Method and system for remotely monitoring hardware state of industrial personal computer

Abstract

The invention provides an industrial personal computer hardware state remote monitoring method and system, and the method comprises the steps: carrying out the synchronous collection of the specified hardware parameters of different hardware units in a target industrial personal computer, and obtaining the original collection data; the method comprises the following steps: preprocessing original collected data to generate an input vector set for hardware health state assessment; inputting the input vector set into a health assessment model, and determining a multi-dimensional health degree index corresponding to each hardware unit; comparing the multi-dimensional health degree index with a preset health threshold value, and generating a hardware abnormal state recognition result; and sending a remote control instruction to the target industrial personal computer for remote control according to the abnormal state recognition result. Through the implementation of the scheme, the key parameters of different hardware units are collected, unified structured data are formed, the health assessment model is utilized to generate the multi-dimensional health indexes, dynamic updating of the remote control strategy is finally driven, and the operation and maintenance efficiency of the industrial personal computer can be effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of industrial computers, and in particular to a method and system for remotely monitoring the hardware status of an industrial computer. Background Art

[0002] With the rapid advancement of information technology in industrial field equipment, industrial computers (ICs), as the core computing and control units in industrial control systems, are widely deployed in key scenarios such as automated production lines, rail transit control, energy dispatch centers, and data acquisition front-ends. These IPCs operate year-round in complex environments characterized by high temperature, high humidity, high dust levels, and strong electromagnetic interference. Their stability is crucial to the safety and continuity of the entire industrial system. However, current mainstream hardware status monitoring methods still rely on regular manual inspections and simple local SNMP monitoring. These methods lack in-depth analysis of hardware operating status and remote management capabilities. In particular, the system cannot provide effective early warnings before potential failures in key components (such as memory, motherboards, and network ports) occur, often leading to reactive repairs after sudden outages, severely impacting equipment availability and operational efficiency. Summary of the Invention

[0003] The present application provides a method and system for remotely monitoring the hardware status of an industrial computer, which is used to solve the problem in related technologies that the system cannot provide effective early warning before potential failures of key components occur.

[0004] In a first aspect, the present application provides a method for remotely monitoring the hardware status of an industrial computer, the method comprising: Synchronously collect the specified hardware parameters of different hardware units in the target industrial computer to obtain the original collected data; Generating an input vector set for hardware health status assessment by preprocessing the original collected data; Inputting the input vector set into a health assessment model to determine a multi-dimensional health index corresponding to each hardware unit; Comparing the multi-dimensional health index with a preset health threshold to generate a hardware abnormality status identification result; Send a remote control instruction to the target industrial computer for remote control according to the abnormal state identification result.

[0005] Optionally, in a first implementation of the first aspect of the present application, the step of synchronously collecting specified hardware parameters of different hardware units in the target industrial computer to obtain raw collected data includes: Collect specified hardware parameters based on multiple types of sensors deployed on different hardware units in the target industrial computer and build an initial perception matrix of hardware parameters; Generate a synchronous acquisition sample set by performing time series reorganization and channel classification on the initial perception matrix of the hardware parameters; The synchronously collected sample set is subjected to field structured encoding according to a preset format template to generate original collected data with field labels.

[0006] Optionally, in a second implementation of the first aspect of the present application, the step of generating an input vector set for hardware health status assessment by preprocessing the original collected data includes: Obtaining a set of health indicator parameters corresponding to each hardware unit by performing field screening and numerical analysis on the original collected data; Performing data cleaning on the health indicator parameter set based on preprocessing rules to construct an intermediate parameter vector sequence that meets the model input requirements; Generate an initial input vector set by performing data normalization processing on the intermediate parameter vector sequence; The initial input vector set is batch reconstructed according to a predefined time window and hardware type labels to generate an input vector set for hardware health status assessment.

[0007] Optionally, in a third implementation of the first aspect of the present application, the step of inputting the input vector set into the health assessment model to determine the multidimensional health index corresponding to each hardware unit includes: Constructing a multi-branch deep feature extraction network according to the hardware type labels of the input vector set; Performing parallel feature extraction on the input vector set through each branch network of the multi-branch deep feature extraction network to generate a feature vector set containing time series correlation features; A multi-head attention weight allocation mechanism is constructed based on the channel dimension of the feature vector set, and a dynamic weight coefficient matrix is ​​generated by calculating the cross-channel feature correlation; Performing attention weighted calculation on the feature vector set according to the dynamic weight coefficient matrix to generate a weighted feature matrix; Inputting the weighted feature matrix into a multi-source fusion network according to a predefined health dimension mapping rule to generate a multi-dimensional health indicator sequence; The final multidimensional health index corresponding to each hardware unit is determined by iteratively compensating the multidimensional health index sequence and the health correction function; wherein the health correction function is a function constructed based on the physical coupling relationship between the hardware units.

[0008] Optionally, in a fourth implementation of the first aspect of the present application, the step of comparing the multi-dimensional health indicator with a preset health threshold to generate a hardware abnormality status identification result includes: Generate an initial abnormal vector set based on the level-by-level matching of the various dimensional parameters of the multi-dimensional health indicator with the preset health threshold; Performing cross-unit correlation correction on the initial abnormal vector set to generate a corrected abnormal vector set; Perform multi-path reasoning on the modified abnormal vector set according to a preset abnormal decision tree to generate an abnormal state intermediate result; Backtrack verification is performed based on the time series characteristics of the abnormal state intermediate result and the historical fault database, and the hardware abnormal state identification result is output.

[0009] Optionally, in a fifth implementation of the first aspect of the present application, the step of sending a remote control instruction to the target industrial computer for remote control according to the abnormal state identification result includes: generating a control instruction set according to the confidence rating and the fault location code in the hardware abnormal state identification result; Performing layered data encapsulation on the control instruction set by an asymmetric encryption algorithm based on a preset security policy library to generate an encrypted instruction data packet; Building a dual-channel redundant transmission link based on the communication protocol characteristics of the target industrial computer, fragmenting and reassembling the encrypted instruction data packet and then synchronously sending it to the target industrial computer; By dynamically reloading the parameter threshold of the encrypted instruction data packet after fragmentation and reorganization, the corresponding monitoring parameter update feedback result is generated.

[0010] Optionally, in a sixth implementation of the first aspect of the present application, the method further includes: Constructing a three-dimensional device connection relationship matrix based on the physical connection topology diagram and communication interface configuration table of the target industrial computer; Generate three-dimensional topological rendering data by performing spatial weight fusion on the three-dimensional device connection relationship matrix and the multi-dimensional health index; Construct a communication status waveform dataset based on the real-time collected communication interface electrical parameter fluctuation data; A dynamic visual monitoring interface is generated by performing multi-layer overlay processing on the three-dimensional topology rendering data, the health status code, and the communication status waveform data set.

[0011] A second aspect of the present application provides a remote monitoring device for the hardware status of an industrial computer, the remote monitoring device for the hardware status of an industrial computer comprising: The acquisition module is used to synchronously acquire the specified hardware parameters of different hardware units in the target industrial computer and obtain the original acquisition data; A preprocessing module, configured to generate an input vector set for hardware health status assessment by preprocessing the original collected data; a determination module, configured to input the input vector set into a health assessment model to determine a multidimensional health index corresponding to each hardware unit; A comparison module, configured to compare the multi-dimensional health indicator with a preset health threshold to generate a hardware abnormality status identification result; The control module is used to send a remote control instruction to the target industrial computer for remote control according to the abnormal state recognition result.

[0012] A third aspect of an embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the processor is used to execute a computer program stored on the memory. When the processor executes the computer program, it implements each step of the method for remotely monitoring the hardware status of an industrial computer provided in the first aspect of the embodiment of the present application.

[0013] The fourth aspect of the embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the remote monitoring method for the hardware status of an industrial computer provided in the first aspect of the embodiment of the present application are implemented.

[0014] In summary, according to the method and system for remote monitoring of the hardware status of an industrial computer provided by the present application, the specified hardware parameters of different hardware units in the target industrial computer are synchronously collected to obtain the original collected data; the input vector set for hardware health status evaluation is generated by preprocessing the original collected data; the input vector set is input into the health evaluation model to determine the multidimensional health index corresponding to each hardware unit; the multidimensional health index is compared with the preset health threshold to generate a hardware abnormal state identification result; and a remote control instruction is sent to the target industrial computer for remote control based on the abnormal state identification result. Through the implementation of the present application, the key parameters of different hardware units are collected to form unified structured data, and the health evaluation model is used to generate multidimensional health indicators, which ultimately drives the dynamic update of the remote control strategy, which can effectively improve the operation and maintenance efficiency of the industrial computer. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flowchart of a method for remotely monitoring the hardware status of an industrial computer provided in an embodiment of the present application; Figure 2 A schematic diagram of a program module of a device for remotely monitoring the hardware status of an industrial computer provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of this application.

[0017] In order to solve the problem in the related art that the system cannot provide effective early warning before the potential failure of key components, the embodiment of the present application provides a remote monitoring method for the hardware status of an industrial computer, such as Figure 1 The flowchart of the method for remotely monitoring the hardware status of an industrial computer provided in this embodiment includes the following steps: Step 110: synchronously collect the specified hardware parameters of different hardware units in the target industrial computer to obtain original collected data.

[0018] Specifically, the system synchronously collects specified hardware parameters from various hardware units within the target industrial computer using parallel sensing technology based on multiple sensor types. By deploying sensor modules with diverse sensing capabilities in key components such as the motherboard, memory, hard drive, and network port, it collects hardware operating status parameters such as voltage, current, clock frequency, error correction code error rate, SMART (Self-Monitoring Analysis and Reporting Technology) attributes, and network data flow rate in real time. To ensure the temporal consistency of collected data, the system incorporates a unified clock driver and channel grouping mechanism, ensuring that all sensor sampling operations are based on the same time base. Data synchronization scheduling logic ensures that the collected data is structurally time-consistent and channel-matched, providing a unified data entry point for subsequent data processing and status analysis.

[0019] In an optional implementation of this embodiment, the steps of synchronously collecting specified hardware parameters of different hardware units in the target industrial computer and obtaining original collected data include: collecting the specified hardware parameters based on multiple types of sensors deployed in different hardware units in the target industrial computer, and constructing an initial perception matrix of hardware parameters; generating a synchronously collected sample set by time series reorganization and channel classification of the initial perception matrix of hardware parameters; and performing field structured encoding on the synchronously collected sample set according to a preset format template to generate original collected data with field labels.

[0020] Specifically, in this embodiment, based on the deployment of multiple sensors, including voltage sensors, clock monitoring modules, error correction code counters, SMART parsing engines, and packet flow detectors, on various hardware units, such as the industrial computer motherboard, memory chips, hard disks, and network interfaces, and leveraging a unified clock source and synchronous trigger circuit, parallel acquisition of specified hardware parameters, such as voltage values, clock frequencies, bit error counts, hard disk health parameters, and network transceiver rates, is achieved. This results in the construction of an initial hardware parameter perception matrix, with rows representing sensor channels and columns representing acquisition time points. This matrix is ​​valuable in quantitatively characterizing the differences in the operating states of various hardware units. For example, it can simultaneously compare the correlation between motherboard voltage and memory error correction code error rates at the same moment. Subsequently, the initial perception matrix is ​​subjected to time series reorganization and channel classification. This involves synchronizing data streams of different sampling frequencies to a common time grid through interpolation or timestamp alignment, and grouping matrix rows by hardware unit category or sensor type, generating a set of synchronously acquired samples with a unified time base. This ensures comparability when subsequent algorithms compare parameters such as temperature fluctuations and network packet loss at the same moment. For example, the voltage data collected every second and the error count of the error correction code updated every two seconds are reorganized into the same time node, so that the health assessment model can process multi-dimensional features simultaneously. Then, the field structured encoding operation is performed on the synchronously collected sample set according to the predefined format template. During the encoding process, each sampling record is mapped to a key-value pair in the form of "hardware type-sensor number-parameter name-value-timestamp". For example, a record can be represented as "motherboard-V1-voltage-12.3V-2025-05-22T14:30:05Z", and the unique device identifier is added to generate the original collected data with field labels.

[0021] Step 120: Generate an input vector set for hardware health status assessment by preprocessing the original collected data.

[0022] Specifically, the input vector set for hardware health status assessment is generated by preprocessing the original collected data, using standardized modeling pre-processing technology. First, the characteristic fields that are representative of the hardware health status are screened out from the original data, and the outliers are removed, and the missing data are interpolated to ensure the integrity and validity of the data set. Then, normalization is performed according to the dimension and variation range of each parameter to unify the data scale and ensure that the numerical differences between different hardware components do not interfere with the judgment logic of the assessment model. Finally, the processed data is batch reorganized according to the device type and acquisition time interval, and converted into a structured multi-dimensional input vector set to meet the format and timing requirements of subsequent model calculations.

[0023] In an optional implementation of this embodiment, the step of generating an input vector set for hardware health status assessment by preprocessing the original collected data includes: obtaining a set of health indicator parameters corresponding to each hardware unit by performing field screening and numerical analysis on the original collected data; performing data cleaning on the health indicator parameter set based on preprocessing rules to construct an intermediate parameter vector sequence that meets the model input requirements; generating an initial input vector set by performing data standardization on the intermediate parameter vector sequence; and batch reconstructing the initial input vector set according to a predefined time window and hardware type label to generate an input vector set for hardware health status assessment.

[0024] Specifically, in this embodiment, during the field screening and numerical parsing stage of the original collected data, it is necessary to extract the corresponding fields from the original records containing the output of multiple sensors based on the hardware unit type and health assessment indicators. For example, the motherboard voltage sensor output field, the system clock frequency field, the memory error correction code count field, the hard disk SMART sub-attribute, and the network port data packet loss rate field are mapped to health indicator parameters, and their numerical types and units are parsed to form a set of health indicator parameters classified by hardware units. The set is stored in the form of a two-dimensional matrix, with each row corresponding to a hardware unit and each column corresponding to an indicator parameter. Through this process, monitoring variables that are not related to health assessment can be targeted and eliminated, and it is ensured that subsequent algorithms only process core health dimensions. When performing data cleaning based on preprocessing rules, missing values, outliers, and noise in the health indicator parameter set need to be uniformly processed. For missing values, linear interpolation based on adjacent time points or interpolation based on historical means can be used to restore data continuity. For outliers that clearly exceed the physical range, refer to the maximum and minimum operating ranges defined in the device manual and mark and remove those values ​​to avoid misleading model decisions. Furthermore, to reduce interference caused by high-frequency jitter, a sliding average or exponential smoothing algorithm can be applied to the parameter sequence to remove short-term peak noise, thereby constructing an intermediate parameter vector sequence that meets the input requirements of the deep learning model. This sequence preserves the temporal evolution trend of the health indicator while eliminating the impact of transient fluctuations on subsequent calculations. Data normalization of the intermediate parameter vector sequence requires converting the numerical scale of each indicator parameter to a uniform interval or distribution. Using the zero-mean unit variance method, each parameter column is subtracted from its historical mean and then divided by the standard deviation to generate a vector that conforms to a standard normal distribution. Alternatively, min-max normalization can be used to map parameter values ​​to the range of 0 to 1 to ensure a consistent numerical range and avoid training bias caused by dimensional differences. For example, mapping voltage values ​​from 0 to 24 volts to the range of 0 to 1, and mapping error counts to the range of the mean plus or minus three standard deviations, not only accelerates the convergence of gradient descent in the neural network but also ensures balanced weight distribution across dimensions in the loss function. During the batch reconstruction phase, the initial set of input vectors needs to be serialized and packaged based on a predefined time window (such as the past 60 seconds or 100 sampling periods) and hardware type labels. Using a sliding window approach, the normalized vectors of adjacent time points are concatenated into fixed-length time series blocks, and a hardware type label is added to each block to form a three-dimensional tensor input structure. This process not only preserves the health evolution information of each hardware unit within a specified time range, but also provides distinguishable component feature identifiers for the deep time series model, enabling subsequent health assessment models to simultaneously learn timing correlations and hardware category differences.

[0025] Step 130: Input the input vector set into a health assessment model to determine a multi-dimensional health index corresponding to each hardware unit.

[0026] Specifically, the input vector set is input into the health assessment model to determine the multi-dimensional health indicators corresponding to each hardware unit, using deep feature modeling and label-guided multi-output prediction technology. The system constructs a multi-branch neural network structure based on the hardware type. Each branch extracts features based on the timing characteristics of specific hardware, and uses the attention mechanism to build dynamic weights between channels to highlight the influence of abnormally sensitive dimensions. After feature aggregation, the fused vector is input into the multi-head output layer to calculate multi-dimensional health indicators such as the stability score, failure trend probability, and operating life estimation of the equipment. This modeling method not only improves the ability to identify different failure modes, but also provides reliable data support for the quantitative expression of system status.

[0027] In an optional implementation of this embodiment, the input vector set is input into the health assessment model to determine the multidimensional health index corresponding to each hardware unit, including: constructing a multi-branch deep feature extraction network according to the hardware type label of the input vector set; performing parallel feature extraction on the input vector set through each branch network of the multi-branch deep feature extraction network to generate a feature vector set containing time-series correlation features; constructing a multi-head attention weight allocation mechanism based on the channel dimension of the feature vector set, and generating a dynamic weight coefficient matrix through cross-channel feature correlation calculation; performing attention weighted calculation on the feature vector set according to the dynamic weight coefficient matrix to generate a weighted feature matrix; inputting the weighted feature matrix into a multi-source fusion network according to a predefined health dimension mapping rule to generate a multidimensional health index sequence; and determining the final multidimensional health index corresponding to each hardware unit through iterative compensation of the multidimensional health index sequence and the health correction function.

[0028] Specifically, the health correction function is constructed based on the physical coupling relationships between hardware units. In this embodiment, to achieve in-depth modeling and analysis of the health status of different hardware units, a multi-branch deep feature extraction network is first constructed based on the hardware type labels contained in the input vector set. This network adopts a distributed architecture, with each branch network extracting features for a specific type of hardware unit. For example, the motherboard, memory, hard drive, and communication interface each correspond to an independent neural network substructure. Through a hardware label encoding mechanism, the type information in the input vector is mapped into a static selection signal, guiding the automatic allocation of the input data stream to the corresponding subnetwork channel. Each subnetwork adopts a different layer and structure configuration to adapt to the temporal feature complexity of the corresponding hardware. For example, the motherboard channel uses a convolutional recurrent hybrid network (CRNN) with residual connections, while the hard drive channel uses a layer-normalized long short-term memory network (LayerNorm LSTM) to enhance the memory capacity of the SMART indicator sequence. Each branch network receives the normalized input tensor in the form of a time series and generates a set of deep feature vectors containing short-term and long-term dependencies during the feature encoding stage. Then, a multi-head attention weight distribution mechanism is constructed based on the distribution of feature vector sets in the channel dimension, and the importance of each dimension is dynamically adjusted by cross-channel feature correlation modeling. For this purpose, the inter-channel collaborative attention tensor is introduced. , where C represents the number of channels and the element Represents the temporal feature similarity between channel i and channel j, through the dot product attention mechanism and combined with the trainable query vector q, key vector k and scaling factor , the attention weight calculation formula is as follows: , in, are the attention query and key vector of channel i and channel j respectively, Represents the transposed form of the key vector, used to compare with the query vector Do the dot product, Indicates the degree of matching between channel i and channel j features. The larger the value, the more similar it is. The weighted feature matrix is ​​a vector dimension. This mechanism enables the system to automatically evaluate the semantic similarity between channels and dynamically weight them at output. By performing a tensor product operation on the feature vector set and the weight matrix, a weighted feature matrix is ​​generated, which strengthens the influence of key dimensions, weakens redundant or noise dimensions, and improves the discrimination accuracy of downstream tasks. Subsequently, according to the predefined health dimension mapping rules, the weighted feature matrix is ​​input into the multi-source fusion network. The network uses a multi-layer nonlinear transformation unit and a residual connection structure to fuse deep features from different channels and different time periods to generate a multi-dimensional health indicator sequence. Each dimension in the sequence corresponds to an independent health assessment sub-indicator, including stability index, failure probability, degradation trend level and predicted remaining life. Each indicator is output in the form of a continuous numerical value to form a complete hardware health description vector.

[0029] To further enhance the overall consistency and logical integrity of health assessments, the system introduces a health correction function to perform global iterative compensation on the initially generated health indicator sequence. This function models the physical coupling relationship between hardware units and their linkage effects on health indicators, and introduces structural compensation terms in multiple indicator dimensions. The formula is as follows: , in, represents the modified health indicator vector of hardware unit i, is the original health index of hardware unit i before correction, It is represented as the original health index vector of hardware unit j before correction, Represents a set of units that have physical connections or logical dependencies with hardware unit i. [0,1] represents the influence weight of hardware unit j on unit i, Indicates the sampling delay difference between the two units, 、 is a global adjustment factor, which is a highly nonlinear transformation h Characterize the difference in health status and perform dynamic coupling compensation considering physical topology and time correlation, so that the health assessment results are more consistent with the actual laws of collaborative degradation and abnormal propagation between devices. The final output is The vector is used as the final multi-dimensional health indicator of each hardware unit, supporting the decision-making logic for anomaly identification and remote control.

[0030] Step 140: Compare the multi-dimensional health index with the preset health threshold to generate a hardware abnormality status identification result. Specifically, the multi-dimensional health indicators are compared with the preset health thresholds to generate hardware abnormal state identification results, relying on an abnormality identification method based on rule matching and dynamic correction mechanisms. By setting the safety threshold range corresponding to each health dimension in the system, the health indicators output by the model are compared item by item to preliminarily determine whether there is an out-of-limit state. In order to address the impact of misjudgment and cross-component coupling, an abnormal vector correction strategy is also introduced to perform linkage corrections on edge abnormal results based on the physical dependency relationship between hardware. Finally, a decision tree and historical sample data are used for multi-path backtracking verification, and abnormal state labels and risk level scores are output to achieve refined identification of abnormal states.

[0031] In an optional implementation of this embodiment, the step of comparing the multidimensional health indicator with a preset health threshold to generate a hardware abnormal state identification result includes: performing step-by-step matching according to each dimensional parameter of the multidimensional health indicator and the preset health threshold to generate an initial abnormal vector set; performing cross-unit association correction on the initial abnormal vector set to generate a corrected abnormal vector set; performing multi-path reasoning on the corrected abnormal vector set according to a preset abnormal decision tree to generate an abnormal state intermediate result; performing backtracking verification according to the time series characteristics of the abnormal state intermediate result and a historical fault database to output a hardware abnormal state identification result.

[0032] Specifically, to accurately identify hardware failure states, the system first needs to match each dimension parameter in the multidimensional health indicator output by the front-end assessment model with the preset health thresholds. Multidimensional health indicators typically include failure probability, performance stability score, degradation trend index, and predicted lifespan. Each dimension corresponds to one or more statically or dynamically set health thresholds. During the matching process, the system performs interval judgment or critical point comparison on the indicator value of each dimension to determine whether there is a potential abnormal state in that dimension, and records the matching result as a Boolean value or risk score. This process is performed independently on each hardware unit. All matching results are concatenated to form an initial anomaly vector set, where each row corresponds to a hardware unit and each column corresponds to the abnormality flag or limit violation level of a health dimension at the current moment. For example, if the error correction code error rate of a memory module exceeds the specified threshold and the degradation trend is in the rapidly increasing range, the abnormal bit of the module in the "stability" and "trend risk" dimensions is marked as "1" or assigned a risk score, such as "0.87", forming an initial anomaly description with quantitative characteristics.

[0033] Subsequently, to avoid misjudgments caused by single-point dimension fluctuations and to more accurately reflect the state coupling relationships between devices, the system performs cross-unit correlation corrections on the initial anomaly vector set. This operation is based on a physical or logical connection graph between hardware, in which edges represent functional dependencies or energy exchange paths between devices. By calculating the propagation consistency of the current anomaly vector along the graph structure, for example, using the message passing mechanism in a graph neural network, high-confidence anomaly vectors are propagated to relevant device nodes via connecting edges. The initial anomaly judgment is then adjusted based on the status scores of upstream and downstream devices. For example, if the motherboard experiences a temperature anomaly while the memory module only experiences marginal fluctuations, yet both operate on the same physical carrier board, the system will increase the risk score of the memory module, indicating that it may be in a potentially abnormal state due to a motherboard issue. This process generates a revised anomaly vector set that retains the initial identification results while incorporating the structural dependencies and dynamic state propagation characteristics between devices. After obtaining the revised anomaly vector set, the system performs multi-path reasoning on it based on a pre-set anomaly decision tree. An anomaly decision tree is a classification inference structure constructed by combining rule-based logic and statistical learning results. Its nodes represent judgment conditions, and leaf nodes correspond to the output intermediate anomaly state labels. This reasoning process supports the simultaneous execution of multiple paths, allowing for the concurrent evaluation of multiple abnormal patterns, covering varying severity levels, from mild degradation to partial failures and critical faults. For example, if a hard drive module experiences a three-cycle decline in "performance stability" and falls below the lower limit of "predicted lifespan," while simultaneously experiencing a CPU module temperature anomaly and triggering an alarm input on the motherboard, the system cross-compares the multiple conditions met in paths A and B, outputting a "compound failure possible" status label with an accompanying confidence score, thereby constructing intermediate abnormality status results tailored to various risk patterns.

[0034] Finally, to ensure the accuracy and credibility of the output status, the system performs time series backtracking verification on the intermediate abnormal state results against the historical fault database. During this backtracking process, the multidimensional health indicator sequence corresponding to the current abnormal state is first aligned with the sequence template of real fault samples in the historical database. The similarity between the sequences is measured using a sliding window dynamic time warping algorithm to determine whether the current state exhibits an evolutionary trend that is highly consistent with past fault samples. If the match exceeds a preset similarity threshold, the system marks the state as a high-confidence anomaly and outputs a complete hardware abnormal state identification result, which includes the hardware unit identifier, fault type label, timestamp, and confidence score. For example, if the error growth trajectory of the current memory module's error correction code matches a pattern of multiple historical outages, and the dynamic time warping distance is less than a threshold, the system identifies the module as being in a "high-risk degradation" state, which serves as the basis for an early warning response. Through this mechanism, the system's final identification output not only has a logical basis for current data judgment but is also validated by historical empirical data, ensuring structural completeness and temporal consistency in the fault judgment results.

[0035] Step 150: Send a remote control instruction to the target industrial computer for remote control according to the abnormal state identification result.

[0036] Specifically, remote control is achieved by sending remote control commands to the target industrial computer based on warning events generated by abnormal state identification. This system utilizes an encrypted command generation and authenticated communication mechanism. Based on the abnormal state type and impact range, the system matches the corresponding control policy template and generates remote control commands to adjust monitoring parameters or execute control actions. Commands are encrypted before issuance, and identity verification is completed through the establishment of an end-to-end secure authentication channel to ensure that the command transmission process is not tampered with or intercepted. After the control command is decrypted and integrity verified on the target device, it automatically triggers operations such as threshold modification, fault shielding, or state reset, establishing a closed-loop control logic from abnormality identification to parameter intervention.

[0037] In an optional implementation of this embodiment, the step of sending remote control instructions to the target industrial computer for remote control based on the abnormal state identification result includes: generating a control instruction set based on the confidence rating and fault location code in the hardware abnormal state identification result; performing layered data encapsulation on the control instruction set through an asymmetric encryption algorithm based on a preset security policy library to generate an encrypted instruction data packet; constructing a dual-channel redundant transmission link based on the communication protocol characteristics of the target industrial computer, fragmenting and reorganizing the encrypted instruction data packet and sending it to the target industrial computer synchronously; and generating corresponding monitoring parameter update feedback results by dynamically reloading the fragmented and reorganized encrypted instruction data packet.

[0038] Specifically, to implement remote response control for abnormal conditions, a control instruction set must first be generated based on the confidence rating and fault location code from the previously identified hardware abnormal condition. The confidence rating refers to the system's assessment of the reliability of a specific abnormality determination, typically expressed as a continuous value between 0 and 1. The fault location code identifies the specific affected hardware unit and its location. For example, "MB01_CPU_TEMP_HIGH" indicates an overtemperature event for CPU 01 on motherboard 01. Based on these two pieces of information, the control instruction set generation process invokes remote handling policies matching the corresponding abnormal condition from a policy template library. These policies include threshold adjustment, module restart, interface blocking, and instruction injection, and a hierarchical response logic is constructed based on the current risk level. For example, if the confidence rating is higher than 0.9 and the location code identifies a critical unit, the system generates high-priority mandatory control instructions. For non-critical modules with medium confidence, only parameter fine-tuning instructions are generated. The final output control instruction set contains multiple fields, such as instruction type, scope, parameter modification value, effective time, and checksum, and has a clearly structured format. After the control instruction set is constructed, to ensure security and tamper resistance during transmission, the system performs layered data encapsulation on it using an asymmetric encryption algorithm based on a pre-set security policy library, generating an encrypted instruction data packet. Asymmetric encryption utilizes a public and private key pair. The control center encrypts the instruction content using the target IPC's public key, which can only be decrypted and read by the target IPC, which possesses the corresponding private key. This layered data encapsulation mechanism encodes the control instruction's main content, instruction summary, timestamp, and source signature into separate data layers. Each layer is accompanied by a hash checksum to ensure integrity. This encapsulated instruction data packet is not only resistant to decomposition by intermediate nodes but also allows the sender's identity and content integrity to be verified at the receiving end, effectively defending against man-in-the-middle and data replay attacks. To ensure highly reliable instruction delivery and mitigate data link instability caused by complex field environments, the system establishes a dual-channel redundant transmission link based on the communication protocol supported by the target IPC. The encrypted instruction data packet is fragmented and reassembled before being synchronously transmitted to the target IPC. The dual-channel transmission link includes a primary link and a backup link, built based on TCP and UDP or 4G and Ethernet respectively, to ensure seamless transmission via the backup link in the event of a primary link interruption. The data fragmentation mechanism splits the encrypted data packet according to the maximum transmission unit (MTU), attaching a sequence number and fragmentation check information to each fragment. During reassembly, the complete data packet is restored based on the sequence number order, ensuring that the instruction content is in order and intact. This transmission design not only enhances the system's robustness to network fluctuations, but also allows multiple instruction packets to be delivered simultaneously on different links, thereby reducing time delays caused by link congestion.After the target industrial computer completes receiving and reassembling the encrypted instruction packet, the device will call the local decryption module to perform data decryption and integrity verification, and perform dynamic parameter threshold reloading based on the instruction field parsing results. Dynamic threshold reloading refers to the real-time update of the currently running monitoring module based on the new threshold parameters contained in the control instruction, without restarting the system or interrupting the task. For example, if the upper limit of the motherboard temperature alarm is changed to 75°C, the system automatically replaces the original threshold and takes effect immediately. At the same time, the update event is recorded in the log module and the monitoring parameter update feedback result containing "overloaded field", "update status", "effective time" and other content is returned. This feedback result will be encapsulated as a receipt data packet and returned to the control center for closed-loop confirmation and status synchronization, ensuring that the remote instruction is not only securely issued, but also effectively executed and tracked and verified, thereby building a closed loop of instruction response for abnormality identification.

[0039] In an optional implementation of this embodiment, a three-dimensional device connection relationship matrix is ​​constructed based on the physical connection topology diagram of the target industrial computer and the communication interface configuration table; three-dimensional topology rendering data is generated by spatially weighting the three-dimensional device connection relationship matrix and the multi-dimensional health index; a communication status waveform data set is constructed based on the real-time collected communication interface electrical parameter fluctuation data; and a dynamic visual monitoring interface is generated by performing multi-layer overlay processing on the three-dimensional topology rendering data, the health status code, and the communication status waveform data set.

[0040] Specifically, to build a monitoring interface that supports dynamic visualization analysis, a three-dimensional device connection matrix must first be constructed based on the target industrial computer's physical connection topology and communication interface configuration table. The physical connection topology is a structural diagram that describes the actual physical connection relationships between devices, representing the connection paths between the industrial computer and its connected sensors, actuators, host computers, or other peripheral devices. The communication interface configuration table records the number and type of each communication port (such as RS232, RS485, Ethernet, USB, etc.), the corresponding device port number, communication protocol, and operating parameters. By mapping the nodes and edges of the topology graph into physical objects in a three-dimensional coordinate system and numbering the input and output channels of each node in conjunction with the interface configuration table, a three-dimensional matrix C(i, j, k) is constructed, where i represents the starting device number, j represents the target device number, and k represents the channel or interface type code. This matrix describes the connection relationship between any two devices in the system under a specific interface channel. This matrix is ​​the foundation for the subsequent rendering engine to recognize the spatial structure and ensures a one-to-one correspondence between the rendered model and the device layout. After constructing the connection matrix, it needs to be spatially weighted and fused with multidimensional health indicators to generate 3D topological rendering data. Multidimensional health indicators are vector sets that assess the operating status of each device unit, typically including dimensions such as stability score, failure probability, failure trend, and lifespan estimate. To integrate these with the 3D topological matrix, visual attributes such as color, transparency, or deformation are applied to the geometric representation of each device node in 3D space, and the representation is dynamically adjusted based on the health indicator. For example, nodes with a failure probability greater than 0.8 are rendered as red translucent spheres in the model, while links with rapid degradation trends are bolded and set to flash. By embedding device health indicators into the node and edge attributes of the connection matrix, a 3D topological rendering dataset is constructed, forming an information model that combines structural logic and operating status. This allows the spatial layout to express anomaly clusters and risk transmission paths. Simultaneously with the generation of the spatial rendering data, a communication status waveform dataset is also constructed based on real-time data collected from the electrical parameter fluctuations of the communication interfaces. This dataset originates from the underlying electrical sampling module of the interface layer. It records physical quantities such as voltage, current, frequency, impedance, and signal integrity in real time and encodes them in time series at the sampling frequency, generating two-dimensional waveform curve data. Fluctuations in the status of different communication interfaces can reveal the stability or level of interference in the communication link. For example, periodic voltage drops on an RS485 bus may indicate grounding anomalies or signal interference. Each interface sampling sequence is mapped into a standard time-series data frame. Each frame contains fields such as timestamp, sample value, fluctuation amplitude, and phase offset, thereby constructing a communication status waveform dataset that supports comparative analysis and dynamic rendering.Finally, to implement a multi-dimensional dynamic visualization system for end users, the generated 3D topology rendering data, health status code information, and communication status waveform datasets are layered together to create a unified dynamic visualization monitoring interface. This multi-layer overlay technology integrates data from different sources through a layered rendering pipeline within the graphics engine. The bottom layer represents the 3D structure layout, the middle layer represents the health status indicators for nodes and edges, and the top layer represents real-time interface waveform animations. Transparency fusion, event linkage, and time synchronization are used to achieve unified responsiveness between these layers, allowing users to clearly visualize device connectivity, health status trends, and abnormal communication fluctuations. For example, when a network port flashes red in the 3D model while a continuous high-frequency interference waveform appears in the layer above, users can identify the current problem area and the scope of the fault, significantly improving the usability and response efficiency of the monitoring system. The resulting monitoring interface is highly interactive, integrated, and real-time, meeting the needs for rapid judgment and decision-making on complex system status in industrial scenarios.

[0041] According to a remote monitoring method for the hardware status of an industrial computer provided by the present application, the specified hardware parameters of different hardware units in the target industrial computer are synchronously collected to obtain the original collected data; the input vector set for hardware health status evaluation is generated by preprocessing the original collected data; the input vector set is input into a health evaluation model to determine the multidimensional health index corresponding to each hardware unit; the multidimensional health index is compared with a preset health threshold to generate a hardware abnormal state identification result; and a remote control instruction is sent to the target industrial computer for remote control based on the abnormal state identification result. Through the implementation of the present application, the key parameters of different hardware units are collected to form unified structured data, and the health evaluation model is used to generate multidimensional health indicators, which ultimately drives the dynamic update of the remote control strategy, which can effectively improve the operation and maintenance efficiency of the industrial computer.

[0042] Figure 2 The embodiment of the present application provides a remote monitoring device for the hardware status of an industrial computer. The remote monitoring device for the hardware status of an industrial computer can be used to implement the remote monitoring method for the hardware status of an industrial computer in the aforementioned embodiment. Figure 2 As shown, the remote monitoring device for the hardware status of the industrial computer mainly includes: The acquisition module 10 is used to synchronously acquire the specified hardware parameters of different hardware units in the target industrial computer to obtain the original acquisition data; A preprocessing module 20 is used to generate an input vector set for hardware health status assessment by preprocessing the original collected data; a determination module 30 for inputting the input vector set into the health assessment model to determine the multi-dimensional health index corresponding to each hardware unit; Comparison module 40, used to compare the multi-dimensional health indicators with the preset health thresholds to generate hardware abnormality status identification results; The control module 50 is used to send remote control instructions to the target industrial computer for remote control according to the abnormal state recognition result.

[0043] In an optional implementation of this embodiment, the acquisition module is specifically used to: collect specified hardware parameters based on multiple types of sensors deployed in different hardware units in the target industrial computer, and construct an initial perception matrix of hardware parameters; generate a synchronous acquisition sample set by performing time series reorganization and channel classification on the initial perception matrix of hardware parameters; perform field structured encoding on the synchronous acquisition sample set according to a preset format template to generate original acquisition data with field labels.

[0044] In an optional implementation of this embodiment, the preprocessing module is specifically used to: obtain a set of health indicator parameters corresponding to each hardware unit by performing field screening and numerical analysis on the original collected data; perform data cleaning on the health indicator parameter set based on preprocessing rules to construct an intermediate parameter vector sequence that meets the model input requirements; generate an initial input vector set by performing data standardization on the intermediate parameter vector sequence; and batch reconstruct the initial input vector set according to a predefined time window and hardware type label to generate an input vector set for hardware health status assessment.

[0045] In an optional implementation of this embodiment, the determination module is specifically used to: construct a multi-branch deep feature extraction network based on the hardware type label of the input vector set; perform parallel feature extraction on the input vector set through each branch network of the multi-branch deep feature extraction network to generate a feature vector set containing time-series correlation features; construct a multi-head attention weight allocation mechanism based on the channel dimension of the feature vector set, and generate a dynamic weight coefficient matrix through cross-channel feature correlation calculation; perform attention weighted calculation on the feature vector set according to the dynamic weight coefficient matrix to generate a weighted feature matrix; input the weighted feature matrix into the multi-source fusion network according to the predefined health dimension mapping rule to generate a multi-dimensional health indicator sequence; perform iterative compensation through the multi-dimensional health indicator sequence and the health correction function to determine the final multi-dimensional health indicator corresponding to each hardware unit; wherein, the health correction function is a function constructed based on the physical coupling relationship between hardware units.

[0046] In an optional implementation of this embodiment, the comparison module is used to: perform step-by-step matching based on the various dimensional parameters of the multidimensional health indicator and the preset health threshold to generate an initial abnormality vector set; perform cross-unit correlation correction on the initial abnormality vector set to generate a corrected abnormality vector set; perform multi-path reasoning on the corrected abnormality vector set according to a preset abnormality decision tree to generate an abnormal state intermediate result; perform backtracking verification based on the time series characteristics of the abnormal state intermediate result and the historical fault database to output the hardware abnormal state identification result.

[0047] In an optional implementation of this embodiment, the control module is specifically used to: generate a control instruction set based on the confidence rating and fault location code in the hardware abnormal state identification result; perform layered data encapsulation on the control instruction set through an asymmetric encryption algorithm based on a preset security policy library to generate an encrypted instruction data packet; construct a dual-channel redundant transmission link based on the communication protocol characteristics of the target industrial computer, fragment and reorganize the encrypted instruction data packet, and send it to the target industrial computer synchronously; generate corresponding monitoring parameter update feedback results by dynamically reloading the fragmented and reorganized encrypted instruction data packet.

[0048] In an optional implementation of this embodiment, the control module is also used to: construct a three-dimensional device connection relationship matrix based on the physical connection topology diagram of the target industrial computer and the communication interface configuration table; generate three-dimensional topology rendering data by spatially weighting the three-dimensional device connection relationship matrix and the multi-dimensional health index; construct a communication status waveform data set based on the real-time collected communication interface electrical parameter fluctuation data; and generate a dynamic visual monitoring interface by performing multi-layer overlay processing on the three-dimensional topology rendering data, health status code and communication status waveform data set.

[0049] According to the remote monitoring device for the hardware status of an industrial computer provided by the present application, the specified hardware parameters of different hardware units in the target industrial computer are synchronously collected to obtain the original collected data; the input vector set for hardware health status evaluation is generated by preprocessing the original collected data; the input vector set is input into the health evaluation model to determine the multidimensional health index corresponding to each hardware unit; the multidimensional health index is compared with the preset health threshold to generate the hardware abnormal state identification result; and a remote control instruction is sent to the target industrial computer for remote control based on the abnormal state identification result. Through the implementation of the present application, the key parameters of different hardware units are collected to form unified structured data, and the health evaluation model is used to generate multidimensional health indicators, which ultimately drives the dynamic update of the remote control strategy, which can effectively improve the operation and maintenance efficiency of the industrial computer.

[0050] According to the application plan Figure 3An electronic device provided in an embodiment of the present application. This electronic device can be used to implement the method for remotely monitoring the hardware status of an industrial computer in the aforementioned embodiment, and mainly includes: Memory 301, processor 302, and computer program 303 stored in memory 301 and executable on processor 302. Memory 301 and processor 302 are connected via communication. When processor 302 executes computer program 303, the method for remotely monitoring the hardware status of an industrial computer described in the aforementioned embodiment is implemented. The number of processors can be one or more.

[0051] The memory 301 can be a high-speed random access memory (RAM) memory or a non-volatile memory such as a disk drive. The memory 301 is used to store executable program code. The processor 302 is coupled to the memory 301 .

[0052] Furthermore, the embodiment of the present application also provides a computer-readable storage medium, which can be provided in the electronic device in the above embodiments. The computer-readable storage medium can be the above Figure 3 Memory in the illustrated embodiment.

[0053] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for remotely monitoring the hardware status of an industrial computer in the aforementioned embodiment. Furthermore, the computer-readable storage medium may be any medium capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), RAM, a magnetic disk, or an optical disk.

[0054] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0055] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0056] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A remote monitoring method for the hardware status of an industrial computer, characterized in that: include: Synchronously collect the specified hardware parameters of different hardware units in the target industrial computer to obtain the original collected data; Generating an input vector set for hardware health status assessment by preprocessing the original collected data; Inputting the input vector set into a health assessment model to determine a multi-dimensional health index corresponding to each hardware unit; Comparing the multi-dimensional health index with a preset health threshold to generate a hardware abnormality status identification result; Send a remote control instruction to the target industrial computer for remote control according to the abnormal state identification result.

2. The method for remotely monitoring the hardware status of an industrial computer according to claim 1, wherein: The step of synchronously collecting the specified hardware parameters of different hardware units in the target industrial computer to obtain original collected data includes: Collect specified hardware parameters based on multiple types of sensors deployed on different hardware units in the target industrial computer and build an initial perception matrix of hardware parameters; Generate a synchronous acquisition sample set by performing time series reorganization and channel classification on the initial perception matrix of the hardware parameters; The synchronously collected sample set is subjected to field structured encoding according to a preset format template to generate original collected data with field labels.

3. The method for remotely monitoring the hardware status of an industrial computer according to claim 1, wherein: The step of generating an input vector set for hardware health status assessment by preprocessing the original collected data includes: Obtaining a set of health indicator parameters corresponding to each hardware unit by performing field screening and numerical analysis on the original collected data; Performing data cleaning on the health indicator parameter set based on preprocessing rules to construct an intermediate parameter vector sequence that meets the model input requirements; Generate an initial input vector set by performing data normalization processing on the intermediate parameter vector sequence; The initial input vector set is batch reconstructed according to a predefined time window and hardware type labels to generate an input vector set for hardware health status assessment.

4. The method for remotely monitoring the hardware status of an industrial computer according to claim 3, wherein: The step of inputting the input vector set into the health assessment model to determine the multi-dimensional health index corresponding to each hardware unit includes: Constructing a multi-branch deep feature extraction network according to the hardware type labels of the input vector set; Performing parallel feature extraction on the input vector set through each branch network of the multi-branch deep feature extraction network to generate a feature vector set containing time series correlation features; A multi-head attention weight allocation mechanism is constructed based on the channel dimension of the feature vector set, and a dynamic weight coefficient matrix is ​​generated by calculating the cross-channel feature correlation; Performing attention weighted calculation on the feature vector set according to the dynamic weight coefficient matrix to generate a weighted feature matrix; Inputting the weighted feature matrix into a multi-source fusion network according to a predefined health dimension mapping rule to generate a multi-dimensional health indicator sequence; The final multidimensional health index corresponding to each hardware unit is determined by iteratively compensating the multidimensional health index sequence and the health correction function; wherein the health correction function is a function constructed based on the physical coupling relationship between the hardware units.

5. The method for remote monitoring of the hardware status of an industrial computer according to claim 1, wherein: The step of comparing the multi-dimensional health indicator with a preset health threshold to generate a hardware abnormality status identification result includes: Generate an initial abnormal vector set based on the level-by-level matching of the various dimensional parameters of the multi-dimensional health indicator with the preset health threshold; Performing cross-unit correlation correction on the initial abnormal vector set to generate a corrected abnormal vector set; Perform multi-path reasoning on the modified abnormal vector set according to a preset abnormal decision tree to generate an abnormal state intermediate result; Backtrack verification is performed based on the time series characteristics of the abnormal state intermediate result and the historical fault database, and the hardware abnormal state identification result is output.

6. The method for remotely monitoring the hardware status of an industrial computer according to claim 5, characterized in that: The step of sending a remote control instruction to the target industrial computer for remote control according to the abnormal state identification result includes: generating a control instruction set according to the confidence rating and the fault location code in the hardware abnormal state identification result; Performing layered data encapsulation on the control instruction set by an asymmetric encryption algorithm based on a preset security policy library to generate an encrypted instruction data packet; Building a dual-channel redundant transmission link based on the communication protocol characteristics of the target industrial computer, fragmenting and reassembling the encrypted instruction data packet and then synchronously sending it to the target industrial computer; By dynamically reloading the parameter threshold of the encrypted instruction data packet after fragmentation and reorganization, the corresponding monitoring parameter update feedback result is generated.

7. The method for remotely monitoring the hardware status of an industrial computer according to claim 1, wherein: The method further comprises: Constructing a three-dimensional device connection relationship matrix based on the physical connection topology diagram and communication interface configuration table of the target industrial computer; Generate three-dimensional topological rendering data by performing spatial weight fusion on the three-dimensional device connection relationship matrix and the multi-dimensional health index; Construct a communication status waveform dataset based on the real-time collected communication interface electrical parameter fluctuation data; A dynamic visual monitoring interface is generated by performing multi-layer overlay processing on the three-dimensional topology rendering data, the health status code, and the communication status waveform data set.

8. A remote monitoring device for the hardware status of an industrial computer, characterized in that: The remote monitoring device for the hardware status of an industrial computer comprises: The acquisition module is used to synchronously acquire the specified hardware parameters of different hardware units in the target industrial computer and obtain the original acquisition data; A preprocessing module, configured to generate an input vector set for hardware health status assessment by preprocessing the original collected data; a determination module, configured to input the input vector set into a health assessment model to determine a multidimensional health index corresponding to each hardware unit; A comparison module, configured to compare the multi-dimensional health indicator with a preset health threshold to generate a hardware abnormality status identification result; The control module is used to send a remote control instruction to the target industrial computer for remote control according to the abnormal state recognition result.

9. An electronic device, characterized in that: Comprising a memory and a processor, wherein: The processor is configured to execute a computer program stored in the memory; When the processor executes the computer program, the steps of the method for remotely monitoring the hardware status of an industrial computer as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for remotely monitoring the hardware status of an industrial computer as claimed in any one of claims 1 to 7 are implemented.

Images