Industrial equipment intelligent operation optimization method and system

Abstract

Multiple embodiments of this specification relate to the field of information technology, and specifically to a method and system for optimizing the intelligent operation of industrial equipment. For use in a distributed computing system, the method comprises the following steps: obtaining an estimated load for the computing task; reading the current loads and reference load intervals of all nodes in the current distributed computing system; obtaining a node whose load does not exceed the reference load interval after being assigned based on the current load and task list, and assigning the computing task to the node; recording the associated data of each node's energy consumption power and resource occupancy rate, and establishing an energy consumption estimation model for each node; responding to subsequent computing tasks based on a pre-established resource prediction model; obtaining a node whose energy consumption increases the least after being assigned the subsequent computing task based on the resource estimate, the node's current resource occupancy rate, and the energy consumption estimation model, recording the node as the target node; and assigning the subsequent computing task to the target node.

Description

Technical Field

[0001] Multiple embodiments of this specification relate to the field of information technology, and specifically to a method and system for intelligent operation optimization of industrial equipment. Background Art

[0002] With the rapid development of information technology, distributed computing systems are increasingly being used in industrial equipment. Distributed computing systems connect multiple computing nodes through a network to collaboratively complete complex computing tasks. However, with the increasing volume and complexity of computing tasks, how to efficiently allocate computing resources, optimize energy consumption, and improve overall system performance have become urgent challenges.

[0003] Traditional distributed computing systems typically use static or semi-static methods to allocate computing tasks, determining task allocation based on pre-defined rules or simple load balancing algorithms. While this approach can alleviate the problem of uneven loads among computing nodes to a certain extent, it struggles to cope with dynamically changing task demands and resource usage. Especially in high-load, multi-tasking scenarios, traditional methods struggle to achieve real-time, accurate resource allocation and energy optimization. Therefore, research into new device operation optimization technologies is necessary. Summary of the Invention

[0004] Multiple embodiments of this specification describe a method and system for intelligent operation optimization of industrial equipment.

[0005] In a first aspect, embodiments of this specification provide an industrial equipment intelligent operation optimization method for a distributed computing system, comprising the steps of:

[0006] Obtaining an estimated load of the computing task based on a response of a pre-established load prediction model to the computing task;

[0007] Read the current load and reference load range of all nodes in the current distributed computing system;

[0008] According to the current load and the task list, obtaining a node whose load does not exceed the reference load range after being assigned the computing task, and assigning the computing task to the node;

[0009] Record the associated data of energy consumption power and resource utilization rate of each node, and after reaching a preset time, establish an energy consumption estimation model for each node based on all the associated data;

[0010] Obtaining an estimated amount of resources for the subsequent computing task based on a response of a pre-established resource prediction model to the subsequent computing task;

[0011] According to the estimated amount of resources, the current resource occupancy rate of the node and the energy consumption estimation model, obtaining a node with the least increase in energy consumption after being assigned the subsequent computing task, and recording it as a target node;

[0012] Allocate the subsequent computing task to the target node.

[0013] In a second aspect, embodiments of this specification provide an industrial equipment intelligent operation optimization system for a distributed computing system, including:

[0014] A first estimation module, which obtains an estimated load of the computing task based on a response of a pre-established load prediction model to the computing task;

[0015] A reading module reads the current load and reference load range of all nodes in the current distributed computing system;

[0016] A first allocation module, based on the current load and the task list, obtains a node whose load does not exceed the reference load range after being assigned the computing task, and allocates the computing task to the node;

[0017] A recording module records the associated data of energy consumption power and resource utilization rate of each node, and after reaching a preset time, establishes an energy consumption estimation model for each node based on all the associated data;

[0018] A second estimation module, which obtains an estimated amount of resources for the subsequent computing task based on a response of a pre-established resource prediction model to the subsequent computing task;

[0019] A second allocation module, based on the estimated resource amount, the current resource occupancy rate of the node and the energy consumption estimation model, obtains a node with the least increase in energy consumption after being assigned the subsequent computing task, and records it as a target node;

[0020] An execution module allocates the subsequent computing task to the target node.

[0021] In a third aspect, embodiments of this specification provide an electronic device, including a processor and a memory;

[0022] The processor is connected to the memory;

[0023] The memory is used to store executable program code;

[0024] The processor reads the executable program code stored in the memory to run a program corresponding to the executable program code, so as to execute the method described in any one of the above aspects.

[0025] In a fourth aspect, an embodiment of this specification provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method described in any one of the above aspects is implemented.

[0026] In a fifth aspect, embodiments of this specification provide a computer program product, including a computer program, which implements the method described in any of the above aspects when executed by a processor.

[0027] The beneficial effects of the technical solutions provided by some embodiments of this specification include at least:

[0028] In various embodiments of this specification, the intelligent operation optimization method and system provided establish a load prediction model based on historical computing task execution records to more accurately predict the system resource requirements of future computing tasks. This model, combined with the energy consumption and resource utilization of each node, is used to establish an energy consumption estimation model. This minimizes the system's total energy consumption while ensuring computing requirements, helping to reduce operating costs and promote the development of green computing. The system can dynamically monitor and adjust resource allocation strategies to adapt to changing task requirements and resource usage, while reducing the additional overhead caused by uneven load or resource waste.

[0029] Other features and advantages of the various embodiments of this specification will be further disclosed in the following detailed description and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of this specification, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 A schematic diagram of a distributed computing system is provided for the embodiments of this specification.

[0032] Figure 2 This is a flow chart of the intelligent operation optimization method provided in the embodiments of this specification.

[0033] Figure 3 This is a schematic diagram of resource utilization provided in the embodiments of this specification.

[0034] Figure 4 Schematic diagram of the intelligent operation optimization system provided in the embodiments of this specification.

[0035] Figure 5 This is a schematic diagram of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0036] The following is an explanation and description of the technical solutions of the embodiments of this specification in conjunction with the drawings of the embodiments of this specification. However, the following embodiments are only preferred embodiments of this specification and are not exhaustive. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without making any creative work are all within the scope of protection of this specification.

[0037] Throughout this specification, the claims, and the accompanying drawings, the terms "first," "second," "third," and the like are used to distinguish between different items, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may include other steps or elements inherent to the process, method, product, or apparatus.

[0038] In the following description, terms such as "inside", "outside", "up", "down", "left", "right", etc. that indicate directions or positional relationships are only used to facilitate the description of the embodiments and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they should not be understood as limitations on this specification.

[0039] The data involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data comply with the relevant laws, regulations and standards of relevant countries and regions.

[0040] Before introducing the technical solution recorded in this specification, the application scenarios of the technical solution and related technologies are introduced.

[0041] In modern intelligent manufacturing systems, a large number of sensors are deployed on various devices on the production line to monitor various parameters in the production process, such as temperature, pressure, speed, etc. These sensors generate massive amounts of data at all times, which need to be processed and analyzed by efficient computing resources to achieve real-time monitoring and optimization of the production process. A distributed computing system 10 is usually used to complete such control computing tasks. The distributed computing system 10 includes one or more scheduling nodes 11 and multiple nodes 12 for executing computing tasks. For example, see the attached Figure 1In an automobile manufacturing plant, industrial robots 31 are responsible for key processes such as welding and assembly. To ensure the accuracy and efficiency of each link, the robot's operating status needs to be monitored in real time, and its operating parameters need to be dynamically adjusted based on the monitoring data. Using intelligent operation optimization methods for industrial equipment, future loads can be predicted based on historical task execution records. Computing tasks are rationally allocated to the most suitable nodes 12, while considering energy consumption factors to achieve optimal performance and minimize energy consumption. In addition, it can also be applied to multiple aspects such as energy management and fault prediction. By conducting deep learning on the historical data of production equipment, an accurate prediction model is established to provide early warning of possible failures, thereby reducing downtime and improving production efficiency.

[0042] Distributed computing is a technology that breaks down large computational tasks into smaller parts and distributes these parts across multiple computers, or nodes, for parallel processing. This distributed computing approach not only significantly improves computational speed but also effectively utilizes computing resources in different locations. It is particularly suitable for processing large datasets or conducting complex simulation experiments. A distributed computing system 10 includes the following key components: a client for initiating computational requests; a scheduler or load balancer responsible for receiving and assigning task requests from clients to appropriate compute nodes 12; servers or workstations where compute nodes 12 actually perform computational tasks; and storage nodes 12, which provide data storage services for compute nodes 12 to access required input data or store results. The intelligent operation optimization method for industrial equipment presented in this specification enables a more intelligent task allocation strategy to be implemented within the distributed computing system 10. This method not only ensures load balancing among compute nodes 12, preventing some nodes from being overloaded while others remain idle, but also minimizes overall system energy consumption while meeting computational requirements, thereby improving overall system efficiency and sustainability. This approach is crucial for promoting the development of Industry 4.0, the Internet of Things (IoT), and big data analytics.

[0043] This specification first proposes an industrial equipment intelligent operation optimization method for a distributed computing system 10, see the attached Figure 2 , including the steps of:

[0044] Step S1) Based on the response of a pre-established load prediction model to the computing task, an estimated load of the computing task is obtained.

[0045] As an example, in this embodiment, the estimated load and load represent CPU usage. In other implementations, the estimated load and load may also represent GPU usage, memory usage, etc. Unlike the resource usage ratio 21 described later, the estimated load and load refer to the usage of a single resource, such as the CPU, GPU, memory, or bandwidth, while the resource usage ratio 21 represents the ratio (not the amount) of usage of multiple resources.

[0046] In this embodiment, the load includes CPU load. The method for establishing a load prediction model provided includes:

[0047] Reading the execution record of the historical computing task, wherein the execution record includes the code, data and CPU load of the historical computing task;

[0048] Obtaining the functions included in the code of the historical computing task, and calculating the number of executions of the function and the average size of data operated during a single execution;

[0049] Generate a characterization coefficient of the function according to the number of executions of the function and the size of data operated during an average single execution;

[0050] Sorting the data according to a preset function, obtaining a characterization coefficient vector of each historical computing task, and associating the characterization coefficient vector with the CPU load as load sample data;

[0051] Establish and use the load sample data to train a machine learning model, and obtain a load prediction model based on the trained machine learning model.

[0052] For example, historical data of welding tasks completed by multiple industrial robots 31 over a period of time are collected. These records include the specific code of each welding task, the welding parameters used, the amount of data processed, and the corresponding CPU load.

[0053] For welding tasks, the following functions are mainly involved: load_welding_parameters, which is used to load the parameters required for a specific welding task, execute_welding_path, which is used to execute the welding operation according to the given path and parameters, and monitor_quality, which is used to monitor the welding quality and adjust the parameters to ensure consistency.

[0054] The load_welding_parameters function is called once before each welding operation and needs to process about 50KB of data (including welding path and parameter settings).

[0055] The execute_welding_path function processes different amounts of data depending on the length of the welding path. For example, for a standard body frame welding task, the function processes approximately 1MB of data.

[0056] The monitor_quality function runs periodically, processing a relatively small amount of data each time, approximately 10KB.

[0057] Define characterization coefficients based on the execution frequency of the above functions and the amount of data they process. For example, if the load_welding_parameters function is called 10 times in a welding task, its characterization coefficient is 10 × 50KB = 500KB. Similarly, the characterization coefficient for execute_welding_path might be 1 × 1MB = 1MB, while the characterization coefficient for monitor_quality is 5 × 10KB = 50KB. The corresponding characterization coefficients can be set to 500, 1024, and 50. Read multiple welding task records and the average CPU load of the robot during each task. For example, the average CPU load during a task is 80%. Associate the CPU load with the characterization coefficient vector [500KB, 1MB, 50KB] to form load sample data. Using the load sample data, train a machine learning model to predict the CPU load required for new welding tasks.

[0058] On the other hand, in another implementation, the method of generating a characterization coefficient of a function according to the number of executions of the function and the average size of data operated during a single execution includes:

[0059] Parsing the code and data to obtain the size of the data operated by each function each time it is executed;

[0060] Calculate the number of executions of each function, the average size of the data operated in a single execution, and the variance of the size of the data operated in a single execution;

[0061] When the variance is smaller than a preset threshold, the size of the data operated during an average single execution of the function is set to a constant value.

[0062] In this embodiment, the method for generating function characterization coefficients is more detailed. It not only considers the number of function executions and the average data size of operations per execution, but also incorporates the variance of the data size to determine whether to set the average data size of operations per execution to a constant value. This helps to more accurately quantify the behavioral characteristics of functions and optimize task allocation and resource management accordingly. The target program is analyzed to obtain the data size of each function's operations during each execution. This can be achieved using static code analysis tools or runtime monitoring tools. The number of function executions, the average data size of operations per execution, and the variance of the data size of operations per execution are calculated.

[0063] Count the number of times each function is called throughout the program's lifetime. Calculate the average size of the data being manipulated across all execution instances. Measure the degree of variability in the size of the data being manipulated between executions. A small variance indicates relatively stable data size; a small variance indicates significant variability.

[0064] A variance threshold is set as a criterion. If a function's variance is below this threshold, it means that the function processes roughly the same amount of data across different executions. In this case, the average data size per execution can be considered a constant. When these conditions are met, the average data size per execution of the function is set to a fixed constant value. This helps simplify feature engineering during subsequent model training and improves prediction accuracy.

[0065] Step S2) Read the current loads and reference load ranges of all nodes 12 in the current distributed computing system 10.

[0066] The reference load range of each node 12 is different and is set using techniques known in the art. For example, the reference load range is obtained by multiplying the economic load rate range of the CPU of each node 12 by the CPU's main frequency.

[0067] Step S3) According to the current load and task list, a node 12 whose load does not exceed the reference load range after being assigned the computing task is obtained, and the computing task is assigned to the node 12.

[0068] Taking a distributed computing system 10 including three nodes 12 as an example, each node 12 has a different reference load range, specifically:

[0069] Node 121: The CPU main frequency is 3.0 GHz, and the economic load rate is 40%-60%. Therefore, the reference load range is 1.2 GHz - 1.8 GHz.

[0070] Node 122: The CPU main frequency is 2.5 GHz, and the economic load rate is 50%-70%. Therefore, the reference load range is 1.25 GHz - 1.75 GHz.

[0071] Node 123: The CPU main frequency is 2.0 GHz, and the economic load rate is 30%-50%. The reference load range is 0.6 GHz-1.0 GHz.

[0072] The economic load rate refers to the CPU usage range that can ensure the best performance and energy consumption ratio. The economic load rate is obtained by performing performance and power consumption tests on devices configured on the same hardware platform as node 12, such as under laboratory conditions.

[0073] The current loads of all nodes 12 in the current distributed computing system 10 are read: Node 121: current load is 1.5 GHz, Node 122: current load is 1.0 GHz, Node 123: current load is 0.8 GHz.

[0074] New computing tasks are expected to increase the load by 0.3 GHz.

[0075] For node 121, the current load is 1.5 GHz, which increases to 1.8 GHz after adding the 0.3 GHz contribution from the new task. Since the reference load range for node 121 is 1.2 GHz - 1.8 GHz, this means that after the new task is assigned to node 121, the load is at the upper limit of the reference load range, but still within the allowable range.

[0076] For node 122, the current load is 1.0 GHz, which increases to 1.3 GHz after adding the 0.3 GHz of the new task. The reference load range for node 122 is 1.25 GHz - 1.75 GHz, so after the new task is assigned to node 122, its load also falls within the reference load range.

[0077] For node 123, the current load is 0.8 GHz, which increases to 1.1 GHz after adding the 0.3 GHz of the new task. The reference load range for node 123 is 0.6 GHz - 1.0 GHz, so assigning the new task to node 123 will cause its load to exceed the reference load range.

[0078] Therefore, both node 121 and node 122 can accept new tasks without exceeding their reference load ranges. However, since node 121's load will eventually reach its upper limit, further program configuration can be used to ultimately select node 122 as the execution node 12 for the new task. This method ensures that each computing task is assigned to a node 12 in the appropriate state, thus achieving load balancing and helping to improve the efficiency and stability of the entire system.

[0079] Step S4) records the associated data of the energy consumption power and resource occupancy rate 21 of each node 12. After a preset time period is reached, an energy consumption estimation model of each node 12 is established based on all the associated data.

[0080] Deploy monitoring tools or services on each node 12 of the distributed computing system 10 to collect real-time energy consumption (e.g., via sensors on the motherboard) and resource utilization 21 (including CPU, GPU, memory, network bandwidth, etc.). These monitoring tools can be specialized software or integrated into the operating system or hardware. Set a reasonable monitoring cycle as the preset duration based on actual needs. For example, data can be collected every 30 minutes or every hour. This interval needs to balance data accuracy with system overhead.

[0081] The resource occupancy rate 21 includes CPU occupancy rate, GPU occupancy rate, memory occupancy rate and network bandwidth occupancy rate.

[0082] The method for establishing the energy consumption estimation model of each node 12 based on all the associated data includes:

[0083] The resource occupancy rate 21 is represented by a vector, which is recorded as an occupancy rate vector;

[0084] Associating the occupancy rate vector with energy consumption power as energy consumption sample data;

[0085] Establish and use the energy consumption sample data to train a machine learning model, and obtain an energy consumption estimation model based on the trained machine learning model.

[0086] For example, during two 30-minute time periods, node 12A's data includes the following: Time period 1: CPU utilization 70%, GPU utilization 20%, memory utilization 65%, network bandwidth utilization 10 Mbps, and power consumption 150 W. Time period 2: CPU utilization 80%, GPU utilization 25%, memory utilization 70%, network bandwidth utilization 12 Mbps, and power consumption 160 W. The utilization vectors can be represented as A1 = [70, 20, 65, 10] and A2 = [80, 25, 70, 12]. The energy consumption samples are ([70, 20, 65, 10], 150) and ([80, 25, 70, 12], 160).

[0087] This data can be used to train a machine learning model using techniques already known in the art. A linear regression model is recommended, as it effectively captures the relationship between input features (resource utilization 21) and output (energy consumption). After training, an energy consumption estimation model is obtained for each node 12. For example, the energy consumption estimation model obtained by the linear regression model is: energy consumption = 0.5 × CPU utilization + 0.3 × GPU utilization + 0.2 × memory utilization + 0.05 × network bandwidth utilization. When we need to assign a new computing task to node 12A, we can estimate the required resource utilization 21 based on the expected task characteristics, and then use the above model to predict the energy consumption that will be incurred by the task.

[0088] On the other hand, in another embodiment, the resource occupancy rate 21 includes CPU occupancy rate, GPU occupancy rate, memory occupancy rate and network bandwidth occupancy rate, and the method of establishing the energy consumption estimation model of each node 12 based on all the related data includes:

[0089] The resource occupancy rate 21 is truncated or rounded according to a preset precision parameter, and then the resource occupancy rate 21 is associated with the energy consumption power;

[0090] The resource occupancy rate 21 associated with the energy consumption power is used as sample data to train a machine learning model to obtain an energy consumption estimation model.

[0091] In this embodiment, to simplify calculations and potentially improve model training efficiency, we first truncate or round the resource utilization ratio 21 according to a preset precision parameter. The precision of truncation or rounding of the resource utilization ratio 21 is determined based on actual needs and model complexity. For example, we can decide to round all percentage values ​​to the nearest integer (i.e., to the nearest whole number) or to round them to one decimal place more precisely.

[0092] Step S5) Based on the response of the pre-established resource prediction model to the subsequent computing task, an estimated amount of resources for the subsequent computing task is obtained.

[0093] Suppose a new computing task (such as a welding control task) is expected to cause node 12A's resource utilization to change to: CPU utilization: 85%, GPU utilization: 30%, memory utilization: 75%, and network bandwidth utilization: 15 Mbps. Based on the energy consumption estimation model described above, the power consumption of node 12A during this task is predicted to be 67.25 W. The energy consumption of a single node 12 when executing a specific task can be predicted, and the energy consumption differences between different nodes 12 can be compared, enabling more intelligent task scheduling decisions and minimizing global energy consumption.

[0094] Step S6) According to the estimated resource amount, the current resource occupancy rate 21 of the node 12 and the energy consumption estimation model, the node 12 with the least increase in energy consumption after being assigned the subsequent computing task is obtained and recorded as the target node 12.

[0095] Step S7 ) Allocate the subsequent computing task to the target node 12 .

[0096] The execution record of the historical computing task also includes resource occupancy rate 21, and the resource occupancy rate 21 includes CPU occupancy rate, GPU occupancy rate, memory occupancy rate and network bandwidth occupancy rate. This embodiment provides a method for establishing a resource prediction model, which specifically includes:

[0097] Parsing the code and data to obtain the size of each function and the data operated each time the function is executed;

[0098] Associating the function and operation data size with the CPU occupancy, GPU occupancy, memory occupancy, and network bandwidth occupancy when the function is executed as resource sample data;

[0099] Establish and use the resource sample data to train a machine learning model to obtain a resource prediction model.

[0100] Analyze the source code of the software module to determine the functionality of each function and the size of the data it operates on. For example, there are three main functions:

[0101] load_welding_path(path_id): loads the welding path with the specified ID from the database.

[0102] calculate_optimal_parameters(welding_path) : Calculates the optimal welding parameters based on a given welding path.

[0103] execute_welding_task(parameters): executes the welding task according to the calculated parameters.

[0104] For each function, the amount of data processed during each execution is required. For example, load_welding_path requires reading approximately 2MB of data. The amount of data processed by calculate_optimal_parameters depends on the complexity of the welding path, averaging approximately 5MB. execute_welding_task involves real-time control instructions and typically processes smaller data sets, approximately 0.5MB.

[0105] Monitor the resource usage of these functions during actual operation and correlate it with the data size of the operations. This can be achieved by integrating performance monitoring tools. Organize the data into a resource sample dataset. Each record in the resource sample dataset represents an execution instance of a function, including the data size of its operations and the corresponding resource usage. Use the resource sample dataset to train machine learning models. You can choose machine learning models such as decision trees, random forests, or neural networks. The function of the resource prediction model is to predict the requirements for CPU, GPU, memory, and network bandwidth based on the input function name and the expected data size of the operations. Some functions do not require the use of the network, while some functions require some network resources. By monitoring the resource usage of these functions during actual operation, this distinction can be reflected in the resource prediction model.

[0106] After model training is complete, the resource prediction model can be used to perform resource forecasting. For example, if a new welding task is planned that involves calling the calculate_optimal_parameters function and expects to process 7MB of data, this information can be input into the model to predict how much CPU, GPU, memory, and network bandwidth resources this call will require. This can help allocate resources more efficiently, avoid overloads, and optimize overall system performance.

[0107] On the other hand, this specification provides an industrial equipment intelligent operation optimization system for a distributed computing system 10, see the attached Figure 4 ,include:

[0108] A first estimation module 100 obtains an estimated load of the computing task based on a response of a pre-established load prediction model to the computing task;

[0109] The reading module 200 reads the current load and reference load range of all nodes 12 of the current distributed computing system 10;

[0110] The first allocation module 300 obtains a node 12 whose load does not exceed the reference load range after being assigned the computing task based on the current load and the task list, and allocates the computing task to the node 12;

[0111] The recording module 400 records the associated data of the energy consumption power and resource utilization rate 21 of each node 12, and after reaching a preset time, establishes an energy consumption estimation model for each node 12 based on all the associated data;

[0112] A second estimation module 500 obtains an estimated amount of resources for a subsequent computing task based on a response of a pre-established resource prediction model to the subsequent computing task;

[0113] The second allocation module 600 obtains the node 12 with the least increase in energy consumption after being assigned the subsequent computing task based on the estimated resource amount, the current resource occupancy rate 21 of the node 12, and the energy consumption estimation model, and records it as the target node 12;

[0114] The execution module 700 allocates the subsequent computing task to the target node 12 .

[0115] See also Figure 5 The figure shows a schematic diagram of the structure of an electronic device provided by an embodiment of this specification.

[0116] like Figure 5 As shown, the electronic device 1100 may include: at least one processor 1101, at least one network interface 1104, a user interface 1103, a memory 1105, and at least one communication bus 1102. The communication bus 1102 may be used to implement communication between the aforementioned components. The user interface 1103 may include buttons, and optionally may also include a standard wired interface or a wireless interface. The network interface 1104 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc. The processor 1101 may include one or more processing cores. The processor 1101 utilizes various interfaces and circuits to connect the various components within the entire electronic device 1100. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 1105 and accessing data stored in the memory 1105, it performs various functions of the routing device 1100 and processes data. Optionally, the processor 1101 may be implemented in hardware using at least one of a DSP, an FPGA, and a PLA. The processor 1101 may integrate one or a combination of a CPU, a GPU, and a modem. The CPU primarily processes the operating system, user interface, and applications; the GPU is responsible for rendering and drawing content displayed on the display; and the modem handles wireless communications.

[0117] It is understandable that the above-mentioned modem may not be integrated into the processor 1101, but may be implemented separately through a chip.

[0118] Memory 1105 may include either RAM or ROM. Optionally, memory 1105 may include non-transitory computer-readable media. Memory 1105 may be used to store instructions, programs, codes, code sets, or instruction sets. Memory 1105 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, sound playback function, image playback function, etc.), instructions for implementing the aforementioned method embodiments, etc.; the data storage area may store data related to the aforementioned method embodiments, etc. Memory 1105 may also optionally be at least one storage device located remotely from the aforementioned processor 1101. Memory 1105, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and application programs. Processor 1101 may be configured to invoke the application programs stored in memory 1105 and execute the methods described in the aforementioned embodiments.

[0119] The embodiments of this specification also provide a computer-readable storage medium having instructions stored therein that, when executed on a computer or processor, cause the computer or processor to perform the steps of the aforementioned embodiments. If the components of the aforementioned electronic device are implemented as software functional units and sold or used as independent products, they may be stored in the computer-readable storage medium.

[0120] The embodiments of this specification also provide a computer program product, including a computer program, which implements multiple steps in the above embodiments when executed by a processor.

[0121] In the absence of conflict, the technical features in this embodiment and implementation scheme can be combined arbitrarily.

[0122] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product comprises multiple computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted via the computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium accessible by a computer or a data storage device such as a server or data center that integrates multiple available media. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state drive (SSD)).

[0123] When implemented via hardware or firmware, the aforementioned method flow is programmed into the hardware circuit to obtain the corresponding hardware circuit structure and realize the corresponding function. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit, whose logical function is determined by the user's device programming. Designers can "integrate" a digital system on a PLD through self-programming, eliminating the need for chip manufacturers to design and manufacture dedicated integrated circuit chips. Moreover, today, instead of manually manufacturing integrated circuit chips, this programming is often performed using "logic compiler" software. This is similar to the software compiler used in program development. Before compilation, the original code must also be written in a specific programming language, called a hardware description language (HDL). There are not just one HDL, but many. Those skilled in the art will also understand that simply by programming the method flow in one of the aforementioned hardware description languages ​​and programming it into the integrated circuit, a hardware circuit that implements the logical method flow can be easily obtained.

[0124] The embodiments described above are merely preferred embodiments of this specification and are not intended to limit the scope of this specification. Without departing from the design spirit of this specification, various modifications and improvements made to the technical solutions of this specification by ordinary technicians in this field should fall within the scope of protection determined by the claims of this specification.

Claims

1. An industrial equipment intelligent operation optimization method for a distributed computing system, characterized in that: Including steps: Obtaining an estimated load of the computing task based on a response of a pre-established load prediction model to the computing task; Read the current load and reference load range of all nodes in the current distributed computing system; According to the current load and the task list, obtaining a node whose load does not exceed the reference load range after being assigned the computing task, and assigning the computing task to the node; Record the associated data of energy consumption power and resource utilization rate of each node, and after reaching a preset time, establish an energy consumption estimation model for each node based on all the associated data; Obtaining an estimated amount of resources for the subsequent computing task based on a response of a pre-established resource prediction model to the subsequent computing task; According to the estimated amount of resources, the current resource occupancy rate of the node and the energy consumption estimation model, obtaining a node with the least increase in energy consumption after being assigned the subsequent computing task, and recording it as a target node; Allocating the subsequent computing task to the target node; The load includes CPU load, and the method for establishing a load prediction model includes: Reading the execution record of the historical computing task, wherein the execution record includes the code, data and CPU load of the historical computing task; Obtaining the functions included in the code of the historical computing task, and calculating the number of executions of the function and the average size of data operated during a single execution; Generate a characterization coefficient of the function according to the number of executions of the function and the size of data operated during an average single execution; Sorting the data according to a preset function, obtaining a characterization coefficient vector of each historical computing task, and associating the characterization coefficient vector with the CPU load as load sample data; Establish and use the load sample data to train a machine learning model, and obtain a load prediction model based on the trained machine learning model.

2. The method for intelligent operation optimization of industrial equipment according to claim 1, characterized in that: The method of generating a characterization coefficient of the function according to the number of executions of the function and the average size of data operated during a single execution includes: Parsing the code and data to obtain the size of the data operated by each function each time it is executed; Calculate the number of executions of each function, the average size of the data operated in a single execution, and the variance of the size of the data operated in a single execution; When the variance is smaller than a preset threshold, the size of the data operated during an average single execution of the function is set to a constant value.

3. The method for intelligent operation optimization of industrial equipment according to claim 1 or 2, characterized in that: The resource occupancy rate includes CPU occupancy rate, GPU occupancy rate, memory occupancy rate and network bandwidth occupancy rate. The method for establishing an energy consumption estimation model for each node based on all the associated data includes: The resource occupancy rate is represented by a vector, which is recorded as an occupancy rate vector; Associating the occupancy rate vector with energy consumption power as energy consumption sample data; Establish and use the energy consumption sample data to train a machine learning model, and obtain an energy consumption estimation model based on the trained machine learning model.

4. The method for intelligent operation optimization of industrial equipment according to claim 1 or 2, characterized in that: The resource occupancy rate includes CPU occupancy rate, GPU occupancy rate, memory occupancy rate and network bandwidth occupancy rate. The method of establishing the energy consumption estimation model of each node based on all the related data includes: performing a truncation or rounding operation on the resource occupancy rate according to a preset precision parameter, and then associating the resource occupancy rate with the energy consumption power; The resource occupancy rate associated with the energy consumption power is used as sample data to train a machine learning model to obtain an energy consumption estimation model.

5. The method for intelligent operation optimization of industrial equipment according to claim 1 or 2, characterized in that: The execution record of the historical computing task also includes resource occupancy rate, which includes CPU occupancy rate, GPU occupancy rate, memory occupancy rate and network bandwidth occupancy rate; Methods for establishing resource forecasting models include: Parsing the code and data to obtain the size of each function and the data operated each time the function is executed; Associating the function and operation data size with the CPU occupancy, GPU occupancy, memory occupancy, and network bandwidth occupancy when the function is executed as resource sample data; Establish and use the resource sample data to train a machine learning model to obtain a resource prediction model.

6. Industrial equipment intelligent operation optimization system, used in distributed computing systems, characterized by: include: A first estimation module, which obtains an estimated load of the computing task based on a response of a pre-established load prediction model to the computing task; A reading module reads the current load and reference load range of all nodes in the current distributed computing system; A first allocation module, based on the current load and the task list, obtains a node whose load does not exceed the reference load range after being assigned the computing task, and allocates the computing task to the node; A recording module records the associated data of energy consumption power and resource utilization rate of each node, and after reaching a preset time, establishes an energy consumption estimation model for each node based on all the associated data; A second estimation module, which obtains an estimated amount of resources for the subsequent computing task based on a response of a pre-established resource prediction model to the subsequent computing task; A second allocation module, based on the estimated resource amount, the current resource occupancy rate of the node and the energy consumption estimation model, obtains a node with the least increase in energy consumption after being assigned the subsequent computing task, and records it as a target node; An execution module allocates the subsequent computing task to the target node; The load includes CPU load, and the method for establishing a load prediction model includes: Reading the execution record of the historical computing task, wherein the execution record includes the code, data and CPU load of the historical computing task; Obtaining the functions included in the code of the historical computing task, and calculating the number of executions of the function and the average size of data operated during a single execution; Generate a characterization coefficient of the function according to the number of executions of the function and the size of data operated during an average single execution; Sorting the data according to a preset function, obtaining a characterization coefficient vector of each historical computing task, and associating the characterization coefficient vector with the CPU load as load sample data; Establish and use the load sample data to train a machine learning model, and obtain a load prediction model based on the trained machine learning model.

7. An electronic device, characterized in that: including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images