Adaptive environmental industrial computer robust intelligent regulation method and system

Abstract

The application relates to the technical field of regulation and control, and discloses a self-adaptive environment industrial computer robust intelligent regulation and control method and system, wherein the self-adaptive environment industrial computer intelligent regulation and control method comprises the following steps: establishing an electromagnetic interference space-time distribution model, the establishment of the electromagnetic interference space-time distribution model comprising collecting multi-point electromagnetic interference data, preprocessing the electromagnetic interference data, constructing an electromagnetic interference space-time distribution interpolation model, and predicting electromagnetic interference time variation; the application cooperates with multiple-level strategies such as real-time updating of the electromagnetic interference space-time distribution model, accurate classification of task sensitivity, dynamic adjustment of risk assessment, and space-time flexible scheduling of task execution, forms a complete closed-loop feedback system, constructs a robust electromagnetic interference adaptation mechanism, can actively perceive environmental changes and make corresponding adjustments, improves the reliability and stability of the industrial control system in a complex electromagnetic environment, and ensures robust execution of key tasks.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of regulation, more particularly, it relates to a self-adaptive environment industrial computer robust intelligent regulation method and system. BACKGROUND

[0002] Industrial computer system is widely used in manufacturing, metallurgy, petrochemical, energy and other industrial fields, and is the core component of modern industrial automation control. However, the industrial field environment usually has complex electromagnetic interference, such as electromagnetic noise generated by large motor start-stop, arc welding, switching power supply, frequency converter and other equipment, which seriously threatens the stability and reliability of the industrial computer system.

[0003] In the prior art, the methods for improving the electromagnetic compatibility of the industrial computer mainly include the following: one is to improve the anti-interference ability of the device itself through hardware reinforcement, such as adopting shielding shell, filter and optical fiber isolation and other measures; two is to optimize the system software design and enhance the data error detection and correction ability; three is to implement task scheduling optimization to avoid executing critical tasks during the peak of interference. However, these methods have obvious deficiencies: the hardware reinforcement scheme is high in cost and poor in flexibility; the software optimization method can reduce the interference effect but is difficult to fundamentally solve the problem; and the traditional task scheduling method lacks accurate perception and quantitative modeling ability of the space-time distribution characteristics of electromagnetic interference, and cannot implement precise regulation and control for complex electromagnetic environment, so the system robustness is insufficient.

[0004] In particular, the existing industrial computer scheduling method has the following significant deficiencies: one is the lack of fine modeling ability of the space-time characteristics of electromagnetic interference, which cannot accurately depict the non-uniform distribution characteristics of the interference; two is the lack of ability to implement differentiated protection for tasks of different importance levels, which cannot allocate resources according to the task sensitivity and importance; three is that in the environment with drastic changes in electromagnetic interference intensity, the real-time performance of critical tasks and the overall robust stability of the system cannot be guaranteed, resulting in high task failure rate and unstable system response time.

[0005] Therefore, there is an urgent need for an industrial computer system that can accurately perceive the space-time distribution characteristics of electromagnetic interference and implement intelligent regulation and control according to the task sensitivity, in order to improve the reliability, robustness and stability of the industrial computer system in complex electromagnetic environment. SUMMARY

[0006] The present application provides a self-adaptive environment industrial computer robust intelligent regulation method and system, which solves the technical problem of self-adaptive environment industrial computer intelligent regulation in related technologies.

[0007] The present application provides a self-adaptive environment industrial computer robust intelligent regulation method, which comprises the following steps:

[0008] establishing an electromagnetic interference spatio-temporal distribution model, the establishing an electromagnetic interference spatio-temporal distribution model comprising collecting multi-point electromagnetic interference data, preprocessing the electromagnetic interference data, constructing an electromagnetic interference spatio-temporal distribution interpolation model, and predicting electromagnetic interference temporal variation;

[0009] based on the electromagnetic interference spatio-temporal distribution model, evaluating task electromagnetic sensitivity and grading;

[0010] based on the electromagnetic interference spatio-temporal distribution model and the task electromagnetic sensitivity, performing task risk assessment and prediction;

[0011] based on the task risk assessment and prediction results, performing robust anti-interference resource dynamic allocation;

[0012] based on the anti-interference resource dynamic allocation, implementing task flexible scheduling execution, realizing robust intelligent scheduling of tasks in time and space dimensions, and improving task execution reliability and robustness.

[0013] As a further optimization scheme of the present application, in the establishing an electromagnetic interference spatio-temporal distribution model, the collecting multi-point electromagnetic interference data comprises:

[0014] deploying multiple electromagnetic interference monitoring sensors at different positions in an industrial field;

[0015] collecting electromagnetic interference intensity data at different spatial positions and time points, forming multi-dimensional spatio-temporal data points containing monitoring position coordinates (x i , y i , z i ), time stamp t i and corresponding electromagnetic interference intensity value I i .

[0016] As a further optimization scheme of the present application, the constructing an electromagnetic interference spatio-temporal distribution interpolation model comprises:

[0017] applying a clustering algorithm to the preprocessed electromagnetic interference data to identify interference source positions;

[0018] establishing a propagation function model based on physical attenuation law for each identified interference source:

[0019]

[0020] where d i (x, y, z, t) is the Euclidean distance from the spatial position (x, y, z) to the i-th interference source, represents the variation function of the i-th interference source in the time dimension;

[0021] constructing a continuous electromagnetic interference spatio-temporal distribution model:

[0022]

[0023] wherein E(x,y,z,t) represents an electromagnetic interference intensity function at spatial position (x,y,z) and time point t, denotes the summation of all interference sources, denotes the intensity of the i-th interference source, denotes the propagation function of the i-th interference source, n E denotes the total number of identified interference sources.

[0024] As a further optimization scheme of the present application, the evaluation task electromagnetic sensitivity and grading includes:

[0025] analyzing the task characteristic parameters of industrial control, and establishing a task characteristic description vector:

[0026] T i =(C i ,K i ,D i ,I i );

[0027] wherein C i is the task computing complexity, indicating the computing load and resource requirement of the task; K i is the task criticality, indicating the importance of the task to the overall function of the system; D i is the task time sensitivity, indicating the tolerance of the task to execution delay; I i is the task data integrity requirement, indicating the sensitivity of the task to data errors;

[0028] evaluating the electromagnetic sensitive characteristics of the industrial computer execution unit, and constructing an execution unit electromagnetic sensitivity description vector:

[0029] U j =(α j ,β j ,γ j );

[0030] wherein α j is the unit type sensitivity coefficient, indicating the inherent electromagnetic sensitivity of different types of hardware units; β j is the unit shielding coefficient, indicating the electromagnetic shielding degree of the hardware unit; γ j is the unit historical failure rate, indicating the historical failure statistical data in the electromagnetic interference environment;

[0031] calculating the sensitivity index S i of the task in a specific electromagnetic environment, and classifying the industrial control task into three levels of high sensitivity, medium sensitivity and low sensitivity.

[0032] As a further optimization scheme of the present application, the execution task risk assessment and prediction includes:

[0033] Combine the electromagnetic interference spatio-temporal distribution model and the task sensitivity index to construct a basic risk assessment model of the task at a specific spatio-temporal position;

[0034] Consider the historical execution of the task, and introduce a task historical failure record correction factor;

[0035] Consider the system resource competition, and introduce a current load state factor of the industrial computer;

[0036] The task risk assessment function is constructed by comprehensively considering the basic risk assessment model of the task at a specific spatio-temporal position, the task historical failure record correction factor, and the current load state factor of the industrial computer, so as to predict the failure risk probability of the task at a specific spatio-temporal position.

[0037] As a further optimization scheme of the present application, the anti-interference resource dynamic allocation includes:

[0038] Define an anti-interference resource pool of the industrial computer system, including computing resources, memory resources, communication bandwidth resources and anti-interference hardware resources;

[0039] Based on the task risk probability and the task importance level, the resource priority index of each task is calculated;

[0040] According to the task priority index, a dynamic resource allocation strategy is implemented;

[0041] Additional anti-interference resources are allocated to high-risk tasks, including processing units with better electromagnetic shielding, data redundancy backup and task execution clock frequency adjustment.

[0042] As a further optimization scheme of the present application, the task elastic scheduling execution includes:

[0043] Construct a spatio-temporal electromagnetic interference map to identify “electromagnetic safety area” and “electromagnetic danger area”;

[0044] Implement a task time rearrangement strategy to adjust the task execution time to avoid high interference period;

[0045] Implement a task space rearrangement strategy to allocate high-risk tasks to hardware units or physical positions with lower electromagnetic interference;

[0046] Construct a backup execution path for critical tasks to realize quick switching and recovery in case of failure.

[0047] As a further optimization scheme of the present application, the task elastic scheduling execution adopts a combination of improved shortest job first and priority scheduling algorithm, and the scheduling algorithm maintains a multi-priority task queue and generates an optimal scheduling scheme through a risk-aware scheduler.

[0048] As a further optimization scheme of the present application, the predicted electromagnetic interference time variation adopts a long short-term memory network model to model the electromagnetic interference intensity sequence of each spatial point, output the interference intensity prediction value of the next time window, and support the prediction of periodic interference, trend interference and burst interference.

[0049] A system for implementing adaptive environmental industrial computer intelligent regulation is used to execute the adaptive environmental industrial computer intelligent regulation method described above, and comprises:

[0050] An electromagnetic interference monitoring module is used to collect multi-point electromagnetic interference data and monitor the electromagnetic environment state of an industrial site in real time.

[0051] An interference modeling module is used to construct an electromagnetic interference space-time distribution model and predict the electromagnetic interference variation trend.

[0052] A task sensitivity evaluation module is used to analyze the characteristics of an industrial control task, evaluate the electromagnetic sensitivity of the task and grade it.

[0053] A risk evaluation module is used to combine the interference model and the sensitivity to predict the task execution risk.

[0054] A resource allocation module is used to dynamically allocate system anti-interference resources according to the task risk and priority.

[0055] A scheduling execution module is used to implement the space-time flexible scheduling of a task and guarantee reliable execution of the task.

[0056] The present application has the beneficial effects that: through the real-time updating of the electromagnetic interference space-time distribution model, the accurate grading of the task sensitivity, the dynamic adjustment of the risk evaluation and the space-time flexible scheduling of the task execution, the present application cooperates with multiple-level strategies to form a complete closed-loop feedback system, to build a robust electromagnetic interference adaptation mechanism, can actively perceive the environmental changes and make corresponding adjustments, improves the reliability and stability of the industrial control system in the complex electromagnetic environment, ensures the robust execution of the key task, specifically, through accurate electromagnetic interference modeling and risk evaluation, combined with differentiated resource allocation strategies, the execution success rate of the key task in the strong electromagnetic interference environment is improved, the system failure and data error caused by electromagnetic interference are effectively reduced; and the dynamic resource allocation and task space-time rearrangement strategy based on risk evaluation reduces the response time fluctuation of the key task in the electromagnetic interference environment, improves the response time guarantee rate in the extreme electromagnetic environment; at the same time, the differentiated resource allocation mechanism based on the task risk and importance realizes the accurate allocation of resources, compared with the traditional uniform task allocation mode, improves the overall throughput of the system, and under the same hardware conditions, more tasks can be processed. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1is a flow chart of a self-adaptive environment industrial computer robust intelligent regulation method of the present application;

[0058] Figure 2 is a detailed flow chart of constructing a time-space distribution model of electromagnetic interference of the present application;

[0059] Figure 3 is a detailed flow chart of constructing a task electromagnetic sensitivity evaluation model of the present application;

[0060] Figure 4 is a detailed flow chart of constructing a task execution risk evaluation model of the present application;

[0061] Figure 5 is a detailed flow chart of constructing a resource dynamic allocation model of the present application;

[0062] Figure 6 is a detailed flow chart of constructing a task flexible scheduling execution model of the present application. DETAILED DESCRIPTION

[0063] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide a better understanding of the subject matter described herein and can include changes, modifications, additions or omissions of the functions and arrangements of the elements discussed without departing from the scope of the present description. Various examples can omit, substitute or add various procedures or components as appropriate, and the examples described can be combined together to describe yet further examples. Additionally, some of the described features can be combined in a single example.

[0064] In at least one embodiment of the present application, a self-adaptive environment industrial computer robust intelligent regulation method is disclosed, as shown in Figures 1 to 5 includes the following steps:

[0065] Step 1, establishing a time-space distribution model of electromagnetic interference, which includes collecting multi-point electromagnetic interference data, preprocessing electromagnetic interference data, constructing an electromagnetic interference time-space distribution interpolation model, and predicting electromagnetic interference time variation;

[0066] This step uses multi-point monitoring technology to collect and quantify electromagnetic interference in an industrial environment, and constructs a time-space distribution model of electromagnetic interference, which specifically includes:

[0067] Step 1.1, collecting multi-point electromagnetic interference data;

[0068] Specifically, through multiple electromagnetic interference monitoring sensors deployed at different positions in the industrial field, electromagnetic interference intensity data at different spatial positions and time points are collected to form an original electromagnetic interference data set. The data recorded by each sensor includes the monitoring position coordinates (x i , y i , z i), timestamp t i and the corresponding electromagnetic interference intensity value I i The generation form is (x i y i , z i , t i I i (Multidimensional spatiotemporal data points)

[0069] Step 1.2: Preprocess electromagnetic interference data;

[0070] Specifically, the collected raw electromagnetic interference data undergoes preprocessing, including outlier detection and correction, data smoothing, and normalization, resulting in a cleaned electromagnetic interference dataset. For outlier data point P... outlier The following function is used for identification:

[0071] P outlier ={(x i y i , z i , t i I i )||I i -μ E |>3σ E};

[0072] Where, μ E σ is the mean value of the electromagnetic interference intensity. E The standard deviation is used. Identified outlier data points are replaced using the weighted average of their nearest neighbors.

[0073] Step 1.3: Construct a spatiotemporal distribution interpolation model for electromagnetic interference;

[0074] Specifically, based on the processed electromagnetic interference data, a continuous spatiotemporal distribution model of electromagnetic interference is constructed using a spatiotemporal interpolation algorithm. This model is represented as:

[0075]

[0076] Where E(x, y, z, t) represents the electromagnetic interference intensity function at spatial location (x, y, z) and time point t. This indicates that for all interference sources (from the 1st to the nth) E Summation of ( ) This represents the intensity of the i-th interference source. Let n represent the propagation function of the i-th interference source. E This indicates the total number of interference sources identified.

[0077] propagation function A model based on physical attenuation laws is adopted:

[0078]

[0079] Where, d i (x, y, z, t) is the Euclidean distance from the spatial location (x, y, z) to the i-th interference source. Let represent the change function of the i-th interference source in the time dimension.

[0080] In the specific implementation, interference source identification is accomplished through clustering algorithms. Based on the spatial distribution characteristics of the monitoring data, a density-based spatial clustering algorithm (DBSCAN) is used to identify the locations of potential interference sources. For each identified interference source, its intensity is fitted using the least squares method. and time-varying function Time-varying function Using Fourier series expansion:

[0081]

[0082] in, Let be the coefficients of the constant term in the Fourier series expansion. This indicates that for k from 1 to m E Summation, For the Fourier series expansion, cos(kω) E The coefficient of the t) term, For the Fourier series expansion sin(kω) E The coefficient of the t) term, ω E Let m be the fundamental frequency of the Fourier series, and m represent the frequency of the periodic variation of the interference. E The order of the Fourier series expansion is the highest harmonic contained in the Fourier series.

[0083] In practical applications in industrial workshops, spatiotemporal interpolation algorithms, combined with factory floor plans and equipment layout information, can accurately draw diagrams such as... Figure 2 The electromagnetic interference heatmap shown intuitively displays the interference intensity distribution in different areas. When transient electromagnetic interference events such as motor start-up and shutdown, and high-power equipment switching occur, this model can update the interference distribution in real time, providing an accurate basis for subsequent task scheduling.

[0084] Step 1.4: Predict the temporal variation of electromagnetic interference;

[0085] Specifically, by analyzing the temporal patterns of historical electromagnetic interference data, a predictive model for the temporal variation of electromagnetic interference is constructed to predict the distribution of electromagnetic interference at future points in time.

[0086] The prediction model employs time series analysis to analyze the electromagnetic interference intensity sequence {E(x, y, z, t1), E(x, y, z, t2), ..., E(x, y, z, t3)} for each spatial point (x, y, z).k )perform modeling, output the predicted value of the next time window wherein E(x, y, z, t1), E(x, y, z, t2), E(x, y, z, tk) represent the electromagnetic interference intensity at the spatial point (x, y, z) at the first time point t1, the second time point t2, the kth time point tk respectively, k k represent the predicted electromagnetic interference intensity value at the spatial point (x, y, z) at the time point t k+1

[0087] In a specific implementation, the electromagnetic interference time-varying prediction model adopts a long short-term memory network (LSTM) structure, which is particularly suitable for processing long-term dependencies of time series data. The LSTM network consists of an input layer, an LSTM layer and an output layer, wherein the input layer receives the normalized historical interference intensity data, the LSTM layer consists of multiple LSTM units, each LSTM unit contains an input gate, a forget gate and an output gate for controlling information flow. The network structure parameters are determined according to the time characteristics of electromagnetic interference in the actual industrial environment, and in most application scenarios, a 2-layer LSTM structure is adopted, each layer contains 64 LSTM units, the time step is 10 minutes, and the prediction window is 1 hour in the future.

[0088] The prediction model can identify and predict the following three typical electromagnetic interference time patterns in practical applications: periodic interference: periodic electromagnetic interference generated by equipment start-stop related to production shifts, regular maintenance, etc.; Trend interference: gradually increasing interference generated by the gradual decline of electromagnetic shielding performance caused by equipment aging; Sudden interference: sudden interference generated by equipment failure, temporary high-power equipment operation, etc.

[0089] In the application case of a metal processing workshop, the prediction model can predict the electromagnetic interference enhancement caused by the start of a large CNC equipment 15-20 minutes in advance, with an accuracy of 92%, providing sufficient preparation time for the advance scheduling of industrial computer tasks, effectively avoiding the failure caused by the execution of critical tasks during high interference periods.

[0090] Step 2, based on the electromagnetic interference space-time distribution model, evaluate the electromagnetic sensitivity of the task and grade it;

[0091] This step is based on the characteristics of industrial control tasks and the characteristics of execution hardware units to build a task electromagnetic sensitivity evaluation model, which divides industrial control tasks into different sensitivity levels, including:

[0092] Step 2.1, analyze the task characteristic parameters;

[0093] ​​​Specifically, the characteristic parameters of various tasks in the industrial computer are analyzed, and a task characteristic description vector T is established i .

[0094] The task characteristic description vector contains the following key parameters:

[0095] Task computing complexity C i : represents the computing load and resource demand of the task;

[0096] Task criticality K i : represents the importance of the task to the overall function of the system;

[0097] Task time sensitivity D i : represents the tolerance of the task to execution delay;

[0098] Task data integrity requirement I i : represents the sensitivity of the task to data errors.

[0099] The task characteristic vector is represented as: T i =(C i , K i , D i , I i );

[0100] Step 2.2, evaluate the electromagnetic sensitive characteristics of the execution unit;

[0101] Specifically, the sensitive characteristics of each execution unit of the industrial computer in the electromagnetic environment are evaluated, and an execution unit electromagnetic sensitivity description vector U j is constructed.

[0102] The execution unit sensitivity vector contains the following parameters:

[0103] Unit type sensitivity coefficient α j : the inherent electromagnetic sensitivity of different types of hardware units (such as CPU, memory, I / O interface);

[0104] Unit shielding coefficient β j : the electromagnetic shielding degree of the hardware unit;

[0105] Unit historical failure rate γ j : historical failure statistics data in the electromagnetic interference environment.

[0106] The execution unit sensitivity vector is represented as: U j =(α j , β j , γ j );

[0107] Step 2.3, calculate the task sensitivity index;

[0108] Specifically, the task characteristic vector is combined The execution unit sensitivity vector U j And the environmental electromagnetic feature E em Calculate the sensitivity index of the task in a specific electromagnetic environment

[0109]

[0110] Wherein, The electromagnetic sensitivity index of task i, f() represents the sensitivity calculation function, is a weight coefficient, used to balance the influence degree of each factor, and ||E em || represents the intensity norm of the environmental electromagnetic interference, and represents the comprehensive size of the electromagnetic interference intensity.

[0111] Step 2.4, divide the tasks into different sensitivity levels;

[0112] Specifically, according to the calculated sensitivity index The industrial control task is divided into three levels of high sensitivity, medium sensitivity and low sensitivity:

[0113] High sensitivity task: Extremely sensitive to electromagnetic interference, requiring the highest level of protection;

[0114] Medium sensitivity task: Has a certain sensitivity to electromagnetic interference and needs appropriate protection;

[0115] Low sensitivity task: Not sensitive to electromagnetic interference, can be executed normally in a high interference environment.

[0116] Wherein, And The preset sensitivity threshold can be dynamically adjusted according to system characteristics and environmental conditions.

[0117] Step 3, based on the electromagnetic interference space-time distribution model and the electromagnetic sensitivity of the task, perform task risk assessment and prediction;

[0118] This step is based on the electromagnetic interference space-time distribution model and the task sensitivity classification result to construct a task execution risk assessment model to predict the failure risk probability of the task in a specific space-time position, which specifically includes:

[0119] Step 3.1, construct a risk assessment basic model;

[0120] Specifically, the electromagnetic interference space-time distribution model E(x, y, z, t) constructed in step 1 and the task sensitivity index calculated in step 2 are combined Construct the task T j In the spatial position (xj y j , z j The basic model for risk assessment at time point t:

[0121]

[0122] Among them, T j Let task j be (x) j y j , z j ) represents the spatial coordinates of task j, t represents the time point, and R base E(x) represents the basic risk index, which is directly proportional to the task sensitivity and the intensity of environmental electromagnetic interference. j y j , z j ,t) represents the position (x j y j , z j The electromagnetic interference intensity at time t.

[0123] Step 3.2: Introduce a task history failure record correction factor;

[0124] Specifically, considering the historical execution of tasks, a task history failure record correction factor is introduced.

[0125] For each task T j The historical failure rate under different electromagnetic interference intensities was statistically analyzed, and a failure probability distribution function P was constructed. fail (T j E). Correction factor The calculation is as follows:

[0126]

[0127] in, This indicates that for k from 1 to m hist Summation, Indicates task T j The failure flag (1 for failure, 0 for success) for the k-th execution. The time decay weight for the k-th execution makes recent failures have a greater impact on the correction factor, m hist This represents the total number of historical records.

[0128] Step 3.3: Introduce the current load status factor of the industrial control computer;

[0129] Specifically, we introduce the current load state factor L(t) of the industrial control computer to consider the impact of system resource contention on task execution risk. The load state factor is expressed as:

[0130]

[0131] where CPU(t), MEM(t) and IO(t) represent the processor usage, memory occupancy and I / O load at time t, respectively, CPU max , MEM max and IO max represent the maximum capacity of processor usage, memory occupancy and I / O load, respectively, α L , β L and γ L are the weight coefficients of processor usage, memory occupancy and I / O load in the load state factor.

[0132] Step 3.4, build a complete task risk assessment function;

[0133] Specifically, based on the above factors, a complete task risk assessment function is built to calculate the failure risk probability of task T j at a specific spatiotemporal location:

[0134]

[0135] where R base (T j , x j , y j , z j , t) represents the basic risk index of task T j at spatial location (x j , y j , z j ) and time point t, reflecting the basic risk level of the task affected by electromagnetic interference at a specific spatiotemporal location; is the task history failure record correction factor; L(t) is the current load state factor of the industrial computer; E(x j , y j , z j , t) represents the electromagnetic interference intensity at location (x j , y j , z j ) and time t; P(failure|T j , E) represents the failure probability of task T j given the electromagnetic interference environment E; σ is a Sigmoid function used to map the risk index to the interval [0, 1], representing the failure probability. Through this function, the system can dynamically assess the execution risk of each task under the current and predicted future electromagnetic environment, providing a basis for subsequent resource allocation and task scheduling decisions.

[0136] In a specific implementation, the task risk assessment model is trained and optimized using the Gradient Boosting Decision Tree (GBDT) algorithm. This algorithm integrates multiple weak learners (decision trees) to build a strong learner, which can effectively handle the complex relationships between task characteristics, historical execution records, system load states, and electromagnetic interference. The training data set includes historical records of the industrial computer system running in different electromagnetic environments, with each record containing electromagnetic interference intensity, task characteristics, system load state, and task execution result (success / failure).

[0137] During model training, the five-fold cross-validation method is used to evaluate model performance, with AUC (Area Under the Curve), precision, and recall as evaluation indicators. In actual deployment, the model is retrained every 3 months using newly collected data to adapt to changes in industrial environments and task characteristics.

[0138] In a distributed control system (DCS) application case in a certain chemical plant, the risk assessment model achieved a failure risk prediction accuracy rate of 94.5% for high-sensitivity tasks such as key parameter acquisition and safety interlock control, which was 22 percentage points higher than the traditional rule-based method. Especially during periods of intense electromagnetic interference (such as large motor start-stop processes), the model can accurately predict the failure risk of tasks in the marginal state, providing a reliable basis for task scheduling and resource allocation, effectively avoiding local system failures caused by electromagnetic interference.

[0139] Step 4, according to the task risk assessment and prediction results, execute robust anti-interference resource dynamic allocation;

[0140] This step is based on the task risk assessment results to build a resource dynamic allocation model to provide differentiated resource protection for tasks of different risk levels, specifically including:

[0141] Step 4.1, define the anti-interference resource pool of the industrial computer system that can be allocated;

[0142] Specifically, define the anti-interference resource pool R pool of the industrial computer system that can be allocated, including computing resources R cpu , memory resources R mem , communication bandwidth resources R bw , and anti-interference hardware resources R anti (such as reserved redundant processing units, backup storage, etc.). The resource pool is represented as:

[0143] R pool ={R cpu , R mem , R bw , R anti};

[0144] Each type of resource has its upper limit on the total amount:

[0145]

[0146] wherein R cpu represents the currently allocated CPU resource amount, represents the maximum available amount of CPU resource;

[0147] R mem represents the currently allocated memory resource amount, represents the maximum available amount of memory resource;

[0148] R bw represents the currently allocated communication bandwidth resource amount, represents the maximum available amount of communication bandwidth resource;

[0149] R anti represents the currently allocated anti-interference hardware resource amount, represents the maximum available amount of anti-interference hardware resource.

[0150] Step 4.2, calculating the task resource priority index;

[0151] Specifically, based on the task risk probability R(T j , t) and the task importance level calculated in step 3, the resource priority index of each task is calculated.

[0152]

[0153] wherein, and are the weight coefficients of risk factors and task importance in the resource priority index, respectively. The task importance level is predefined by the system administrator, reflecting the importance of the task to the overall function of the system.

[0154] Step 4.3, implementing a dynamic resource allocation strategy;

[0155] Specifically, according to the task priority index , a dynamic resource allocation strategy is implemented. For each type of resource R k , the resource share obtained by the task T j is calculated as follows:

[0156]

[0157] wherein, represents the maximum available amount of the kth type of resource, is the task T j ​the demand weight of the kth resource, the demand weight of the kth resource for the ith task, n task the total number of tasks in the current system, the resource priority index of the ith task at time t, the resource priority index of task T j the resource priority index of task T task at time t, the summation of i from 1 to n task .

[0158] This step ensures that high-priority tasks can obtain more resource shares.

[0159] Step 4.4, additional allocation of anti-interference resources for high-risk tasks;

[0160] Specifically, for high-risk tasks ( where, the benchmark threshold of task risk probability, when the task risk probability exceeds this threshold, it is determined as a high-risk task), additional allocation of anti-interference resources R anti , including: allocating processing units with better electromagnetic shielding for computing tasks; increasing redundant backup of data storage and transmission; adjusting the clock frequency of task execution to reduce sensitivity to electromagnetic interference.

[0161] For particularly critical and high-risk tasks, implement a resource exclusive strategy to avoid sharing critical hardware resources with other tasks and minimize the impact of electromagnetic interference. The resource allocation strategy can be expressed as a conditional allocation function:

[0162]

[0163] where, the anti-interference resource allocation amount obtained by task T j at time t; the total amount of exclusive anti-interference resources; the total amount of shareable anti-interference resources; R(T j , t) represents the risk probability of task T j at time t; the critical threshold of risk; the benchmark threshold of risk; the importance level of task T j ; the critical threshold of importance; the resource priority index of task T j at time t; H risk represents the set of high-risk tasks, the summation of all tasks in the set of high-risk tasks.

[0164] Step 5, based on anti-interference resource dynamic allocation, implement task flexible scheduling execution, realize the robust intelligent scheduling of tasks in time and space dimensions, improve the reliability and robustness of task execution;

[0165] This step is based on the aforementioned electromagnetic interference space-time distribution model, task risk assessment results and resource priority allocation strategy analysis results, etc. The task flexible scheduling execution model is constructed to realize the intelligent scheduling of tasks in time and space dimensions to avoid high electromagnetic interference areas and time periods, improve the reliability of task execution, which specifically includes:

[0166] Step 5.1, construct a space-time electromagnetic interference map;

[0167] Specifically, the space-time electromagnetic interference map is constructed to identify the "electromagnetic safety area" and "electromagnetic danger area" in the industrial computer running environment. Based on the electromagnetic interference space-time distribution model E(x,y,z,t) established in step 1, the industrial computer environment is divided into different levels of areas:

[0168] Zone safe ={(x,y,z,t)|E(x,y,z,t)<E safe};

[0169] Zone warning ={(x,y,z,t)|E safe ≤E(x,y,z,t)<E danger};

[0170] Zone danger ={(x,y,z,t)|E(x,y,z,t)≥E danger};

[0171] Where (x,y,z,t) represents the spatial position coordinates (x,y,z) and time t, E safe and E danger are the electromagnetic interference safety threshold and electromagnetic interference danger threshold respectively, which can be dynamically adjusted according to the overall tolerance of the system to electromagnetic interference, Zone safe represents the electromagnetic safety area, i.e. the area with interference intensity lower than the safety threshold; Zone warning represents the electromagnetic warning area, i.e. the area with interference intensity between the safety threshold and the danger threshold; Zone danger represents the electromagnetic danger area, i.e. the area with interference intensity higher than the danger threshold.

[0172] Step 5.2, implement task time rearrangement strategy;

[0173] Specifically, the task time rearrangement strategy is implemented, and the task execution time is adjusted according to the time distribution characteristics of electromagnetic interference to avoid high interference periods. For task T j , its optimal execution time window t opt is calculated as follows:

[0174]

[0175] where t current is the current system time, t deadline is the deadline that the task must complete, R(T j , t) is the task risk probability calculated in step 3, and argmin represents the argument value when the objective function takes the minimum value.

[0176] Step 5.3, implement the task space rearrangement strategy;

[0177] Specifically, the task space rearrangement strategy is implemented, and high-risk tasks are assigned to hardware units or physical locations with lower electromagnetic interference. For task T j , its optimal execution location (X opt , y opt , z opt ) is calculated as follows:

[0178]

[0179] where (X opt , y opt , z opt represents the optimal execution location coordinates of task T j , argmin represents the argument value when the objective function takes the minimum value, AvailableUnits represents the set of available hardware execution units, the elements in the set are hardware units that can be used for task execution, E(x, y, z, t scheduled ) represents the electromagnetic interference intensity at location (x, y, z) and planned execution time t scheduled , is the task sensitivity index.

[0180] Step 5.4, construct a backup execution path for critical tasks;

[0181] Specifically, a backup execution path is constructed for critical tasks to achieve rapid switching and recovery in case of failure. For high-importance task T j ( where represents the importance level of task T j , represents the critical threshold of importance), backup execution resources B jA state synchronization algorithm is established between the primary and backup execution units. When an anomaly is detected in the primary execution path, the automatic switchover condition is triggered.

[0182]

[0183] Among them, Switch(T) j ) represents task T j Whether to switch to the backup execution path (Boolean value, E) switch To switch the trigger electromagnetic interference threshold, if the threshold is higher than this value, a backup execution path switch will be triggered, resulting in Error(T). j ) represents task T j The number of errors detected during execution.

[0184] The task elastic scheduling execution model employs a combination of an improved Shortest Job First (SJF) and priority scheduling algorithms in its implementation. The system maintains a multi-priority task queue, with each queue using a time-space risk-weighted sorting strategy. The core of the scheduling algorithm is a dynamically adaptive risk-aware scheduler. This scheduler receives input from a risk assessment model and, combined with task deadline constraints, resource requirements, and the current system state, generates the optimal scheduling scheme.

[0185] The main steps of the scheduling algorithm are as follows:

[0186] All tasks to be scheduled are initially sorted according to risk assessment results and priority;

[0187] For high-risk tasks ( Among them, R(T) j ,t) represents task T j The risk assessment value at time t, (This represents the risk threshold for high-risk tasks) Execution time window optimization, finding the execution time period with the lowest electromagnetic interference within the future time interval Δt;

[0188] Optimize the spatial location of medium- and high-risk tasks by assigning tasks to hardware execution units with lower electromagnetic interference.

[0189] Generate primary and backup execution plans for critical tasks to ensure a quick switch to the backup path in case the primary execution path fails; periodically (every 30 seconds) reassess the current environment and task status and dynamically adjust the scheduling plan.

[0190] By executing steps 1 to 5, the adaptive environmental industrial control computer intelligent control method provided in this embodiment can achieve accurate perception, quantitative assessment and intelligent response to electromagnetic interference in industrial sites, ensuring the reliable operation of the industrial control computer system in complex electromagnetic environments.

[0191] The embodiment realizes accurate perception and intelligent response to electromagnetic interference in an industrial environment through technical means such as constructing an electromagnetic interference space-time distribution model, task electromagnetic sensitivity evaluation and classification, task risk assessment and prediction, anti-interference resource dynamic allocation, and task elastic scheduling execution, and has the following significant technical effects:

[0192] System reliability is significantly improved: through accurate electromagnetic interference modeling and risk assessment, combined with differentiated resource allocation strategies, the success rate of critical task execution in a strong electromagnetic interference environment is improved, effectively reducing system failures and data errors caused by electromagnetic interference. Especially for high sensitivity tasks, the failure rate is reduced.

[0193] Real-time assurance capability is enhanced: dynamic resource allocation and task space-time rearrangement strategies based on risk assessment are adopted, which reduces the response time fluctuation of critical tasks in the electromagnetic interference environment, improves the response time guarantee rate in extreme electromagnetic environments, and is much higher than traditional methods.

[0194] System resource utilization efficiency is improved: a differentiated resource allocation mechanism based on task risk and importance is implemented, which realizes precise resource allocation. Compared with the traditional uniform task allocation method, the overall throughput of the system is improved, more tasks can be processed under the same hardware conditions, and the balance between electromagnetic interference protection and system performance is achieved.

[0195] Strong environmental adaptability: through real-time monitoring and prediction of electromagnetic interference, combined with adaptive task scheduling strategies, the system can automatically adjust the protection strategy with the change of the environment, and can adapt to electromagnetic interference environments of different degrees and different distribution characteristics without human intervention, and the applicability is greatly improved.

[0196] Reliability cost is reduced: compared with the traditional method of improving electromagnetic compatibility through hardware reinforcement, the embodiment adopts an intelligent control strategy combining software and hardware, which reduces the system hardware cost while obtaining the same reliability, reduces the time and energy consumption caused by task retry and system restart, and improves the service life of industrial control equipment in harsh environments.

[0197] The adaptive environment industrial computer intelligent control method provided by the embodiment effectively solves the problem of reliable operation of industrial computer systems in an industrial electromagnetic interference environment, and has significant technical progress and practical value.

[0198] Real application examples of the embodiment

[0199] Application scenario description

[0200] The embodiment has been practically applied in an industrial control system of a rolling mill in a large steel plant. The rolling mill covers an area of about 15000 square meters, and a large number of high-power electrical equipment are deployed inside, including variable frequency motors, electric arc furnaces, large transformers, etc. These devices generate intense electromagnetic interference during operation and show obvious spatiotemporal non-uniform distribution characteristics. The industrial computer system in the workshop is responsible for controlling the key parameters of the rolling production line, including rolling pressure, speed control, temperature monitoring, safety interlocking, etc. The system contains 24 industrial computers distributed in different positions in the workshop, connecting more than 500 I / O points.

[0201] Before applying the method, the rolling mill industrial control system was facing serious electromagnetic interference problems. Especially during the start-stop process of large motors, problems such as control parameter acquisition errors, communication interruptions, and control instruction execution failures often occurred, with an average of 3-5 system partial failures per week, seriously affecting production efficiency and product quality. After applying the method, the system reliability has been significantly improved, achieving stable operation in a complex electromagnetic environment.

[0202] Implementation process instance

[0203] Step 1 instance: In the rolling mill, we first deployed electromagnetic interference monitoring equipment.

[0204] As shown in Table 1, 12 electromagnetic interference monitoring sensors were installed in the workshop, covering the areas around key equipment and the areas where industrial computers were concentrated.

[0205] Table 1: Electromagnetic interference monitoring sensor deployment

[0206] Sensor number Installation location Monitoring range Data acquisition frequency S01 Main rolling mill control cabinet Radius 5m range 100Hz S02 Frequency converter area Radius 8m range 100Hz S03 Electric arc furnace control area Radius 10m range 200Hz S04 Industrial computer server room Radius 3m range 100Hz ... ... ... ... S12 Finished product area control cabinet Radius 4m range 50Hz

[0207] By continuously collecting data for 30 days, more than 25 million data points were accumulated to form the original electromagnetic interference data set. Applying clustering algorithm for interference source identification, 7 main electromagnetic interference sources in the workshop were identified, as shown in Table 2.

[0208] Table 2: Characteristics of identified main electromagnetic interference sources

[0209] Interference source number Location Interference type Interference intensity (relative value) Time characteristic I01 Main motor area Frequency converter harmonic 0.85 Related to motor start-stop I02 Electric arc furnace Pulse interference 0.92 Periodic fluctuation I03 Main transformer Low-frequency interference 0.65 Related to load ... ... ... ... ... I07 Welding area High-frequency interference 0.58 Intermittent

[0210] Based on the identified interference source information, a spatiotemporal interpolation algorithm was applied to construct a rolling mill electromagnetic interference distribution model, generating an electromagnetic interference heat map to visually display the interference intensity distribution in different areas and different time periods. Especially for the interference changes during the start-up process of the main motor, the LSTM model successfully predicted the trend of interference intensity, with a prediction accuracy of 91.6%.

[0211] Example for Step 2: In this application, we performed sensitivity assessment and classification for all tasks in the steel rolling control system.

[0212] First, we analyzed the characteristics of tasks in the system, as shown in Table 3.

[0213] Table 3: Characteristic parameters of typical tasks in the system

[0214] Task type Computational complexity Criticality Time sensitivity Data integrity requirement Rolling pressure control 0.7 0.95 0.9 0.9 Temperature monitoring 0.3 0.7 0.5 0.8 Speed adjustment 0.6 0.9 0.95 0.85 Safety interlock 0.4 1.0 1.0 1.0 Data recording 0.5 0.4 0.3 0.7

[0215] Combining the sensitive characteristics of execution units and the electromagnetic features of the environment, we calculated the sensitivity index of each task and classified them according to the preset threshold (S high = 0.75, S medium = 0.5). The results are shown in Table 4.

[0216] Table 4: Classification results of task sensitivity

[0217] Task type Sensitivity index Sensitivity level Rolling pressure control 0.82 High sensitivity Temperature monitoring 0.61 Medium sensitivity Speed adjustment 0.88 High sensitivity Safety interlock 0.93 High sensitivity Data recording 0.48 Low sensitivity

[0218] Example for Step 3: Based on the established electromagnetic interference model and task sensitivity classification, we performed risk assessment for critical tasks in the system. For the "rolling pressure control" task, we considered the base risk index, historical failure records, and system load status, and established a risk assessment model through gradient boosting decision tree algorithm.

[0219] In practical application, this model successfully predicted the impact of electromagnetic interference caused by motor starting on the rolling pressure control task. The predicted peak time of electromagnetic interference was 17 minutes earlier than the actual time, and the prediction accuracy of task failure risk probability was 93.4%.

[0220] Examples for Step 4 and Step 5: Based on the risk assessment results, the system implemented differentiated resource allocation strategies and task flexible scheduling.

[0221] Table 5 shows the resource allocation and scheduling optimization for different types of tasks in the system.

[0222] Table 5: Resource allocation and scheduling optimization for different types of tasks

[0223]

[0224] When the intensity of electromagnetic interference exceeds the preset threshold, the system automatically increases the resource allocation proportion of critical tasks and enables backup execution units. In a strong interference event caused by an electric arc furnace starting, the system successfully transferred high-sensitivity tasks to low-interference areas for execution, ensuring the stable operation of tasks.

[0225] Technical effect verification

[0226] The implementation mode has achieved remarkable technical effects in the actual application of the steel rolling workshop of the steel plant. We focus on verifying the two key technical effects of system reliability improvement and real-time guarantee.

[0227] 1. System reliability improvement effect:

[0228] We counted the system failure conditions within 6 months before and after the application, focusing on system abnormalities and task failures caused by electromagnetic interference.

[0229] As shown in Table 6, after applying the method, the overall system failure rate is significantly reduced.

[0230] Table 6: System failure statistics comparison before and after application

[0231] Fault type Fault times per month before application Fault times per month after application Reduction ratio Control parameter acquisition error 15.3 1.2 92.2％ Communication interruption 8.6 0.5 94.2％ Control command execution failure 6.2 0.3 95.2％ System local restart 4.5 0.2 95.6％ Total 34.6 2.2 93.6％

[0232] Especially for high sensitivity tasks, the execution success rate in strong electromagnetic interference environment is significantly improved. In the strongest interference environment (relative intensity > 0.8), the execution success rate before application is only 68.5%, while after application it is improved to 98.7%, an increase of 30.2 percentage points.

[0233] 2. Real-time guarantee effect:

[0234] We also verified the real-time guarantee capability of the system in the electromagnetic interference environment.

[0235] As shown in Table 7, after applying the method, the response time fluctuation of key tasks is significantly reduced, especially in strong interference environment.

[0236] Table 7: Task response time statistics before and after application (unit: ms)

[0237]

[0238] In the extreme electromagnetic interference environment (relative intensity > 0.8), the system response time guarantee rate (probability of responding within the required time window) before application is only 76.3%, while after application it is improved to 99.5%, an increase of 23.2 percentage points.

[0239] The application results of the implementation mode in the actual industrial environment fully prove its effectiveness and practical value, significantly improving the reliability and real-time performance of industrial control systems in complex electromagnetic environments, providing reliable control guarantee for industrial production.

[0240] The above describes the embodiments of the present application, but the embodiments are not limited to the above specific embodiments, and the above specific embodiments are only illustrative but not restrictive, and the ordinary skilled in the art can make more equivalent embodiments under the inspiration of the embodiments, which are all within the protection scope of the embodiments.

Claims

1. A robust intelligent control method for an adaptive environment industrial control computer, characterized in that, Includes the following steps: Establishing a spatiotemporal distribution model for electromagnetic interference (EMI) involves collecting multi-point EMI data, preprocessing EMI data, constructing an EMI spatiotemporal distribution interpolation model, and predicting the temporal variation of EMI. Based on the spatiotemporal distribution model of electromagnetic interference, the electromagnetic susceptibility of the task is assessed and classified. Assessing and classifying the electromagnetic susceptibility of a task includes: Analyze the characteristic parameters of industrial control tasks and establish task characteristic description vectors: in, The computational complexity of the task represents the computational load and resource requirements of the task. The criticality of a task indicates its importance to the overall functionality of the system. Task time sensitivity represents the task's tolerance for execution delays; For task data integrity requirements, this indicates the task's sensitivity to data errors; Evaluate the electromagnetic susceptibility characteristics of industrial control computer execution units and construct an electromagnetic susceptibility description vector for the execution units: in, This is the unit type sensitivity coefficient, representing the inherent electromagnetic sensitivity of different types of hardware units; The shielding coefficient represents the degree of electromagnetic shielding of the hardware unit. The historical failure rate of the unit represents the historical failure statistics in an electromagnetic interference environment; The sensitivity index of the computational task in the electromagnetic environment Industrial control tasks are divided into three levels: high sensitivity, medium sensitivity, and low sensitivity. Based on the spatiotemporal distribution model of electromagnetic interference and the electromagnetic sensitivity of the task, risk assessment and prediction of the task are performed. Based on the results of the mission risk assessment and prediction, implement robust and anti-interference dynamic allocation of resources; Based on the dynamic allocation of anti-interference resources, elastic scheduling and execution of tasks are implemented to achieve robust intelligent scheduling of tasks in the time and space dimensions, thereby improving the reliability and robustness of task execution.

2. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, In establishing the spatiotemporal distribution model of electromagnetic interference, the collection of multi-point electromagnetic interference data includes: Multiple electromagnetic interference monitoring sensors are deployed at different locations in the industrial site; Collect electromagnetic interference intensity data at different spatial locations and time points to form a data set including the coordinates of the monitoring locations. timestamp and the corresponding electromagnetic interference intensity value Multidimensional spatiotemporal data points.

3. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, The construction of the spatiotemporal distribution interpolation model for electromagnetic interference includes: Clustering algorithms are applied to the preprocessed electromagnetic interference data to identify the location of interference sources; For each identified interference source, establish a propagation function model based on the physical attenuation law: ； in, For spatial location To the Euclidean distance between each interference source Indicates the first The function of change of each interference source in the time dimension; Constructing a continuous spatiotemporal distribution model of electromagnetic interference: ； in, Indicates spatial location and time point Electromagnetic interference intensity function at the location, This represents the summation over all interference sources. Indicates the first The intensity of each interference source, Indicates the first The propagation function of each interference source This indicates the total number of interference sources identified.

4. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, The risk assessment and prediction for the execution task includes: By combining the spatiotemporal distribution model of electromagnetic interference and the mission sensitivity index, a basic model for risk assessment of missions in spatiotemporal locations is constructed. Considering the historical execution of tasks, a task failure record correction factor is introduced; To address system resource contention, a current load state factor for the industrial control computer is introduced. By integrating the risk assessment model of the mission's spatiotemporal location, the correction factor of the mission's historical failure records, and the current load status factor of the industrial control computer, a complete mission risk assessment function is constructed to predict the probability of mission failure at its spatiotemporal location.

5. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, The implementation of robust and interference-resistant dynamic resource allocation includes: Define the anti-interference resource pool that can be allocated to the industrial control computer system, including computing resources, memory resources, communication bandwidth resources, and anti-interference hardware resources; Calculate the resource priority index for each task based on the task risk probability and task importance level; Implement a dynamic resource allocation strategy based on task priority index; Additional anti-interference resources are allocated to high-risk tasks, including processing units with better electromagnetic shielding, data redundancy backup, and task execution clock frequency adjustment.

6. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, The implementation of elastic scheduling and execution of tasks includes: Construct a spatiotemporal electromagnetic interference map to identify "electromagnetic safe zones" and "electromagnetic hazardous zones"; Implement a task time rescheduling strategy to adjust task execution times and avoid periods of high interference. Implement a task space rearrangement strategy to allocate high-risk tasks to hardware units or physical locations with low electromagnetic interference. Build backup execution paths for critical tasks to enable rapid switching and recovery in case of failure.

7. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, The elastic scheduling of tasks adopts a combination of improved shortest job first and priority scheduling algorithms. The scheduling algorithm maintains a multi-priority task queue and generates the optimal scheduling scheme through a risk-aware scheduler.

8. The adaptive environment industrial control computer robust intelligent control method according to claim 1, characterized in that, The prediction of electromagnetic interference time variation adopts a long short-term memory network model, which models the electromagnetic interference intensity sequence of each spatial point and outputs the predicted interference intensity value for the next time window, supporting the prediction of periodic interference, trend interference and sudden interference.

9. A system for realizing intelligent control of an adaptive environment industrial control computer, used to execute the robust intelligent control method of an adaptive environment industrial control computer as described in any one of claims 1-8, characterized in that, include: Electromagnetic interference monitoring module is used to collect electromagnetic interference data from multiple points and monitor the electromagnetic environment status of industrial sites in real time. The interference modeling module is used to construct a spatiotemporal distribution model of electromagnetic interference and predict the changing trend of electromagnetic interference. The task sensitivity assessment module is used to analyze the characteristics of industrial control tasks, assess the electromagnetic sensitivity of tasks, and classify them. The risk assessment module is used to combine interference models and sensitivity to predict the risks of task execution; The resource allocation module is used to dynamically allocate system anti-interference resources based on task risk and priority. The scheduling and execution module is used to implement spatiotemporal elastic scheduling of tasks to ensure reliable task execution.

Images