A method, system, device, medium and product for preventing misoperation of a direct current system

Abstract

The application discloses a kind of direct current system anti-misoperation discrimination method, system, equipment, medium and product, it is related to direct current system monitoring technical field, obtains the operation identification data and electrical operation data of direct current system, based on the operation intention library of preestablished operation identification data is intended to parse, obtain intention credibility and action mark, electrical characteristic extraction is carried out to electrical operation data, obtain electrical feature vector and electrical mark, according to intention credibility and electrical feature vector carries out causal coupling processing, obtains causal coupling degree and coupling mark, according to causal coupling degree, action mark, electrical mark and coupling mark carries out operation risk detection, obtains corresponding anti-misoperation discrimination result.The existing direct current system anti-misoperation technology is only around the single dimension step verification of operation behavior compliance, which cannot accurately identify the explicit human misoperation such as skip, mis-touch and operation object error, and reduces the reliability of the direct current system operation.

Description

Technical Field

[0001] This invention relates to the field of DC system monitoring technology, and in particular to a method, system, device, medium and product for identifying DC system misoperation. Background Technology

[0002] As the core hub of the power grid, the DC system of a substation undertakes the task of uninterrupted power supply to critical loads such as relay protection, automatic devices, control circuits, and emergency lighting. It is the core basic power source to ensure the reliable operation of the secondary system and prevent the escalation of power grid accidents. With the continuous expansion of the power grid and the increasing automation and unmanned operation of substations, switching operations such as switching of DC system operation modes, battery bank and charger commissioning and decommissioning, and bus switching are becoming increasingly frequent. The operation process is complex and has many risk points. Once a misoperation occurs, it can easily lead to serious events such as DC bus undervoltage, protection failure to operate, and circuit breaker tripping, directly threatening the safe and stable operation of the power grid, and even causing regional power outages and significant economic losses. Current power safety management and control place higher demands on the whole process, real-time, and intelligent prevention of errors in DC operation. The traditional mode of relying on manual verification, mechanical interlocking, and post-event alarm is no longer suitable for the high reliability and high safety operation requirements of modern power grids.

[0003] Currently, existing DC system anti-misoperation technologies mostly focus on single-dimensional step verification of operational compliance, or only achieve post-event anomaly monitoring through electrical quantity threshold exceeding limits. They cannot accurately identify explicit human misoperations such as skipping steps, accidental contact, and incorrect operation objects, thus reducing the reliability of DC system operation. Summary of the Invention

[0004] This invention provides a method, system, device, medium, and product for identifying misoperation in DC systems. It solves the technical problem that existing DC system misoperation prevention technologies mostly focus on single-dimensional step verification of operational compliance or only achieve post-event anomaly monitoring through electrical quantity threshold exceedances. These technologies cannot accurately identify explicit human misoperations such as skipping steps, accidental contact, and incorrect operation objects, thus reducing the reliability of DC system operation.

[0005] The first aspect of this invention provides a method for identifying and preventing misoperation in a DC system, comprising:

[0006] Acquire operation identification data and electrical operation data of the DC system, and perform intent parsing on the operation identification data based on a preset operation intent library to obtain the corresponding intent credibility and action identifier;

[0007] Electrical features are extracted from the electrical operation data to obtain the corresponding electrical feature vectors and electrical identifiers;

[0008] Based on the intent credibility and the electrical feature vector, causal coupling processing is performed to obtain the corresponding causal coupling degree and coupling identifier;

[0009] Operational risk detection is performed based on the causal coupling degree, the action identifier, the electrical identifier, and the coupling identifier to obtain the corresponding anti-misoperation judgment result.

[0010] Optionally, the operation intent library includes an action template library and a timing rule library. The step of parsing the operation recognition data based on the preset operation intent library to obtain the corresponding intent credibility and action identifier includes:

[0011] Based on a pre-trained target detection model, target detection is performed sequentially on each frame of the operation image in the operation recognition data to obtain multiple action data.

[0012] Based on a preset action matching function, the corresponding action matching degree is determined according to the action template library and each action data;

[0013] Based on a preset temporal coherence function, the corresponding temporal coherence coefficient is determined according to the temporal rule base and each action data.

[0014] The action matching degree and the temporal coherence coefficient are weighted based on the preset intent weight to obtain the initial intent credibility, and the initial intent credibility is normalized to obtain the corresponding intent credibility.

[0015] Misoperation detection is performed based on the action matching degree, the temporal coherence coefficient, and the intent credibility to obtain the corresponding action identifier.

[0016] By employing the above technical solution, action data is extracted frame by frame from the operation image based on a pre-trained target detection model. This accurately captures operation details and ensures the accuracy of the original behavior data collection. The action matching degree and temporal coherence coefficient are calculated using an action template library and a temporal rule library, respectively, to verify the standardization of the operation from both action form and execution flow dimensions. Then, a standardized quantified intent credibility is obtained through intent weighting and normalization. Misoperation detection is performed based on the action matching degree, the temporal coherence coefficient, and the intent credibility to obtain the corresponding action identifier. This effectively reduces human judgment errors, accurately identifies human misoperations, and provides objective and reliable detection results. Furthermore, the model and standardized algorithms improve detection efficiency and versatility, allowing for stable adaptation to the operation behavior recognition needs of different scenarios.

[0017] Optionally, the step of detecting erroneous operations based on the action matching degree, the temporal coherence coefficient, and the intent credibility to obtain the corresponding action identifier includes:

[0018] The action matching degree, the temporal coherence coefficient, and the intent credibility are averaged to obtain the corresponding first average.

[0019] The difference between the preset action benchmark value and the first mean value is processed to obtain the corresponding action abnormality score;

[0020] When the abnormal action score is greater than or equal to the preset action threshold, the preset first action identifier score is determined as the corresponding action identifier.

[0021] When the abnormal action score is less than the action threshold, the preset second action identifier score is determined as the corresponding action identifier.

[0022] By adopting the above technical solution, the average values ​​of action matching degree, temporal coherence coefficient, and intent credibility are first calculated. Multi-dimensional evaluation indicators are then integrated to form a comprehensive evaluation metric. Next, an error score is obtained by combining this metric with the action baseline value, intuitively quantifying the overall deviation of the operational behavior. Based on action thresholds, interval judgments are made, and corresponding action identifier scores are matched according to rules to differentiate completion states. The entire calculation logic is simple and clear, with low computational overhead. The judgment criteria are unified and highly quantifiable, objectively distinguishing between normal and abnormal operations, effectively avoiding subjective judgment bias. Simultaneously, it facilitates program implementation, quickly outputting stable and reusable action identifiers, and improving the standardization and real-time performance of misoperation identification.

[0023] Optionally, the step of extracting electrical features from the electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers includes:

[0024] The electrical operating data is input into a preset electrical steady-state deviation function to obtain the corresponding electrical steady-state deviation;

[0025] The electrical response time and the corresponding action execution time in the electrical operation data are processed by difference to obtain the corresponding response delay;

[0026] The electrical operation data is input into a preset electrical fluctuation function to obtain the corresponding electrical fluctuation coefficient;

[0027] The electrical steady-state deviation, the response delay, and the electrical fluctuation coefficient are used as the corresponding electrical feature vectors;

[0028] The electrical steady-state deviation, the response delay, and the electrical fluctuation coefficient are weighted according to preset electrical weights to obtain the corresponding electrical anomaly score;

[0029] When the electrical anomaly score is greater than or equal to a preset electrical threshold, the preset first electrical identification score is determined as the corresponding electrical identification.

[0030] When the electrical anomaly score is less than the electrical threshold, the preset second electrical identifier score is determined as the corresponding electrical identifier.

[0031] By adopting the above technical solution, electrical operation data is comprehensively analyzed from three dimensions: steady-state deviation, response delay, and fluctuation coefficient. A complete electrical feature vector is constructed, which can fully cover three types of state information: static operating conditions of equipment, action response timing, and operational stability. The feature representation is comprehensive and without omission. By weighting and integrating various indicators with electrical weights, an electrical anomaly score is obtained, taking into account the judgment priority of different electrical features. The quantitative results are more in line with the on-site operation standards. Combined with electrical thresholds, a graded judgment is completed and an electrical identifier is output. This can accurately identify various anomalies such as electrical circuit faults, transmission lag, and operational fluctuations, and facilitate the deployment of automated programs, thereby improving the accuracy and real-time performance of electrical condition detection.

[0032] Optionally, the step of performing causal coupling processing based on the intent credibility and the electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier includes:

[0033] The response delay is input into a preset time-series causal function to obtain the corresponding time-series causal decay value;

[0034] The preset coupling reference value and the electrical steady-state deviation are compared to obtain the corresponding first difference value.

[0035] The coupling reference value and the electrical fluctuation coefficient are compared to obtain the corresponding second difference value.

[0036] Based on preset coupling weights, the intent credibility, the temporal causal decay value, the first difference, and the second difference are weighted and calculated to obtain the corresponding causal coupling degree.

[0037] The coupling benchmark value and the causal coupling degree are compared to obtain the corresponding coupling anomaly score.

[0038] When the coupling anomaly score is greater than or equal to a preset coupling threshold, the preset first coupling identifier score is determined as the corresponding coupling identifier.

[0039] When the coupling anomaly score is less than the coupling threshold, the preset second coupling identifier score is determined as the corresponding coupling identifier.

[0040] By adopting the above technical solution, the response delay is processed according to the time-series causal function to obtain the time-series causal attenuation value, which effectively identifies the real correlation and false interference between actions and electrical responses, and strengthens the timing logic verification. The first difference and the second difference are calculated by combining the coupling benchmark value to quantify the deviation of electrical steady state, operation fluctuations and ideal state. Then, based on the preset coupling weight, the credibility of intent, the time-series causal attenuation value, the first difference and the second difference are weighted to obtain the corresponding causal coupling degree. By integrating multi-dimensional information such as human operation, timing correlation and electrical state, the coordinated relationship between operation and electrical response is fully reflected. By processing the difference between the coupling benchmark value and the causal coupling degree, the corresponding coupling anomaly score is obtained and the coupling identifier is determined by the coupling threshold. The quantification standard is unified and the logic is rigorous. It can accurately identify the causal mismatch between actions and electrical states. At the same time, the overall calculation process is simple and real-time, which further improves the system's ability to identify hidden faults and linkage anomalies.

[0041] Optionally, the step of performing operational risk detection based on the causal coupling degree, the action identifier, the electrical identifier, and the coupling identifier to obtain the corresponding anti-misoperation judgment result includes:

[0042] The causal coupling degree is used to retrieve a preset risk classification list and match the corresponding risk level;

[0043] When the risk level is normal operation, the risk level is determined as the corresponding anti-misoperation judgment result;

[0044] When the risk level is not normal operation, the corresponding target key is generated using the action identifier, the electrical identifier, and the coupling identifier;

[0045] The target key is used to retrieve a preset list of key-value pairs for accidental operation risks, and the corresponding accidental operation prevention judgment result is matched.

[0046] By adopting the above technical solution, risk levels can be quickly determined by retrieving a risk classification list based on causal coupling degree. This allows for a direct differentiation of the overall operational risk level. For normal operations, the judgment result is directly output, simplifying the process and improving response efficiency. For abnormal operating conditions, action identifiers, electrical identifiers, and coupling identifiers are integrated to generate target keys, enabling the encoding and summarization of multi-dimensional abnormal information. By retrieving the results from the list of misoperation risk key-value pairs, the abnormality type and root cause of the fault can be accurately located. This approach ensures both the comprehensiveness and accuracy of the judgment results and facilitates rapid retrieval and classification of results by the system, adapting to the application requirements of online real-time anti-misoperation detection in DC systems.

[0047] A second aspect of the present invention provides a DC system anti-misoperation detection system, comprising:

[0048] The acquisition module is used to acquire operation identification data and electrical operation data of the DC system, and to perform intent parsing on the operation identification data based on a preset operation intent library to obtain the corresponding intent credibility and action identifier;

[0049] The extraction module is used to extract electrical features from the electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers;

[0050] The causal coupling module is used to perform causal coupling processing based on the intent credibility and the electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier.

[0051] The detection module is used to perform operational risk detection based on the causal coupling degree, the action identifier, the electrical identifier, and the coupling identifier, and obtain the corresponding anti-misoperation judgment result.

[0052] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the DC system anti-misoperation judgment method as described above.

[0053] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, it implements the DC system anti-misoperation judgment method as described above.

[0054] The fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer performs the DC system anti-misoperation judgment method as described above.

[0055] As can be seen from the above technical solutions, the present invention has the following advantages:

[0056] This invention acquires operation identification data and electrical operation data of a DC system, performs intent parsing on the operation identification data based on a pre-set operation intent database to obtain corresponding intent credibility and action identifiers, extracts electrical features from the electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers, performs causal coupling processing based on intent credibility and electrical feature vectors to obtain corresponding causal coupling degree and coupling identifiers, and performs operation risk detection based on causal coupling degree, action identifiers, electrical identifiers, and coupling identifiers to obtain corresponding anti-misoperation judgment results. This overcomes the technical problem that existing DC system anti-misoperation technologies often only focus on single-dimensional step verification of operational behavior compliance, failing to accurately identify explicit human errors such as skipping steps, accidental touches, and incorrect operation objects, thus reducing the reliability of DC system operation. Compared with traditional methods for preventing misoperation in DC systems, this invention uses a pre-set operation intent database to perform intent parsing, which can accurately identify the operator's initial intention and the compliance of the action, and quickly output the intent credibility and action identifier. Then, it performs multi-dimensional feature extraction on electrical operation data to construct electrical feature vectors and generate electrical identifiers, which can comprehensively capture abnormal states such as steady-state, dynamic response, and operational fluctuations of equipment. Then, it combines the intent credibility and electrical feature vectors to perform causal coupling processing, quantifies the intrinsic relationship between operation behavior and electrical feedback, and obtains causal coupling degree and coupling identifier, effectively distinguishing between normal linkage and abnormal mismatch. Finally, it performs operation risk detection based on causal coupling degree, action identifier, electrical identifier, and coupling identifier to obtain the corresponding misoperation prevention judgment result. This realizes the joint identification of multiple risks such as human error, equipment electrical failure, and causal mismatch between action and electrical response, improving the reliability of DC system operation. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart of the steps of a DC system anti-misoperation judgment method provided in Embodiment 1 of the present invention;

[0059] Figure 2 This is a flowchart illustrating the steps of a DC system anti-misoperation detection method provided in Embodiment 2 of the present invention.

[0060] Figure 3 This is a structural block diagram of a DC system anti-misoperation discrimination system provided in Embodiment 3 of the present invention;

[0061] Figure 4This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0062] This invention provides a method, system, device, medium, and product for identifying misoperation in DC systems. It addresses the technical problem that existing DC system misoperation prevention technologies often only focus on single-dimensional step verification of operational compliance or only achieve post-event anomaly monitoring through electrical quantity threshold exceedances. These technologies are unable to accurately identify explicit human misoperations such as skipping steps, accidental contact, or incorrect operation objects, thus reducing the reliability of DC system operation.

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a DC system anti-misoperation detection method provided in Embodiment 1 of the present invention.

[0065] This invention provides a method for identifying and preventing misoperation in a DC system, comprising:

[0066] Step 101: Obtain the operation identification data and electrical operation data of the DC system, and perform intent parsing on the operation identification data based on the preset operation intent library to obtain the corresponding intent credibility and action identifier.

[0067] Operation recognition data refers to the video frame sequence of DC system operation site collected by industrial cameras, which includes visual motion information such as switch position, handle angle, and operation action trajectory. It is the core visual data for judging whether the operation behavior is compliant.

[0068] Electrical operating data refers to high-frequency sampling data of parameters such as DC system voltage, current, insulation resistance, closing circuit voltage, and electrical quantity change rate, which are used to reflect the electrical status and response characteristics of equipment.

[0069] The operation intent library refers to a database that includes a standard action template library and a timing rule library. It stores the standard action forms and timing specifications of compliant operation of DC systems, serving as a reference benchmark for intent parsing.

[0070] Intent credibility refers to the numerical value that represents the compliance of the operational intent after weighted normalization. The closer it is to 1, the more standard and compliant the operational intent is. The value range is [0,1].

[0071] Action identifiers are binary identifiers used to mark whether there is an explicit human error in an operation. 1 indicates that there is a human error and 0 indicates that the operation is normal.

[0072] In this embodiment of the invention, video images of the DC system operation process are first acquired in real time using a high-definition industrial camera to obtain operation recognition data. Simultaneously, electrical operation data such as voltage, current, and insulation resistance of the DC system are acquired through an electrical acquisition unit. The operation recognition data and electrical operation data are synchronized spatiotemporally using an IEEE 1588PTP unified clock source to ensure precise alignment in time and space. The synchronized operation recognition data is then input into a pre-trained target detection neural network model to detect and track targets such as switch handles and operating devices frame by frame, extracting action data such as switch position, handle angle, and motion trajectory. Based on a preset action matching function, the corresponding action matching degree is determined according to an action template library and each action data. Based on a preset temporal coherence function, the corresponding temporal coherence coefficient is determined according to a temporal rule library and each action data. The action matching degree and temporal coherence coefficient are weighted according to a preset intent weight to obtain an initial intent credibility, which is then normalized to obtain the corresponding intent credibility. Misoperation detection is performed based on the action matching degree, temporal coherence coefficient, and intent credibility to obtain the corresponding action identifier.

[0073] Step 102: Extract electrical features from the electrical operation data to obtain the corresponding electrical feature vectors and electrical identifiers.

[0074] The electrical characteristic vector is a three-dimensional vector composed of three indicators: electrical steady-state deviation, response delay, and electrical fluctuation coefficient, which comprehensively describes the electrical response characteristics during operation.

[0075] Electrical identification refers to a binary identifier that marks whether the electrical condition is abnormal, with 1 indicating an abnormal electrical condition and 0 indicating a normal electrical condition.

[0076] In this embodiment of the invention, electrical operating data is sequentially input into an electrical steady-state deviation function, a dynamic response delay calculation function, and an electrical fluctuation function to obtain, respectively, an electrical steady-state deviation reflecting the degree to which electrical quantities deviate from the standard state, a response delay reflecting the lag of electrical response behind operational actions, and an electrical fluctuation coefficient reflecting the severity of electrical quantity fluctuations. Based on preset electrical weights, the electrical steady-state deviation, response delay, and electrical fluctuation coefficient are weighted to obtain corresponding electrical anomaly scores. These scores are compared with preset electrical thresholds to determine whether there are problems such as electrical circuit anomalies or latent equipment faults, and an electrical identifier is output to mark whether the electrical state is abnormal.

[0077] It should be noted that the dynamic response delay calculation function is as follows:

[0078] ;

[0079] in, To respond to the delay, For electrical response time, This refers to the moment when the action is executed.

[0080] Step 103: Perform causal coupling processing based on intent credibility and electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier.

[0081] Causal coupling degree refers to the comprehensive matching value obtained by integrating intentional compliance, electrical deviation, timing delay, and fluctuation intensity. The value range is [0,1]. The higher the value, the better the operation matches the electrical state.

[0082] The coupling identifier is a binary identifier that marks whether there is a causal anomaly between the operational intent and the electrical response. 1 indicates a causal mismatch, and 0 indicates a normal causal match.

[0083] In this embodiment of the invention, the response delay in the electrical feature vector is substituted into a preset temporal causal function to obtain a temporal causal decay value used to distinguish between genuine causal relationships and spurious correlations. A preset coupling benchmark value is compared with the electrical steady-state deviation to obtain a corresponding first difference value. The coupling benchmark value is compared with the electrical fluctuation coefficient to obtain a corresponding second difference value. Based on preset coupling weights, the intent credibility, temporal causal decay value, first difference value, and second difference value are weighted to obtain a corresponding causal coupling degree. The coupling benchmark value is compared with the causal coupling degree to obtain a corresponding coupling anomaly score. The coupling anomaly score is compared with a preset coupling threshold to determine whether there is a causal mismatch between the operational action and the electrical response, and a coupling identifier used to mark whether the causal relationship is abnormal is output.

[0084] Step 104: Perform operational risk detection based on causal coupling degree, action identifier, electrical identifier, and coupling identifier to obtain the corresponding anti-misoperation judgment result.

[0085] In this embodiment of the invention, a preset risk classification list is retrieved using causal coupling degree to match the corresponding risk level. When the risk level is determined to be normal operation, the risk level is identified as the corresponding anti-misoperation judgment result. When the risk level is not normal operation, a target key is generated by combining action identifier, electrical identifier, and coupling identifier. The target key is then matched against a preset list of misoperation risk key-value pairs to accurately locate the specific risk root cause, such as human error, electrical fault, hidden transmission failure, or multiple risks. Simultaneously, a complete anti-misoperation judgment result containing the risk level, risk root cause, and protection command is output.

[0086] In this embodiment of the invention, by acquiring operation identification data and electrical operation data of a DC system, the operation identification data is parsed based on a preset operation intent library to obtain the corresponding intent credibility and action identifier. Electrical features are extracted from the electrical operation data to obtain the corresponding electrical feature vector and electrical identifier. Causal coupling processing is performed based on the intent credibility and electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier. Operational risk detection is performed based on the causal coupling degree, action identifier, electrical identifier, and coupling identifier to obtain the corresponding anti-misoperation judgment result. This overcomes the technical problem that existing DC system anti-misoperation technologies often only perform single-dimensional step verification around the compliance of operational behavior, failing to accurately identify explicit human errors such as skipping steps, accidental contact, and incorrect operation objects, thus reducing the reliability of DC system operation. Compared with traditional methods for preventing misoperation in DC systems, this invention uses a pre-set operation intent database to perform intent parsing, which can accurately identify the operator's initial intention and the compliance of the action, and quickly output the intent credibility and action identifier. Then, it performs multi-dimensional feature extraction on electrical operation data to construct electrical feature vectors and generate electrical identifiers, which can comprehensively capture abnormal states such as steady-state, dynamic response, and operational fluctuations of equipment. Then, it combines the intent credibility and electrical feature vectors to perform causal coupling processing, quantifies the intrinsic relationship between operation behavior and electrical feedback, and obtains causal coupling degree and coupling identifier, effectively distinguishing between normal linkage and abnormal mismatch. Finally, it performs operation risk detection based on causal coupling degree, action identifier, electrical identifier, and coupling identifier to obtain the corresponding misoperation prevention judgment result. This realizes the joint identification of multiple risks such as human error, equipment electrical failure, and causal mismatch between action and electrical response, improving the reliability of DC system operation.

[0087] Please see Figure 2 , Figure 2 The flowchart illustrates the steps of a DC system anti-misoperation judgment method provided in Embodiment 2 of the present invention.

[0088] This invention provides a method for identifying and preventing misoperation in a DC system, comprising:

[0089] Step 201: Obtain the operation identification data and electrical operation data of the DC system, and perform intent parsing on the operation identification data based on the preset operation intent library to obtain the corresponding intent credibility and action identifier.

[0090] Furthermore, the operation intent library includes an action template library and a timing rule library, and step 201 includes the following sub-steps:

[0091] S11. Based on the pre-trained target detection model, target detection is performed on each frame of operation image in the operation recognition data in sequence to obtain multiple action data.

[0092] Motion data refers to quantitative operational information such as switch position, handle angle, and motion trajectory offset obtained through target detection and tracking.

[0093] In this embodiment of the invention, based on a pre-trained target detection model, target detection is performed sequentially on each frame of operation images in the operation recognition data. Key targets such as switch handles, operating mechanisms, and operating positions in the DC system operation site are accurately located and identified. Quantitative information such as switch open / closed status, handle rotation angle, and motion trajectory offset are output in real time. This information is then integrated to form multiple sets of continuous motion data.

[0094] It should be noted that the training process of the object detection model is as follows: First, historical operation images of DC systems with different types of handles under various states are collected to construct an image dataset covering multiple operating conditions and states. Then, a target state label library is established, and the images are labeled according to specifications to form an operation target image sample set that can be used for training, which is then divided into training set, validation set, and test set according to proportions. Subsequently, the YOLO lightweight deep learning model is selected, and pre-trained weights from a public dataset are loaded based on the transfer learning strategy. The backbone network is first frozen for warm-up training, and then the network is unfrozen for full training. During the training process, various data augmentation strategies are used to improve the model's generalization ability, and the accuracy is checked periodically using the validation set until the relevant detection indicators of the model reach the preset threshold. Finally, an independent test set is used to verify the generalization. After meeting the requirements for recognition accuracy and inference latency, the model is converted into a format suitable for edge deployment, completing the construction and training of the entire object detection model.

[0095] S12. Based on the preset action matching function, determine the corresponding action matching degree according to the action template library and each action data.

[0096] The action template library refers to a database that is pre-built and stores the characteristics of various compliant standard operating actions of DC systems, including benchmark data such as standard switch positions, handle angles, and action trajectories.

[0097] Action matching degree refers to the similarity between the actual operation action and the standard action template. The value ranges from [0,1]. The closer the value is to 1, the more standard and compliant the action is.

[0098] In this embodiment of the invention, each action data is compared with the standard operation actions stored in the preset action template library in a dimension-by-dimensional manner. The similarity between the actual action and the standard action is calculated by the preset action matching function in terms of shape, position, angle and trajectory features. The deviation caused by instantaneous interference and noise is eliminated, and the action matching degree that can intuitively reflect the consistency between the actual operation and the standard operation is output.

[0099] It should be noted that the action matching function is as follows:

[0100] ;

[0101] in, For action matching degree, This represents the total number of sampling times. The action data collected at time t, For the standard action template data at time t, This is a time-by-time matching function. It is 1 when the motion data matches the standard motion template data, and 0 when the motion data does not match the standard motion template data. t is the time index.

[0102] S13. Based on the preset temporal coherence function, determine the corresponding temporal coherence coefficient according to the temporal rule base and each action data.

[0103] The timing rule base refers to a pre-defined database of standard operating procedures, execution order, action intervals, and logical constraints for DC systems.

[0104] The temporal continuity coefficient is a quantitative indicator that represents whether the operation is in a standardized and reasonable time sequence. The value range is [0,1], and the closer it is to 1, the more compliant the temporal sequence is.

[0105] In this embodiment of the invention, based on each action data and the standard operation step sequence, action interval and execution logic specified in the timing rule base, the timing change amount, step connection rationality and process integrity of the actual operation are calculated segment by segment through a preset timing coherence function. The timing interference caused by occasional jitter and temporary pauses is eliminated, and a timing coherence coefficient that can accurately reflect whether the operation process conforms to the standard sequence is obtained.

[0106] It should be noted that the timing coherence function is specifically as follows:

[0107] ;

[0108] in, For time series coherence coefficient, Let be the change in motion at time t. Let be the standard motion change at time t. The action data collected at time t-1 This is the standard motion template data at time t-1.

[0109] S14. Based on the preset intent weight, the action matching degree and the temporal coherence coefficient are weighted to obtain the initial intent credibility, and the initial intent credibility is normalized to obtain the corresponding intent credibility.

[0110] Intent weight refers to the proportional coefficient that is pre-set according to the DC system operation specifications to reflect the importance of action matching degree and timing continuity coefficient in intent judgment. The sum of the two weights is 1.

[0111] Initial intent credibility refers to the intermediate result after weighted calculation without normalization, which is used to initially characterize the compliance level of the operational intent.

[0112] Intent credibility refers to the final quantitative value obtained after weighting and normalization, which is used to characterize the true compliance of the operational intent. The value ranges from [0,1], and the closer it is to 1, the more normal the intent is.

[0113] In this embodiment of the invention, the action matching degree and the temporal coherence coefficient are weighted and summed based on a preset intent weight to obtain the initial intent credibility, which reflects the initial compliance level of the operation intent. The initial intent credibility is then normalized to obtain the corresponding intent credibility.

[0114] It should be noted that the expression for the credibility of the initial intent is as follows:

[0115] ;

[0116] in, To assess the credibility of the initial intent, The first intention weight coefficient, This is the weighting coefficient for the second intention.

[0117] S15. Perform erroneous operation detection based on action matching degree, temporal coherence coefficient and intent credibility to obtain the corresponding action identifier.

[0118] Furthermore, S15 includes the following sub-steps:

[0119] S151. The action matching degree, temporal coherence coefficient and intent credibility are averaged to obtain the corresponding first mean.

[0120] In this embodiment of the invention, the mean value among the action matching degree, the temporal coherence coefficient and the intent credibility is calculated to obtain the corresponding first mean value.

[0121] S152. The preset action baseline value and the first mean value are processed to obtain the corresponding action abnormality score.

[0122] The action baseline value refers to the pre-set ideal operating state reference value, which is usually set to 1. It is used to measure the degree of deviation between the actual action and the standard action.

[0123] In this embodiment of the invention, the difference between a preset action benchmark value and a first mean value is calculated to obtain the corresponding action abnormality score.

[0124] It should be noted that the expression for the abnormality score is as follows:

[0125] ;

[0126] in, This represents the score for abnormal behavior.

[0127] S153. When the abnormal action score is greater than or equal to the preset action threshold, the preset first action identifier score is determined as the corresponding action identifier.

[0128] Action threshold refers to the critical judgment value set in advance according to the DC system operation safety specifications, used to distinguish between normal operation and abnormal action.

[0129] The first action identification score refers to the preset identification value used to mark the existence of human error. It is usually set to 1, which represents an abnormal action.

[0130] In this embodiment of the invention, when the abnormal action score is greater than or equal to the action threshold, it is determined that there is an obvious human error such as accidental touch, skipping step, or incorrect operation object in the current operation, and the preset first action identifier score is determined as the action identifier.

[0131] S154. When the abnormal action score is less than the action threshold, the preset second action identifier score is determined as the corresponding action identifier.

[0132] The second action identification score refers to the preset identification value used to mark the normal operation action. It is usually set to 0, which means the action is compliant.

[0133] In this embodiment of the invention, when the abnormal action score is less than the action threshold, the operation action is determined to meet the specifications, and the preset second action identifier score is determined as the corresponding action identifier.

[0134] It's worth noting that the algorithm first calculates the average of action matching degree, temporal coherence coefficient, and intent credibility, integrating multi-dimensional evaluation indicators to form a comprehensive evaluation score. Then, it performs a difference calculation based on the action baseline value to obtain an action anomaly score, intuitively quantifying the overall deviation of the operational behavior. Based on action thresholds, it performs interval judgments and matches corresponding action identifier scores according to rules to differentiate completion states. The entire calculation logic is simple and clear, with low computational overhead. The judgment standard is unified and highly quantifiable, objectively distinguishing between normal and abnormal operations, effectively avoiding subjective judgment bias. It is also easy to implement, quickly outputting stable and reusable action identifiers, improving the standardization and real-time performance of misoperation identification.

[0135] It should be noted that by parsing and extracting action data frame by frame from the operation image using a pre-trained target detection model, operation details can be accurately captured, ensuring the accuracy of the original behavior data collection. The action matching degree and temporal coherence coefficient are calculated using an action template library and a temporal rule library, respectively, to verify the standardization of the operation from both the action form and execution flow dimensions. Then, a standardized quantified intent credibility is obtained through intent weighting and normalization. Misoperation detection is performed based on the action matching degree, the temporal coherence coefficient, and the intent credibility, resulting in corresponding action identifiers. This effectively reduces human judgment errors, accurately identifies human misoperations, and provides objective and reliable detection results. Furthermore, relying on the model and standardized algorithms improves detection efficiency and versatility, and can stably adapt to the operation behavior recognition needs in different scenarios.

[0136] Step 202: Extract electrical features from the electrical operation data to obtain the corresponding electrical feature vectors and electrical identifiers.

[0137] Further, step 202 includes the following sub-steps:

[0138] S21. Input the electrical operation data into the preset electrical steady-state deviation function to obtain the corresponding electrical steady-state deviation.

[0139] Electrical steady-state deviation refers to the quantified deviation between the actual electrical steady-state value and the standard reference value. The larger the value, the more abnormal the electrical condition.

[0140] In this embodiment of the invention, electrical operating data is input into a preset electrical steady-state deviation function. The deviation between the actual electrical quantity and the ideal steady-state value is calculated based on a standard electrical reference template. Errors caused by instantaneous fluctuations and sampling interference are eliminated to obtain an electrical steady-state deviation that can accurately reflect the electrical state's deviation from the normal operating range.

[0141] It should be noted that the electrical steady-state deviation function is specifically as follows:

[0142] ;

[0143] in, For electrical steady-state deviation, The electrical quantity vector at time t (including parameters such as DC system voltage, current, insulation resistance, closing circuit voltage, and rate of change of electrical quantities at time t). Let be the standard electrical reference vector at time t.

[0144] S22. Perform difference processing on the electrical response time and the corresponding action execution time in the electrical operation data to obtain the corresponding response delay.

[0145] Response delay refers to the time difference between the execution of an operation and the generation of an effective electrical response. It is used to determine whether there are hidden failures in the transmission mechanism, such as jamming, stripping, or shaft breakage.

[0146] The electrical response time refers to the precise point in time when key electrical quantities such as voltage and current in electrical operating data begin to change effectively, indicating that the equipment has generated electrical feedback.

[0147] The execution time of an action refers to the precise point in time when a switch, handle, or other mechanism in the operation action data completes its actual action and triggers a change in the state of the equipment.

[0148] In this embodiment of the invention, the difference between the electrical response time and the corresponding action execution time in the electrical operation data is calculated to obtain a response delay that reflects whether the mechanism transmission is lagging, whether there is jamming or failure.

[0149] S23. Input the electrical operation data into the preset electrical fluctuation function to obtain the corresponding electrical fluctuation coefficient.

[0150] The electrical fluctuation coefficient is a quantitative value that characterizes the magnitude and stability of electrical quantity fluctuations during operation. The larger the value, the more severe the electrical fluctuations and the more unstable the state.

[0151] In this embodiment of the invention, electrical operation data is input into a preset electrical fluctuation function to obtain an electrical fluctuation coefficient that accurately characterizes the smoothness of electrical circuit operation.

[0152] It should be noted that the electrical fluctuation function is specifically as follows:

[0153] ;

[0154] in, This is the electrical fluctuation coefficient.

[0155] S24. Electrical steady-state deviation, response delay, and electrical fluctuation coefficient are used as the corresponding electrical characteristic vectors.

[0156] In this embodiment of the invention, electrical steady-state deviation, response delay, and electrical fluctuation coefficient are determined as the corresponding electrical feature vectors.

[0157] S25. Based on the preset electrical weights, perform weighted calculations on the electrical steady-state deviation, response delay, and electrical fluctuation coefficient to obtain the corresponding electrical anomaly score.

[0158] Electrical weight refers to a proportional coefficient pre-set according to the DC system safety specifications, used to reflect the importance of electrical steady-state deviation, response delay, and electrical fluctuation coefficient in anomaly determination. The sum of the weights of the three is 1.

[0159] In this embodiment of the invention, the electrical steady-state deviation, response delay and electrical fluctuation coefficient are weighted based on preset electrical weights to obtain an electrical anomaly score that can comprehensively reflect the degree of electrical anomaly.

[0160] S26. When the electrical anomaly score is greater than or equal to the preset electrical threshold, the preset first electrical identification score is determined as the corresponding electrical identification.

[0161] Electrical threshold refers to the critical judgment value pre-set according to the safety operation standard of DC system, which is used to distinguish between normal electrical state and abnormal electrical state.

[0162] The first electrical identification score refers to the preset identification value used to mark abnormal electrical conditions. It is usually set to 1, which means that there is an electrical fault or abnormality.

[0163] In this embodiment of the invention, when the electrical anomaly score is greater than or equal to a preset electrical threshold, it is determined that the current DC system has an electrical circuit anomaly, equipment failure, or unstable state, and the preset first electrical identifier score is determined as the corresponding electrical identifier.

[0164] S27. When the electrical anomaly score is less than the electrical threshold, the preset second electrical identifier score is determined as the corresponding electrical identifier.

[0165] The second electrical identification score refers to the preset identification value used to mark the normal electrical condition. It is usually set to 0, which means that the electrical operation is stable and compliant.

[0166] In this embodiment of the invention, when the electrical anomaly score is less than the electrical threshold, the electrical operating state is determined to be normal and stable, and the preset second electrical identifier score is determined as the corresponding electrical identifier.

[0167] It is worth mentioning that by comprehensively analyzing electrical operation data from three dimensions—steady-state deviation, response delay, and fluctuation coefficient—a complete electrical feature vector is constructed, which can fully cover three types of state information: static operating conditions, action response timing, and operational stability of equipment, with comprehensive and complete feature representation. By weighting and fusing various indicators with electrical weights, an electrical anomaly score is obtained, taking into account the judgment priority of different electrical features, and the quantitative results are more in line with on-site operating standards. Combined with electrical thresholds, a graded judgment is completed and an electrical identifier is output. This can accurately identify various anomalies such as electrical circuit faults, transmission lag, and operational fluctuations, and facilitates the deployment of automated programs, improving the accuracy and real-time performance of electrical condition detection.

[0168] Step 203: Perform causal coupling processing based on intent credibility and electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier.

[0169] Furthermore, step 203 includes the following sub-steps:

[0170] S31. Input the response delay into the preset time-series causal function to obtain the corresponding time-series causal decay value.

[0171] The temporal causal decay value refers to the weighting coefficient output by the temporal causal function, which is used to strengthen the effective causal relationship and suppress misjudgments caused by excessive delay. The value range is [0,1].

[0172] In this embodiment of the invention, the response delay is input into a preset time-series causal function, and the delay value is calculated using the standard response time as a reference. This automatically weakens the spurious correlation signal caused by the excessive delay and strengthens the true causal relationship of the effective response, thereby obtaining a time-series causal attenuation value that can distinguish between true causality and spurious correlation.

[0173] It should be noted that the time-series causal function is specifically as follows:

[0174] ;

[0175] in, This represents the time-series causal decay value. This is the standard response time, typically taken as 100ms. The response was delayed.

[0176] S32. Perform difference processing on the preset coupling reference value and the electrical steady-state deviation to obtain the corresponding first difference value.

[0177] The coupling reference value refers to a pre-set ideal electrical state reference value, which is usually set to 1 and is used to measure the degree to which electrical characteristics deviate from the normal level.

[0178] In this embodiment of the invention, the difference between the preset coupling reference value and the electrical steady-state deviation is calculated to obtain the corresponding first difference.

[0179] S33. Perform difference processing on the coupling reference value and the electrical fluctuation coefficient to obtain the corresponding second difference value.

[0180] In this embodiment of the invention, the difference between the coupling reference value and the electrical fluctuation coefficient is calculated to obtain the corresponding second difference.

[0181] S34. Based on the preset coupling weights, perform weighted calculations on the intent credibility, temporal causal decay value, first difference and second difference to obtain the corresponding causal coupling degree.

[0182] Coupling weight refers to the proportional coefficient pre-set according to the DC system's anti-misoperation rules, used to reflect the importance of the four indicators in the causal coupling judgment, and the sum of all weights is 1.

[0183] In this embodiment of the invention, based on a preset coupling weight, the intention credibility, the temporal causal decay value, the first difference and the second difference are weighted and calculated to obtain the causal coupling degree that can uniformly quantify the overall synergy between the operation action and the electrical response.

[0184] It should be noted that the expression for the degree of causal coupling is as follows:

[0185] ;

[0186] in, For causal coupling degree, The first coupling weight coefficient, The second coupling weight coefficient, The third coupling weight coefficient, This is the fourth coupling weight coefficient. For the credibility of the intent.

[0187] S35. The coupling baseline value and the causal coupling degree are compared to obtain the corresponding coupling anomaly score.

[0188] The coupling anomaly score is a quantitative score that characterizes the deviation of the causal match between the operation action and the electrical response from the ideal state. The higher the score, the more severe the causal mismatch.

[0189] In this embodiment of the invention, the difference between the coupling benchmark value and the causal coupling degree is calculated to obtain a coupling anomaly score that can be directly used for coupling identification.

[0190] S36. When the coupling anomaly score is greater than or equal to the preset coupling threshold, the preset first coupling identifier score is determined as the corresponding coupling identifier.

[0191] The coupling threshold refers to a critical judgment value pre-set according to the DC system safety standard, used to distinguish between normal causal matching and abnormal causal mismatch.

[0192] The first coupling flag score refers to the preset flag value used to mark causal mismatches. It is usually set to 1, which means that there is an abnormal correlation between the action and the electrical response.

[0193] In this embodiment of the invention, when the coupling anomaly score is greater than or equal to a preset coupling threshold, it is determined that there is a significant causal mismatch between the operation action and the electrical response, a hidden transmission failure, or a logic anomaly, and the preset first coupling identifier score is determined as the corresponding coupling identifier.

[0194] S37. When the coupling anomaly score is less than the coupling threshold, the preset second coupling identifier score is determined as the corresponding coupling identifier.

[0195] The second coupling identifier score refers to the preset identifier value used to mark normal causal matching. It is usually set to 0, which means that the action and electrical response logic are compliant.

[0196] In this embodiment of the invention, when the coupling anomaly score is less than the coupling threshold, it is determined that the causal relationship between the action and the electrical response is normal and the matching is effective, and the preset second coupling identifier score is determined as the corresponding coupling identifier.

[0197] It is worth mentioning that by processing the response delay according to the time-series causal function to obtain the time-series causal attenuation value, the true correlation and false interference between the action and the electrical response are effectively identified, and the time-series logic verification is strengthened. The first difference and the second difference are calculated by combining the coupling benchmark value to quantify the deviation of the electrical steady state, operation fluctuation and ideal state. Then, based on the preset coupling weight, the credibility of the intent, the time-series causal attenuation value, the first difference and the second difference are weighted to obtain the corresponding causal coupling degree. By integrating multi-dimensional information of human operation, time-series correlation and electrical state, the coordinated relationship between operation and electrical response is fully reflected. By processing the difference between the coupling benchmark value and the causal coupling degree, the corresponding coupling anomaly score is obtained and the coupling identifier is determined by the coupling threshold. The quantification standard is unified and the logic is rigorous. It can accurately identify the causal mismatch between the action and the electrical state. At the same time, the overall calculation process is simple and has strong real-time performance, which further improves the system's ability to identify hidden faults and linkage anomalies.

[0198] Step 204: Use causal coupling degree to retrieve the preset risk classification list and match the corresponding risk level.

[0199] The risk classification list refers to a classification table that is pre-set according to the DC system safety specifications and uses causal coupling degree as the retrieval basis. It includes four levels of risk: normal, minor, severe, and emergency.

[0200] In this embodiment of the invention, a preset risk classification list is retrieved using causal coupling degree, and the corresponding risk level is quickly located and output according to the coupling degree value range.

[0201] It should be noted that the specific expression for the numerical range of coupling degree is as follows:

[0202] ;

[0203] in, It is classified as a risk level.

[0204] Step 205: When the risk level is normal operation, the risk level is determined as the corresponding anti-misoperation judgment result.

[0205] In this embodiment of the invention, when the risk level is normal operation, the current operation action, operation intention and electrical response all meet the specification requirements, and the risk level is determined as the corresponding anti-misoperation judgment result.

[0206] Step 206: When the risk level is not normal operation, the corresponding target key is generated using action identifier, electrical identifier, and coupling identifier.

[0207] The target key refers to the retrieval code generated by combining three types of identifiers according to rules, which is used to quickly retrieve the corresponding risk type.

[0208] In this embodiment of the invention, when the risk level is not normal operation, it indicates that there is a safety hazard in the current operation. Then, the previously generated action identifier, electrical identifier and coupling identifier are retrieved and combined according to the preset coding rules to form a unique target key.

[0209] Step 207: Use the target key to retrieve the preset list of key-value pairs for accidental operation risks and match the corresponding anti-accidental operation judgment results.

[0210] In this embodiment of the invention, a target key is used to retrieve a preset list of key-value pairs for erroneous operation risks, match the anomaly type and judgment conclusion corresponding to the current operation, and generate a complete anti-erroneous operation judgment result.

[0211] It should be noted that the list of key-value pairs for accidental operation risks is shown in Table 1.

[0212] Table 1

[0213]

[0214] It is worth mentioning that by quickly classifying risk levels based on the risk grading list according to causal coupling degree, the overall operational risk level can be intuitively distinguished. For normal operations, the judgment result is directly output, simplifying the process and improving response efficiency. For abnormal operating conditions, action identifiers, electrical identifiers, and coupling identifiers are integrated to generate target keys, realizing the encoding and summary of multi-dimensional abnormal information. Then, by searching the list of misoperation risk key-value pairs and matching the results, the abnormality type and fault root cause can be accurately located. This ensures both the comprehensiveness and accuracy of the judgment results and facilitates the system's rapid retrieval and classification of results, adapting to the application requirements of online real-time anti-misoperation detection in DC systems.

[0215] In this embodiment of the invention, by acquiring operation identification data and electrical operation data of a DC system, the operation identification data is parsed based on a preset operation intent library to obtain the corresponding intent credibility and action identifier. Electrical features are extracted from the electrical operation data to obtain the corresponding electrical feature vector and electrical identifier. Causal coupling processing is performed based on the intent credibility and electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier. Operational risk detection is performed based on the causal coupling degree, action identifier, electrical identifier, and coupling identifier to obtain the corresponding anti-misoperation judgment result. This overcomes the technical problem that existing DC system anti-misoperation technologies often only perform single-dimensional step verification around the compliance of operational behavior, failing to accurately identify explicit human errors such as skipping steps, accidental contact, and incorrect operation objects, thus reducing the reliability of DC system operation. Compared with traditional methods for preventing misoperation in DC systems, this invention uses a pre-set operation intent database to perform intent parsing, which can accurately identify the operator's initial intention and the compliance of the action, and quickly output the intent credibility and action identifier. Then, it performs multi-dimensional feature extraction on electrical operation data to construct electrical feature vectors and generate electrical identifiers, which can comprehensively capture abnormal states such as steady-state, dynamic response, and operational fluctuations of equipment. Then, it combines the intent credibility and electrical feature vectors to perform causal coupling processing, quantifies the intrinsic relationship between operation behavior and electrical feedback, and obtains causal coupling degree and coupling identifier, effectively distinguishing between normal linkage and abnormal mismatch. Finally, it performs operation risk detection based on causal coupling degree, action identifier, electrical identifier, and coupling identifier to obtain the corresponding misoperation prevention judgment result. This realizes the joint identification of multiple risks such as human error, equipment electrical failure, and causal mismatch between action and electrical response, improving the reliability of DC system operation.

[0216] Please see Figure 3 , Figure 3 This is a structural block diagram of a DC system anti-misoperation discrimination system provided in Embodiment 3 of the present invention.

[0217] This invention provides a DC system anti-misoperation detection system, comprising:

[0218] The acquisition module 301 is used to acquire operation identification data and electrical operation data of the DC system, and perform intent parsing on the operation identification data based on a preset operation intent library to obtain the corresponding intent credibility and action identifier.

[0219] Extraction module 302 is used to extract electrical features from electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers;

[0220] The causal coupling module 303 is used to perform causal coupling processing based on the intent credibility and electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier.

[0221] The detection module 304 is used to detect operational risks based on causal coupling degree, action identifier, electrical identifier and coupling identifier, and obtain the corresponding anti-misoperation judgment result.

[0222] Furthermore, the operation intent library includes an action template library and a timing rule library, and the acquisition module 301 includes:

[0223] The intent parsing submodule is used to perform target detection on each frame of the operation image in the operation recognition data sequentially based on a pre-trained target detection model to obtain multiple action data.

[0224] Based on a preset action matching function, the corresponding action matching degree is determined according to the action template library and each action data;

[0225] Based on the preset temporal coherence function, the corresponding temporal coherence coefficient is determined according to the temporal rule base and each action data.

[0226] The action matching degree and temporal coherence coefficient are weighted based on the preset intent weight to obtain the initial intent credibility. The initial intent credibility is then normalized to obtain the corresponding intent credibility.

[0227] The action identification submodule is used to detect erroneous operations based on action matching degree, temporal coherence coefficient and intent credibility, and obtain the corresponding action identification.

[0228] Furthermore, the action identifier submodule includes:

[0229] The mean unit is used to average the action matching degree, temporal coherence coefficient, and intent credibility to obtain the corresponding first mean.

[0230] The difference between the preset action baseline value and the first mean value is processed to obtain the corresponding action abnormality score;

[0231] The identification unit is used to determine the preset first action identification score as the corresponding action identification when the action abnormality score is greater than or equal to the preset action threshold.

[0232] When the abnormal action score is less than the action threshold, the preset second action identifier score is determined as the corresponding action identifier.

[0233] Furthermore, the extraction module 302 includes:

[0234] The electrical feature extraction submodule is used to input electrical operating data into a preset electrical steady-state deviation function to obtain the corresponding electrical steady-state deviation.

[0235] The difference between the electrical response time and the corresponding action execution time in the electrical operation data is processed to obtain the corresponding response delay;

[0236] Input the electrical operation data into the preset electrical fluctuation function to obtain the corresponding electrical fluctuation coefficient;

[0237] Electrical steady-state deviation, response delay, and electrical fluctuation coefficient are used as the corresponding electrical feature vectors;

[0238] The electrical steady-state deviation, response delay, and electrical fluctuation coefficient are weighted based on preset electrical weights to obtain the corresponding electrical anomaly score;

[0239] The electrical identification submodule is used to determine the preset first electrical identification score as the corresponding electrical identification when the electrical anomaly score is greater than or equal to the preset electrical threshold.

[0240] When the electrical anomaly score is less than the electrical threshold, the preset second electrical identifier score is determined as the corresponding electrical identifier.

[0241] Furthermore, the causal coupling module 303 includes:

[0242] The causal coupling submodule is used to input the response delay into a preset time-series causal function to obtain the corresponding time-series causal decay value;

[0243] The preset coupling reference value and the electrical steady-state deviation are processed to obtain the corresponding first difference value;

[0244] The second difference value is obtained by subtracting the coupling reference value from the electrical fluctuation coefficient.

[0245] Based on the preset coupling weights, the intent credibility, temporal causal decay value, first difference and second difference are weighted and calculated to obtain the corresponding causal coupling degree.

[0246] The coupling benchmark value and the causal coupling degree are subtracted to obtain the corresponding coupling anomaly score;

[0247] The coupling identifier submodule is used to determine the preset first coupling identifier score as the corresponding coupling identifier when the coupling anomaly score is greater than or equal to the preset coupling threshold.

[0248] When the coupling anomaly score is less than the coupling threshold, the preset second coupling identifier score is determined as the corresponding coupling identifier.

[0249] Furthermore, the detection module 304 includes:

[0250] The retrieval submodule is used to retrieve a pre-defined risk classification list based on causal coupling degree and match the corresponding risk level.

[0251] When the risk level is normal operation, the risk level is determined as the corresponding anti-misoperation judgment result;

[0252] The matching submodule is used to generate a corresponding target key using action identifier, electrical identifier, and coupling identifier when the risk level is not normal operation.

[0253] The target key is used to retrieve a pre-defined list of key-value pairs for accidental operation risks and match the corresponding accidental operation prevention judgment results.

[0254] Please see Figure 4 , Figure 4 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention.

[0255] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402. The memory 401 stores a computer program. When the computer program is executed by the processor 402, the processor 402 performs the DC system anti-misoperation judgment method as described in any of the above embodiments.

[0256] Memory 401 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 401 has storage space 403 for program code 413 for performing any of the method steps described above. For example, storage space 403 for program code may include individual program codes 413 for implementing the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When these codes are run by a computing device, they cause the computing device to perform the various steps in the DC system anti-misoperation detection method described above.

[0257] Embodiment 5 of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the DC system anti-misoperation judgment method as described in any of the above embodiments.

[0258] Embodiment 6 of the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the DC system anti-misoperation judgment method as described in any of the above embodiments.

[0259] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0260] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0261] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0262] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0263] If the integrated unit is implemented as 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 invention, in essence, 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. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0264] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying misoperation prevention in a DC system, characterized in that, include: Acquire operation identification data and electrical operation data of the DC system, and perform intent parsing on the operation identification data based on a preset operation intent library to obtain the corresponding intent credibility and action identifier; Electrical features are extracted from the electrical operation data to obtain the corresponding electrical feature vectors and electrical identifiers; Based on the intent credibility and the electrical feature vector, causal coupling processing is performed to obtain the corresponding causal coupling degree and coupling identifier; Based on the causal coupling degree, the action identifier, the electrical identifier, and the coupling identifier, operational risk detection is performed to obtain the corresponding anti-misoperation judgment result; The step of extracting electrical features from the electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers includes: The electrical operating data is input into a preset electrical steady-state deviation function to obtain the corresponding electrical steady-state deviation; The electrical response time and the corresponding action execution time in the electrical operation data are processed by difference to obtain the corresponding response delay; The electrical operation data is input into a preset electrical fluctuation function to obtain the corresponding electrical fluctuation coefficient; The electrical steady-state deviation, the response delay, and the electrical fluctuation coefficient are used as the corresponding electrical feature vectors; The step of performing causal coupling processing based on the intent credibility and the electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier includes: The response delay is input into a preset time-series causal function to obtain the corresponding time-series causal decay value; The preset coupling reference value and the electrical steady-state deviation are compared to obtain the corresponding first difference value. The coupling reference value and the electrical fluctuation coefficient are compared to obtain the corresponding second difference value. Based on preset coupling weights, the intent credibility, the temporal causal decay value, the first difference, and the second difference are weighted and calculated to obtain the corresponding causal coupling degree. The coupling benchmark value and the causal coupling degree are compared to obtain the corresponding coupling anomaly score. When the coupling anomaly score is greater than or equal to a preset coupling threshold, the preset first coupling identifier score is determined as the corresponding coupling identifier. When the coupling anomaly score is less than the coupling threshold, the preset second coupling identifier score is determined as the corresponding coupling identifier.

2. The DC system anti-misoperation judgment method according to claim 1, characterized in that, The operation intent library includes an action template library and a timing rule library. The step of parsing the operation recognition data based on the preset operation intent library to obtain the corresponding intent credibility and action identifier includes: Based on a pre-trained target detection model, target detection is performed sequentially on each frame of the operation image in the operation recognition data to obtain multiple action data. Based on a preset action matching function, the corresponding action matching degree is determined according to the action template library and each action data; Based on a preset temporal coherence function, the corresponding temporal coherence coefficient is determined according to the temporal rule base and each action data. The action matching degree and the temporal coherence coefficient are weighted based on the preset intent weight to obtain the initial intent credibility, and the initial intent credibility is normalized to obtain the corresponding intent credibility. Misoperation detection is performed based on the action matching degree, the temporal coherence coefficient, and the intent credibility to obtain the corresponding action identifier.

3. The DC system anti-misoperation judgment method according to claim 2, characterized in that, The step of detecting erroneous operations based on the action matching degree, the temporal coherence coefficient, and the intent credibility to obtain the corresponding action identifier includes: The action matching degree, the temporal coherence coefficient, and the intent credibility are averaged to obtain the corresponding first average. The difference between the preset action benchmark value and the first mean value is processed to obtain the corresponding action abnormality score; When the abnormal action score is greater than or equal to the preset action threshold, the preset first action identifier score is determined as the corresponding action identifier. When the abnormal action score is less than the action threshold, the preset second action identifier score is determined as the corresponding action identifier.

4. The DC system anti-misoperation judgment method according to claim 1, characterized in that, The step of extracting electrical features from the electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers further includes: The electrical steady-state deviation, the response delay, and the electrical fluctuation coefficient are weighted according to preset electrical weights to obtain the corresponding electrical anomaly score; When the electrical anomaly score is greater than or equal to a preset electrical threshold, the preset first electrical identification score is determined as the corresponding electrical identification. When the electrical anomaly score is less than the electrical threshold, the preset second electrical identifier score is determined as the corresponding electrical identifier.

5. The DC system anti-misoperation judgment method according to claim 1, characterized in that, The step of performing operational risk detection based on the causal coupling degree, the action identifier, the electrical identifier, and the coupling identifier to obtain the corresponding anti-misoperation judgment result includes: The causal coupling degree is used to retrieve a preset risk classification list and match the corresponding risk level; When the risk level is normal operation, the risk level is determined as the corresponding anti-misoperation judgment result; When the risk level is not normal operation, the corresponding target key is generated using the action identifier, the electrical identifier, and the coupling identifier; The target key is used to retrieve a preset list of key-value pairs for accidental operation risks, and the corresponding accidental operation prevention judgment result is matched.

6. A DC system misoperation prevention and discrimination system, used to implement the DC system misoperation prevention and discrimination method according to any one of claims 1-5, characterized in that, include: The acquisition module is used to acquire operation identification data and electrical operation data of the DC system, and to perform intent parsing on the operation identification data based on a preset operation intent library to obtain the corresponding intent credibility and action identifier; The extraction module is used to extract electrical features from the electrical operation data to obtain corresponding electrical feature vectors and electrical identifiers; The causal coupling module is used to perform causal coupling processing based on the intent credibility and the electrical feature vector to obtain the corresponding causal coupling degree and coupling identifier. The detection module is used to perform operational risk detection based on the causal coupling degree, the action identifier, the electrical identifier, and the coupling identifier, and obtain the corresponding anti-misoperation judgment result.

7. An electronic device, characterized in that, The system includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the DC system anti-misoperation judgment method as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the DC system anti-misoperation judgment method as described in any one of claims 1-5.

9. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the DC system anti-misoperation discrimination method as described in any one of claims 1-5.

Images