps in the above method embodiment, which will not be repeated here.

[0101] To achieve the above objectives, the present invention also provides an efficient and dynamic processing platform for park neural network data, such as Figure 5 As shown, it includes a processor, a memory, and a control program for an efficient and dynamic processing platform for park neural network data; wherein, the control program for an efficient and dynamic processing platform for park neural network data is executed by the processor, the control program for an efficient and dynamic processing platform for park neural network data is stored in the memory, and the control program for an efficient and dynamic processing platform for park neural network data implements the steps of the method for efficient and dynamic processing of park neural network data. For example: S1. Acquire the neural network data corresponding to the park and to be processed in real time, and generate corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is the variable parameter data of the neural network data; S2. Based on the first parameter data, establish a programmable program architecture corresponding to the park, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; S3. Based on the second parameter data, dynamically process and adjust the third parameter data corresponding to the park in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0102] The specific details of the steps have been explained above and will not be repeated here.

[0103] In the embodiment of the present invention, the built-in processor of the efficient and dynamic processing platform for park neural network data can be composed of integrated circuits, such as a single packaged integrated circuit, or a plurality of integrated circuits with the same or different functions, including one or more central processing units (CPU), microprocessors, digital processing chips, graphics processors, and various control chips. The processor uses various interfaces and lines to connect various components, and executes various functions and processes data for efficient and dynamic processing of park neural network data by running or executing programs or units stored in the memory, and calling data stored in the memory; The memory is used to store program codes and various data. It is installed in an efficient and dynamic processing platform for the campus neural network data, and realizes high-speed and automatic access to programs or data during operation.

[0104] The memory includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electronically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memory, magnetic disk memory, magnetic tape memory, or any other computer-readable medium that can be used to carry or store data.

[0105] To achieve the above object, the present invention also provides a computer readable storage medium, such as Figure 6 As shown, the computer-readable storage medium stores a control program for an efficient and dynamic processing platform for park neural network data, and the control program for an efficient and dynamic processing platform for park neural network data implements the steps of the efficient and dynamic processing method for park neural network data, for example: S1. Acquire the neural network data corresponding to the park and to be processed in real time, and generate corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is the variable parameter data of the neural network data; S2. Based on the first parameter data, establish a programmable program architecture corresponding to the park, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; S3. Based on the second parameter data, dynamically process and adjust the third parameter data corresponding to the park in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0106] The specific details of the steps have been explained above and will not be repeated here.

[0107] In the description of the embodiments of the present invention, it should be noted that any process or method description in the flowchart or otherwise described herein may be understood as representing a module, fragment or portion of a code comprising one or more executable instructions for implementing steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations, in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.

[0108] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processing module, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM).

[0109] In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0110] In an embodiment of the present invention, to achieve the above-mentioned purpose, the present invention further provides a chip system, wherein the chip system includes at least one processor, and when the program instructions are executed in the at least one processor, the chip system executes the steps of the efficient and dynamic processing method for campus neural network data, for example: S1. Acquire the neural network data corresponding to the park and to be processed in real time, and generate corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is the variable parameter data of the neural network data; S2. Based on the first parameter data, establish a programmable program architecture corresponding to the park, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; S3. Based on the second parameter data, dynamically process and adjust the third parameter data corresponding to the park in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

[0111] The specific details of the steps have been explained above and will not be repeated here.

[0112] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0113] The present invention uses a method to obtain in real time the neural network data corresponding to the park and to be processed, and generates corresponding first parameter data of at least five in number based on the neural network data; wherein the first parameter data is variable parameter data of the neural network data; based on the first parameter data, a programmable program architecture corresponding to the park is established, and corresponding second parameter data is generated according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; according to the second parameter data, the third parameter data corresponding to the park is dynamically processed and adjusted in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park, and the system, platform and storage medium corresponding to the method can realize efficient processing of the neural network data; and the operation strategy of the neural network data can be dynamically adjusted according to the changes in the park parameters to ensure the coordination of the management plan with the actual situation.

[0114] That is to say, in terms of coping with parameter changes, the solution of the present invention: dynamically adjusts the strategy to effectively cope with parameter changes in the park, ensuring the coordination of management planning with actual conditions. In terms of improving management effectiveness: through the central architecture and programmable hardware architecture, an efficient neural network data processing and park management system is formed. In terms of flexible resource scheduling: the collaborative work of the cloud platform and the hardware platform enables flexible scheduling and efficient use of resources. In terms of ensuring data security: during data processing and storage, data security is ensured through cloud security technology.

[0115] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An efficient and dynamic processing method for park neural network data, characterized in that: The method comprises the steps of: Acquire in real time the neural network data corresponding to the park and to be processed, and generate corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is the variable parameter data of the neural network data; Based on the first parameter data, a programmable program framework corresponding to the park is established, and corresponding second parameter data is generated according to the program framework; wherein the second parameter data is the preferred optimal parameter data; According to the second parameter data, the third parameter data corresponding to the park is dynamically processed and adjusted in real time; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

2. According to claim 1, a highly efficient and dynamic processing method for park neural network data is characterized in that: The real-time acquisition of neural network data corresponding to the park and to be processed, and the generation of corresponding at least five first parameter data based on the neural network data, further includes: Acquire application data corresponding to the park and in different scenarios, and interactively process the neural network data under each platform based on the application data; wherein the application data includes industry application data and basic application data.

3. According to claim 1, a highly efficient and dynamic processing method for park neural network data is characterized in that: The method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: Based on the program architecture and in combination with the first parameter data, corresponding fourth parameter data is generated; wherein the fourth parameter data is conditional data corresponding to and affecting each variable parameter; According to the fourth parameter data, the first parameter data is adjusted in real time, and corresponding fifth parameter data is generated; wherein the fifth parameter data is the minimum value data of the sum of all parameters.

4. An efficient and dynamic processing method for park neural network data according to claim 1 or 3, characterized in that: The method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: The resource data corresponding to the park is virtualized, and resource scheduling and service issuance processing are performed on the resource data in real time.

5. According to claim 4, an efficient and dynamic processing method for park neural network data is characterized in that: The method of establishing a programmable program framework corresponding to the park based on the first parameter data, and generating corresponding second parameter data according to the program framework, further includes: Build a server-centric neural network data architecture, and use the server as the basis for data storage, network calls, and data security protection.

6. According to claim 1, a highly efficient and dynamic processing method for park neural network data is characterized in that: The real-time dynamic processing and adjustment of the third parameter data corresponding to the park according to the second parameter data also includes: According to the second parameter data and in combination with the first parameter data, sixth parameter data corresponding to the third parameter data is generated; wherein the sixth parameter data is data for specifying a neural network data operation strategy; Based on the sixth parameter data, first control data corresponding to the park is generated; wherein the first control data is control instruction data for obtaining the effect of the strategy.

7. An efficient and dynamic processing system for campus neural network data, characterized in that: The system is used to implement the efficient and dynamic processing method for park neural network data as described in any one of claims 1 to 6, and the system includes: A data acquisition and generation unit, used for acquiring in real time the neural network data corresponding to the park and to be processed, and generating corresponding at least five first parameter data based on the neural network data; wherein the first parameter data is variable parameter data of the neural network data; A data establishment and generation unit, used to establish a programmable program architecture corresponding to the park based on the first parameter data, and generate corresponding second parameter data according to the program architecture; wherein the second parameter data is the preferred optimal parameter data; A data processing and adjustment unit is used to dynamically process and adjust the third parameter data corresponding to the park in real time according to the second parameter data; wherein the third parameter data is the neural network data operation strategy corresponding to the park.

8. The efficient and dynamic processing system for campus neural network data according to claim 7, characterized in that: The data acquisition generation unit further includes: A first data processing module is used to obtain application data corresponding to the park and in different scenarios, and interactively process the neural network data under each platform based on the application data; wherein the application data includes industry application data and basic application data; And / or, the data establishment generation unit further includes: A first data generating module, configured to generate corresponding fourth parameter data based on the program architecture and in combination with the first parameter data; wherein the fourth parameter data is conditional data corresponding to and influencing each variable parameter; A second data generating module, used to adjust the first parameter data in real time according to the fourth parameter data, and generate corresponding fifth parameter data; wherein the fifth parameter data is the minimum value data of the sum of all parameters; The second data processing module is used for virtualizing the resource data corresponding to the park and performing resource scheduling and service issuance processing on the resource data in real time; The third data processing module is used to build a server-centric neural network data architecture and perform data storage, network calls and data security protection operations based on the server; And / or, the data processing adjustment unit further includes: A third data generating module, used to generate sixth parameter data corresponding to the third parameter data according to the second parameter data and in combination with the first parameter data; wherein the sixth parameter data is data for specifying a neural network data operation strategy; The fourth data generation module is used to generate first control data corresponding to the park based on the sixth parameter data; wherein the first control data is control instruction data for obtaining the effect of the strategy.

9. An efficient and dynamic processing platform for campus neural network data, characterized in that: It includes a processor, a memory, and a platform control program for efficient and dynamic processing of park neural network data; wherein, the platform control program for efficient and dynamic processing of park neural network data is executed on the processor, the platform control program for efficient and dynamic processing of park neural network data is stored in the memory, and the platform control program for efficient and dynamic processing of park neural network data implements the efficient and dynamic processing method for park neural network data as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an efficient and dynamic processing platform control program for campus neural network data, and the efficient and dynamic processing platform control program for campus neural network data implements the efficient and dynamic processing method for campus neural network data as described in any one of claims 1 to 6.

Images