Industrial production resource allocation system and method

Abstract

The invention relates to the technical field of industrial automation, and discloses an industrial production resource distribution system and method, and the system comprises an operation data collection module, a demand matrix construction module, a resource distribution optimization module, a local distribution optimization module, and a feedback adjustment module. Compared with the prior art which depends on a static resource allocation mode, and particularly under the conditions of real-time fluctuation of resource requirements in a process link and uneven load of equipment tasks in industrial production, the technical problem that accurate, dynamic and efficient resource allocation cannot be realized is solved. The self-adaptive adjustment of resource allocation is realized by adopting a reinforcement learning and feedback mechanism, and the real-time performance and the utilization rate of resource allocation are remarkably improved, so that the problems of non-uniform resource allocation, low efficiency and response lag are avoided, and the overall efficiency and stability of industrial production are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial automation, and particularly relates to an industrial production resource allocation system and method. Background Art

[0002] At present, there are problems of insufficient accuracy and limited dynamic adjustment ability in the resource allocation method in industrial production. For example, in complex industrial scenarios such as iron and steel smelting and petrochemical production, the demands for computing power, network bandwidth, and storage resources in different process links such as ironmaking, steelmaking, and continuous casting are significantly different. Existing technologies usually adopt resource allocation schemes based on preset rules or static models, either unable to accurately allocate resources according to the importance and complexity of each process link, resulting in uneven resource allocation or waste, or difficult to flexibly adjust the allocation strategy according to real-time working conditions and changes in equipment task loads, and also failing to achieve an optimization mechanism that combines the global and local aspects, making it difficult to balance the global resource utilization rate and the task performance of a single device. Therefore, there is an urgent need for a system and method that can dynamically collect and analyze real-time data and still achieve precise matching and dynamic adjustment of resource allocation in the case of complex process resource demands, so as to improve the resource utilization efficiency, response speed, and overall operation stability of industrial production. Summary of the Invention

[0003] Aiming at the above-mentioned technical deficiencies, the purpose of the present invention is to propose an industrial production resource allocation system, aiming to solve the technical problem that the existing technology relies on a static resource allocation mode, especially under the conditions of real-time fluctuations in process link resource demands and uneven equipment task loads in industrial production, and is unable to achieve precise, dynamic, and efficient resource allocation.

[0004] To solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides an industrial production resource allocation system, including:

[0005] An operation data acquisition module, configured to collect real-time operation data of the i-th process link at time t through industrial equipment sensors, including: computing power utilization rate C i,t , network bandwidth occupancy rate B i,t , storage utilization rate S i,t , and task load volatility σ i,t ;

[0006] A demand matrix construction module, configured to calculate the dynamic complexity DCF of the i-th process link according to the real-time operation data of the i-th process link at time t i , and generate a resource demand priority matrix P according to the value of DCF i ;

[0007] A resource allocation optimization module, configured to construct a first resource allocation matrix, and optimize the resource allocation matrix according to the generated resource requirement priority matrix P in combination with a swarm intelligence algorithm to obtain a second resource allocation matrix;

[0008] A local allocation optimization module, configured to, for a process step with a task load volatility σ i,t greater than a preset task load volatility threshold, pre-construct a reinforcement learning model, and based on the second resource allocation matrix in combination with the task load volatility σ i,t use the reinforcement learning model to perform local optimization of the allocation matrix to obtain a third resource allocation matrix B;

[0009] A feedback adjustment module, configured to collect actual usage status feedback data of the third resource allocation matrix B, including the usage conditions of network resources, storage resources, and computing resources of each process step, and dynamically feedback and adjust the third resource allocation matrix B according to the actual usage status feedback data.

[0010] Preferably, in the demand matrix construction module, wherein, w 1 , w 2 , w 3 are respectively the weight coefficients of network resources, storage resources, and computing resources, which are set according to the requirements of different process steps; B max , S max , C max are respectively the maximum values of the preset corresponding resource requirements.

[0011] Preferably, in the resource allocation optimization module, optimizing the resource allocation matrix according to the generated resource requirement priority matrix P in combination with a swarm intelligence algorithm to obtain a second resource allocation matrix specifically includes:

[0012] Calculating the allocation cost F(A) according to the generated resource requirement priority matrix P and the first resource allocation matrix, and the formula is:

[0013]

[0014] wherein, p ij is the priority of the i-th process step in the resource requirement priority matrix P for the j-th type of resource; a ij is the amount of the j-th type of resource allocated to the i-th process step in the first resource allocation matrix; U ij is the actual utilization efficiency of the preset resource j; C ij is the allocation cost of the amount of the j-th type of resource allocated to the i-th process step; m is the total number of process steps; n is the total number of resource types;

[0015] Use the first resource allocation matrix as the initial population in the swarm intelligence algorithm, and use the allocation cost F9A as the objective function in the swarm intelligence algorithm. Repeat the iteration until the value of the objective function is less than the preset allocation cost threshold, and output the second resource allocation matrix.

[0016] Preferably, in the local allocation optimization module, based on the second resource allocation matrix A combined with the task load volatility σ i,t Use the reinforcement learning model to perform local optimization of the allocation matrix to obtain the third resource allocation matrix B, specifically including:

[0017] Construct the input data of the reinforcement learning model: Combine the second resource allocation matrix A with the task load volatility σ i,t As the input data of the reinforcement learning model;

[0018] Construct the state space S of the reinforcement learning model;

[0019] Construct the action space A of the reinforcement learning model;

[0020] Construct the reward function: Construct the reward function R(S,A) according to the state space S and the action space A;

[0021] Execute iterative optimization: Use the Q-learning algorithm to update the state-action value function Q(S,A), select the action space A according to the state space S, execute the adjustment operation to generate a new allocation matrix, calculate the reward value R(S,A) and update the state-action value function Q(S,A). When the state-action value function Q(S,A) converges, output the optimized third resource allocation matrix B.

[0022] Preferably, in the local allocation optimization module, the reward function R(S,A) is defined according to the improvement of resource utilization and the change of allocation cost:

[0023] R(S,A) = w 1 ·ΔU(S,A) - w 2 ·C adjust (S,A)

[0024] Where, ΔU(S,A) is the increased amount of resource utilization after adjustment; C adjust (S,A) is the increased amount of allocation cost caused by the adjustment; w 1 , w 2 Are the weight parameters of resource utilization improvement and allocation cost.

[0025] Preferably, in the local allocation optimization module, S = {B i,t , σ i,t , C i,t}

[0026] Preferably, in the local allocation optimization module, A = {aij +Δa ij ∣Δa ij ∈{-δ, 0, +δ}}, where a ij is the resource quantity of the jth type of resource allocated to the ith process step in the first resource allocation matrix; Δa ij is the adjusted resource allocation amount, and δ is the resource allocation step size.

[0027] Preferably, the present invention provides an industrial production resource allocation method, including:

[0028] Step S10: Collect real-time operation data through industrial equipment sensors. The real-time operation data of the ith process step at time t includes: computing power utilization rate C i,t , which is used to represent the computing power usage of the equipment in the ith process step; network bandwidth occupancy rate B i,t , which is used to represent the data transmission volume usage of the equipment in the ith process step; storage utilization rate S i,t , which is used to represent the storage capacity usage of the equipment in the ith process step; task load volatility σ i,t , which is used to reflect the load change range of the equipment tasks in the ith process step within time t; and form an initial resource demand matrix by normalizing the collected real-time operation data.

[0029] Step S20: Calculate the dynamic complexity DCF in the ith process step according to the real-time operation data of the ith process step in Step S10 i , and generate a resource demand priority matrix P according to the value of DCF i .

[0030] Step S30: Construct a first resource allocation matrix, and optimize the resource allocation matrix according to the resource demand priority matrix P generated in Step S20 by combining with the swarm intelligence algorithm to obtain a second resource allocation matrix A;

[0031] Step S40: For the steelmaking and continuous casting process steps, based on the second resource allocation matrix A and combining with the task load volatility σ i,t , use a reinforcement learning model to perform local optimization of the allocation matrix to obtain a third resource allocation matrix B;

[0032] Step S50: Collect the actual usage status feedback data of the third resource allocation matrix B in Step S40, including the network resources, storage resources, and computing power resources usage of each process step, and dynamically feedback and adjust the third resource allocation matrix B according to the actual usage status feedback data.

[0033] The beneficial effects of the present invention are as follows: Compared with the prior art, there are problems such as insufficient multi-source data integration ability, fault diagnosis relying on manual analysis and slow response speed. Especially under the conditions of complex equipment operation environment and frequent parameter changes, it is impossible to achieve rapid diagnosis and efficient operation and maintenance. Since the present application constructs a closed-loop optimization mechanism based on knowledge graph and transfer learning, it realizes the deep association of multi-source data and the dynamic update of knowledge, thus avoiding the problems of response lag and high misdiagnosis rate in traditional operation and maintenance methods, and improving the accuracy of fault diagnosis, response speed and the intelligent operation and maintenance ability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0035] Figure 1 It is a schematic flowchart of the first embodiment of an industrial production resource allocation system provided by the present invention.

[0036] Figure 2 It is a schematic diagram of the equipment of an industrial production resource allocation system provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0038] Embodiment 1: As Figure 1 shown, it is a schematic flowchart of the first embodiment of the industrial production resource allocation system of the present invention, and the first embodiment of the industrial production resource allocation system of the present invention is proposed.

[0039] In the first embodiment, the industrial production resource allocation system includes:

[0040] An operation data acquisition module, configured to collect real-time operation data of the i-th process link at time t through industrial device sensors, including: computing power utilization rate C i,t , network bandwidth occupancy rate B i,t , storage utilization rate S i,t and task load volatility σ i,t ;

[0041] It should be noted that the real-time data acquisition module is implemented by deploying various industrial sensor devices such as temperature sensors, flow meters, PLC controllers, industrial cameras, etc. in the technological process i;

[0042] It can be understood that the resource requirements of each technological process are significantly different. For example, the ironmaking process has a priority need for computing power resources to support the automated operation of the PLC controller; the steelmaking process has a priority need for network resources to meet the high-bandwidth requirements of high-definition cameras and data transmission; the continuous casting process has a priority need for storage resources to record and store the real-time generated production data. The initial matrix of resource requirements after normalization is an important input data for the subsequent demand matrix construction module, which not only unifies the data ranges of different resource dimensions but also provides a basis for the calculation of dynamic complexity.

[0043] The demand matrix construction module is used to calculate the dynamic complexity DCF in the i-th technological process according to the real-time operation data of the i-th technological process at time t i , and generate the resource demand priority matrix P according to the value of DCF i ;

[0044] It should be noted that in the demand matrix construction module, where w 1 , w 2 , w 3 are the weight coefficients of network resources, storage resources, and computing power resources respectively, which are set according to the requirements of different technological processes; B max , S max , C max are the preset maximum values of the corresponding resource requirements respectively.

[0045] It should be understood that the priority matrix P provides the relative importance ranking of the resource requirements of each technological process, guiding the resource allocation optimization module to allocate limited resources to the most needed processes.

[0046] For example, the real-time data of the steelmaking process is: the network bandwidth occupancy rate B i,t = 70%; the task load volatility σ i,t = 30%; the computing power utilization rate C i,t = 50%; the weight parameters are w 1 = 0.5, w 2 = 0.3, w 3 = 0.2; calculate DCF i = 0.5·0.7 + 0.3·0.3 + 0.2·0.5 = 0.56, normalize the above calculation result to ensure that the total value is 1, and finally according to DCF iThe value generation priority matrix P = [0.8 0.6 0.4], where the first column is the computing power priority, the second column is the network priority, and the third column is the storage priority.

[0047] A resource allocation optimization module, used to construct a first resource allocation matrix, and optimize the resource allocation matrix according to the generated resource demand priority matrix P combined with the swarm intelligence algorithm to obtain a second resource allocation matrix;

[0048] It should be noted that based on the second resource allocation matrix A combined with the task load volatility σ i,t Use a reinforcement learning model to perform local optimization of the allocation matrix to obtain a third resource allocation matrix B, specifically including:

[0049] Construct the input data of the reinforcement learning model: Combine the second resource allocation matrix A with the task load volatility σ i,t As the input data of the reinforcement learning model;

[0050] Construct the state space S of the reinforcement learning model;

[0051] Construct the action space A of the reinforcement learning model;

[0052] Construct a reward function: Construct a reward function R(S,A) according to the state space S and the action space A;

[0053] Execute iterative optimization: Use the Q-learning algorithm to update the state-action value function Q(S,A), select the action space A according to the state space S, execute adjustment operations to generate a new allocation matrix, calculate the reward value R(S,A) and update the state-action value function Q(S,A). When the state-action value function Q(S,A) converges, output the optimized third resource allocation matrix B.

[0054] It should be understood that the swarm intelligence algorithm is used to optimize the first resource allocation matrix, and its purpose is to maximize the resource utilization efficiency and task completion degree of the process link while ensuring the total resource constraint.

[0055] A local allocation optimization module, used for the process link where the task load volatility σ i,t is greater than the preset task load volatility threshold, pre-construct a reinforcement learning model, and based on the second resource allocation matrix combined with the task load volatility σ i,t Use the reinforcement learning model to perform local optimization of the allocation matrix to obtain a third resource allocation matrix B;

[0056] It should be noted that the task load volatility refers to the fluctuation range of resource requirements in the process link within time t, and is used to measure the stability of task execution.

[0057] It can be understood that the task load volatility of the process link fluctuates in real time due to changes in task types or execution states. For example, in the steelmaking process, the real-time data transmission of high-definition cameras may cause a sharp increase in bandwidth demand due to unexpected situations in the production process; in the continuous casting process, the data storage demand may fluctuate significantly due to abnormal data collection.

[0058] It should be understood that the third resource allocation matrix B generated after being adjusted by reinforcement learning can better meet the requirements of high-volatility process links while maintaining the balance of global resource allocation.

[0059] The feedback adjustment module is used to collect the actual usage status feedback data of the third resource allocation matrix B, including the usage of network resources, storage resources, and computing power resources in each process link, and dynamically feedback and adjust the third resource allocation matrix B according to the actual usage status feedback data.

[0060] It should be understood that the feedback adjustment module provides a closed-loop control mechanism, which ensures the continuous optimization of the resource allocation scheme through real-time data collection and dynamic adjustment.

[0061] In addition, the present invention also provides an industrial production resource allocation device. Please refer to Figure 2, An industrial production resource allocation device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute an industrial production resource allocation system in the first embodiment above. An industrial production resource allocation device in an embodiment of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. An industrial production resource allocation device is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of the present invention. An industrial production resource allocation device may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of an industrial production resource allocation device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 may allow an industrial production resource allocation device to communicate with other devices wirelessly or wiredly to exchange data. Although an industrial production resource allocation device with various systems is shown in the figure, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems may be alternatively implemented or had.

[0062] The present invention also provides a computer program product, including a computer program which, when executed by a processor, implements an industrial production resource allocation system as described above. The computer program product provided by the present invention can solve the technical problem of industrial production resource allocation. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as those of the industrial production resource allocation system provided by the above embodiments, and will not be elaborated herein.

[0063] Specifically, according to the embodiments disclosed by the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed by the present invention include a computer program product which includes a computer program carried on a computer-readable medium. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by a processing device 1001, the above functions defined in the system of the embodiments disclosed by the present invention are executed.

[0064] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0065] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these changes and modifications.

Claims

1. An industrial production resource allocation system, characterized in that: The system includes: The operation data collection module is used to collect the real-time operation data of the i-th process link at time t through industrial equipment sensors, including: computing power utilization rate C i,t , Network bandwidth utilization rate B i,t , Storage utilization S i,t and task load fluctuation rate σ i,t ; Demand matrix building module, used to calculate the dynamic complexity DCF of the i-th process link based on the real-time operation data of the i-th process link at time t i , according to DCF i The value of generates the resource demand priority matrix P; A resource allocation optimization module is used to construct a first resource allocation matrix, and optimize the resource allocation matrix according to the generated resource demand priority matrix P in combination with a swarm intelligence algorithm to obtain a second resource allocation matrix; The local allocation optimization module is used to optimize the task load fluctuation rate σ i,t For process links with a task load fluctuation rate greater than a preset threshold, a reinforcement learning model is pre-built based on the second resource allocation matrix combined with the task load fluctuation rate σ i,t The third resource allocation matrix B is obtained by using the reinforcement learning model to locally optimize the allocation matrix; The feedback adjustment module is used to collect actual usage status feedback data of the third resource allocation matrix B, and dynamically feedback and adjust the third resource allocation matrix B according to the actual usage status feedback data.

2. An industrial production resource allocation system as claimed in claim 1, characterized in that: In the demand matrix building module, Among them, w1, w2, and w3 are the weight coefficients of network resources, storage resources, and computing resources, respectively, which are set according to the requirements of different process links; B max ,S max ,C max They are respectively the preset maximum values ​​of the corresponding resource requirements.

3. An industrial production resource allocation system as claimed in claim 1, characterized in that: In the resource allocation optimization module, the resource allocation matrix is ​​optimized according to the generated resource demand priority matrix P in combination with the swarm intelligence algorithm to obtain the second resource allocation matrix, which specifically includes: The allocation cost F(A) is calculated based on the generated resource demand priority matrix P and the first resource allocation matrix. The formula is: Among them, p ij is the priority of the i-th process link to the j-th resource in the resource demand priority matrix P; a ij is the resource amount of the j-th resource allocated to the i-th process link in the first resource allocation matrix; U ij is the actual utilization efficiency of the preset resource j; C ij is the allocation cost of the amount of resources allocated to the jth type of resources in the preset i-th process link; m is the total number of process links; n is the total number of resource types; The first resource allocation matrix is ​​used as the initial population in the swarm intelligence algorithm, and the allocation cost F(A) is used as the objective function in the swarm intelligence algorithm. The iteration is repeated until the value of the objective function is less than the preset allocation cost threshold, and the second resource allocation matrix is ​​output.

4. An industrial production resource allocation system as claimed in claim 3, characterized in that: In the local allocation optimization module, based on the second resource allocation matrix A combined with the task load fluctuation rate σ i,t The third resource allocation matrix B is obtained by locally optimizing the allocation matrix using the reinforcement learning model, which specifically includes: Construct the input data of the reinforcement learning model: combine the second resource allocation matrix A with the task load volatility σ i,t As input data for reinforcement learning models; Construct the state space S of the reinforcement learning model; Construct the action space A of the reinforcement learning model; Construct reward function: Construct reward function R(S,A) based on state space S and action space A; Perform iterative optimization: Use the Q-learning algorithm to update the state-action value function Q(S,A), select the action space A according to the state space S, perform adjustment operations to generate a new allocation matrix, calculate the reward value R(S,A) and update the state-action value function Q(S,A). When the state-action value function Q(S,A) converges, output the optimized third resource allocation matrix B.

5. An industrial production resource allocation system as claimed in claim 4, characterized in that: In the local allocation optimization module, the reward function R(S,A) is defined based on the improvement of resource utilization and the change of allocation cost: R(S,A)=w1·ΔU(S,A)-w2·C adjust (S,A) Among them, ΔU(S,A) is the improvement of resource utilization after adjustment; C adjust (S, A) is the increase in allocation cost caused by the adjustment; w1, w2 are the resource utilization improvement and allocation cost weight parameters.

6. An industrial production resource allocation system as claimed in claim 4, characterized in that: In the local allocation optimization module, S = {B i,t ,σ i,t ,C i,t }.

7. An industrial production resource allocation system as claimed in claim 4, characterized in that: In the local allocation optimization module, A={a ij +Δa ij ∣Δa ij ∈{-δ,0,+δ}}, where a ij is the amount of resources of the jth type allocated to the ith process link in the first resource allocation matrix; Δa ij To adjust the resource allocation amount, δ is the resource allocation step size.

8. An industrial production resource allocation method, applied to an industrial production resource allocation system according to any one of claims 1 to 7, characterized in that: Methods include: Step S10: Collect real-time operation data through industrial equipment sensors. The real-time operation data of the i-th process link at time t includes: computing power utilization rate C i,t , used to indicate the computing power usage of the equipment in the i-th process link; network bandwidth occupancy rate B i,t , used to indicate the data transmission usage of the equipment in the i-th process link; storage utilization rate S i,t , used to represent the storage capacity usage of the equipment in the i-th process link; task load fluctuation rate σ i,t , used to reflect the load change amplitude of the equipment task in the i-th process link within time t; Step S20: Calculate the dynamic complexity DCF of the i-th process link according to the real-time operation data of the i-th process link at time t in step S10 i , according to DCF i The value of generates the resource demand priority matrix P; Step S30: construct a first resource allocation matrix, and optimize the resource allocation matrix according to the resource demand priority matrix P generated in step S20 in combination with a swarm intelligence algorithm to obtain a second resource allocation matrix A; Step S40: For the steelmaking and continuous casting process, based on the second resource allocation matrix A and the task load fluctuation rate σ i,t The third resource allocation matrix B is obtained by using the reinforcement learning model to locally optimize the allocation matrix; Step S50: Collect actual usage status feedback data of the third resource allocation matrix B in step S40, including the usage of network resources, storage resources and computing resources in each process link, and dynamically adjust the third resource allocation matrix B according to the actual usage status feedback data.

Images