Abnormal monitoring method and device for full performance integrated test system based on electric energy meter

Abstract

The present invention discloses a method and device for monitoring abnormalities of a full-performance integrated test system based on an electric energy meter. The method for monitoring abnormalities of a full-performance integrated test system of the present invention comprises: constructing a historical state matrix, recording past state changes, using time attenuation and disturbance correction, comparing the current state with the historical state, calculating the difference function, judging whether there is an abnormality, and analyzing the abnormal trend through a backtracking mechanism to confirm the continuity of the abnormality; after confirming that the abnormality has continuity, using multivariable collaborative reasoning to perform in-depth analysis, capturing the complex interactions between variables through nonlinear convolution, constructing a comprehensive state representation, and combining the current state, determining the root cause of the abnormality through a collaborative reasoning function, designing a feedback abnormality processing algorithm, calculating the optimal recovery path, and gradually adjusting the state to restore the system to normal. The present invention can not only detect and reason about abnormalities, but also adaptively adjust the current state, gradually repair the abnormality and return to a normal state.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent industrial automation testing, and in particular to an abnormality monitoring method and device for a full-performance integrated test system based on an electric energy meter. Background Art

[0002] With the development of modern industrial automation and intelligence, full-performance integrated test systems have been widely used in various complex test scenarios. Such systems integrate multiple subsystems and equipment and can perform multiple tasks at the same time, such as environmental monitoring, mechanical equipment operation, test device operation, etc., with extremely high test efficiency and versatility. As the test scenarios become more complex, the large amount of data dimensions that the system needs to process also increases, such as temperature and humidity, mechanical equipment status, test equipment output, software operation status, etc. These data are not only diverse, but also highly dynamic in space and time.

[0003] The widespread application of full-performance integrated test systems has improved the automation level of industrial testing, but it has also brought new challenges. Due to the complexity and versatility of the system, the system may have various types of anomalies during long-term operation, such as unstable operation of mechanical equipment, abnormal output of test equipment, influence of environmental factors, software operation errors, etc. These abnormal phenomena may not only lead to inaccurate test results, but may even cause equipment damage, test interruption or other more serious consequences. Therefore, timely and accurate detection and processing of these anomalies is the key to ensuring stable operation of the system.

[0004] The existing technology has at least the following technical problems: it is impossible to dynamically adjust the detection standard according to the change rate of the full-performance integrated test system, ignores the interrelationships and linkage effects between different variables, lacks consideration of recovery costs during exception handling, resulting in low exception handling efficiency and inability to effectively expand to handle more exception detection needs. The full-performance integrated test system has poor flexibility and adaptability, which limits the ability to accurately detect, infer, judge and repair anomalies, making it impossible to effectively guarantee the safety and reliability of the full-performance integrated test system. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the defects of the above-mentioned prior art and provide an abnormality monitoring method and device for a full-performance integrated test system based on an electric energy meter to improve the accuracy of abnormality detection and avoid false alarms caused by accidental noise or instantaneous disturbances; by calculating the optimal path for abnormality recovery through feedback of the abnormality processing algorithm, the full-performance integrated test system can adaptively adjust the current state, gradually repair the abnormality and return to normal state.

[0006] To this end, the present invention adopts the following technical solution.

[0007] In a first aspect, the present invention provides a method for monitoring abnormality of a full performance integrated test system based on an electric energy meter, which comprises the steps of:

[0008] S1. Construct a historical state matrix to record past state changes, use time decay and disturbance correction to compare the current state with the historical state, calculate the difference function, determine whether there is an anomaly, and analyze the anomaly trend through the backtracking mechanism to confirm the continuity of the anomaly;

[0009] S2. After confirming that the anomaly is continuous, use multivariate collaborative reasoning to perform in-depth analysis, capture the complex interactions between variables through nonlinear convolution, build a comprehensive state representation, and combine the current state to determine the root cause of the anomaly through collaborative reasoning functions, design a feedback anomaly handling algorithm, calculate the optimal recovery path, and gradually adjust the state to restore the system to normal.

[0010] Furthermore, in S1, a difference function is introduced to compare the difference between the current state of the full-performance integrated test system and the historical state matrix. Assuming that the full-performance integrated test system has been in a normal state, the difference between the current state of the full-performance integrated test system and the historical state will not exceed a preset threshold; if the current state of the full-performance integrated test system deviates from the historical state by more than a preset threshold, it means that an abnormality has occurred in the full-performance integrated test system.

[0011] Furthermore, in S1, when the difference function exceeds the set threshold, the backtracking mechanism is triggered to analyze the state changes and trends between multiple moments, try to trace the possible origin of the anomaly, and determine whether the anomaly is continuous; the concept of the backtracking window is introduced to analyze the changing trend of the state in the backtracking window.

[0012] Furthermore, in S1, the consistency coefficient of state change is defined to measure the continuity of the state change of the full performance integrated test system in the entire backtracking window; if the state change trend of the full performance integrated test system in the backtracking window is consistent, that is, the state continues to rise or fall, it represents a gradually accumulated abnormality, which is judged by calculating the cumulative trend of the state change.

[0013] Furthermore, in S1, the continuity of the anomaly is judged by combining the consistency coefficient and the trend analysis result. If the consistency coefficient exceeds the set standard and the trend analysis result shows that the full performance integration test state continues to change in a certain direction, the existence of the anomaly is confirmed and the anomaly is marked as a "continuity anomaly".

[0014] Furthermore, in S2, after determining that the anomaly is persistent, multivariate collaborative reasoning is performed; nonlinear convolution operations are used to capture the interactive relationship between different variables, and a comprehensive state representation is obtained through convolution operations to discover the abnormal linkage effects hidden in different variables, especially when there are complex nonlinear relationships between the variables.

[0015] Furthermore, in S2, in order to further improve the accuracy of abnormal reasoning, the comprehensive state representation is combined with the system state, and a collaborative reasoning function is used to make the final judgment, ensuring that the system can consider the complex relationship between multiple variables during the reasoning process while maintaining sensitivity to the original data.

[0016] Furthermore, in S2, if the bi-norm of the collaborative reasoning result exceeds the preset abnormal threshold, the system will record the abnormal flag and enter the abnormality handling stage; when the abnormality is confirmed, the system must take effective feedback measures to prevent the abnormality from further deteriorating. Therefore, the present invention designs a feedback abnormality handling algorithm to ensure that the system is restored to a normal state at the minimum cost by calculating the optimal path for abnormal recovery.

[0017] Furthermore, in S2, a smooth and efficient recovery path is found by feedback of the exception handling algorithm, and finally the self-repair of the system is achieved by adjusting the current system state, and the current state is gradually adjusted according to the feedback path.

[0018] In a second aspect, the present invention provides a full performance integrated test system abnormality monitoring device based on an electric energy meter, which is used to implement the above-mentioned full performance integrated test system abnormality monitoring method, and comprises the steps of:

[0019] Abnormal judgment unit: constructs a historical state matrix, records past state changes, uses time decay and disturbance correction to compare the current state with the historical state, calculates the difference function, determines whether there is an abnormality, and analyzes the abnormal trend through the backtracking mechanism to confirm the continuity of the abnormality;

[0020] Abnormal recovery unit: After confirming that the abnormality is continuous, it uses multivariable collaborative reasoning to perform in-depth analysis, captures the complex interactions between variables through nonlinear convolution, builds a comprehensive state representation, and combines the current state to determine the root cause of the abnormality through collaborative reasoning functions, designs feedback abnormality handling algorithms, calculates the optimal recovery path, and gradually adjusts the state to restore the system to normal.

[0021] The present invention has the following beneficial effects:

[0022] 1. The present invention can effectively capture the difference between the current and historical data of the full-performance integrated test system; compared with the anomaly detection method of single-point data, the retrospective analysis of historical states helps to improve the accuracy of anomaly detection and avoid false alarms caused by accidental noise or instantaneous disturbances.

[0023] 2. The present invention further analyzes historical status changes through a backtracking mechanism, which can not only capture the time point when the anomaly occurs, but also determine the trend change of the anomaly, providing more comprehensive information for system maintenance and repair.

[0024] 3. The present invention captures the complex interactive relationship between multiple variables through nonlinear convolution operations, and can discover the abnormal linkage effects hidden in different variables; the collaborative reasoning function combines the comprehensive state and the original state of multivariate data, thereby improving the accuracy of abnormal reasoning.

[0025] 4 The present invention can not only detect and infer anomalies, but also calculate the optimal path for anomaly recovery by feedback of the anomaly handling algorithm. The full-performance integrated test system can adaptively adjust the current state, gradually repair the anomaly and return to the normal state. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of the abnormality monitoring method of the full performance integrated test system based on the electric energy meter of the present invention;

[0027] Figure 2 It is a structural diagram of the abnormality monitoring device of the full performance integrated test system based on the electric energy meter of the present invention. DETAILED DESCRIPTION

[0028] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0030] The specific technical solution for abnormal monitoring of a full-performance integrated test system based on an electric energy meter provided by the present invention is described in detail below in conjunction with the accompanying drawings.

[0031] See attached Figure 1, which shows a flow chart of an abnormality monitoring method for a full performance integrated test system based on an electric energy meter provided by the present invention, the method comprising the following steps:

[0032] S1. Construct a historical state matrix to record past state changes, use time attenuation and disturbance correction to compare the current state with the historical state, calculate the difference function, determine whether there is an anomaly, and analyze the abnormal trend through the backtracking mechanism to confirm the continuity of the anomaly.

[0033] In the abnormal monitoring process of the full-performance integrated test system based on the electric energy meter, data collection is the basis for the construction of the historical state matrix, because the core function of the historical state matrix is ​​to calculate the state changes of the full-performance integrated test system at past moments by tracing back the collected multi-dimensional data, and use it for comparative analysis with the current state to determine whether an abnormality occurs.

[0034] Construct a historical state matrix to record the state of the full performance integrated test system at multiple moments in the past, ensuring that past change trends can be effectively referenced when making judgments at the current moment. The data that constitute the historical state matrix includes the following parts:

[0035] In the monitoring of mechanical structures, photoelectric sensors, proximity switches and other devices are used to obtain the position information, action status and feedback signals of mechanical equipment in real time during operation. Taking the robot grabbing the electric energy meter as an example, monitor whether each action of the robot is completed within the specified time. If the robot does not reach the specified position on time, the full performance integrated test system will store this abnormal mark in the data stream and use it together with the normal time series data as part of the subsequent analysis. At the same time, the operating status of the mechanical equipment (such as action time, position change, etc.) will be stored in the system in the form of a continuous time series for the construction of the historical state matrix.

[0036] For the operation data of each device that tests the electric energy meter, the serial port server collects the status information and output data of the signal source, standard meter, withstand voltage instrument and other equipment. For example, the withstand voltage instrument may continuously update its voltage and current data during the test of the electric energy meter. By collecting and storing the dynamic data of the test equipment in real time, it is ensured that the historical information in the historical status matrix can accurately reflect the status of the equipment at each time point.

[0037] Software operation status data is also an important part of the historical status matrix, which obtains the log information of the scheduling system, the connection status of the database, the execution status of the processing flow, etc. from the software level. For example, database problems or data format errors in software operation will also be recorded as time series information; when the full performance integration test system enters the retrospective analysis stage, the software abnormality information is analyzed together with other equipment status data as the basis for deducing the cause of the abnormality.

[0038] Use historical status to help determine current anomalies and avoid relying on data at a single moment. Define the historical status matrix for:

[0039]

[0040] in, Indicates that the full performance integrated test system is The state vector at time, is a time index, indicating the past time index from the current time t. It’s time; Indicates the length of the historical time period considered, that is, the duration range of the historical state; is the time decay factor, It is a regulating parameter used to control the impact of past time data on the current state; is a random disturbance term. By constructing a historical state matrix, the collective wisdom of historical data can be used in current abnormal judgments, rather than a single data point.

[0041] After the calculation of the historical state matrix is ​​completed, in order to detect whether there is an abnormality at the current moment, it is necessary to compare the current state of the full performance integrated test system With the historical state matrix If the system is always in a normal state, the difference between the current data change and the historical state will not exceed the preset threshold; if the current state deviates from the historical state by more than the preset threshold, it means that the system has become abnormal. Therefore, the difference function is introduced. :

[0042]

[0043] in, Represents the Euclidean distance between the current state and the historical state matrix, measuring the difference between the current state and the historical state; in order to prevent inconsistent standards caused by different historical states, the standard deviation is introduced The differences were standardized to eliminate the effects of the numerical scale; It is a regulating parameter that controls the weight of the state change rate part in the difference degree; the integral term The rate of change of the system in the past period of time is further calculated, representing the dynamic characteristics of the state. is the time parameter of the integration. The change rate is introduced as an additional judgment basis to ensure that the potential risks brought by rapid changes can be captured.

[0044] The difference function is used as a preliminary judgment tool for anomaly detection. When the difference function exceeds the preset threshold, the full performance integrated test system will consider that there is a potential anomaly. The difference function exceeding the threshold may be caused by a variety of reasons, such as noise, instantaneous fluctuations of the data collector, short-term system disturbances, etc., and it is not necessarily a true anomaly every time. Therefore, after the difference exceeds the threshold, an "abnormal" judgment will not be made immediately, but a backtracking mechanism will be triggered to understand whether the full performance integrated test system has continuous abnormal signs at multiple time points through backtracking analysis.

[0045] When the difference function exceeds the set threshold When the backtracking mechanism is triggered, the state changes and trends between multiple moments are analyzed, and the possible origin of the anomaly is tried to be traced back and whether the anomaly is continuous. The concept of the backtracking window is introduced. Assume that the backtracking window is , backtracking time steps (such as the first 10 seconds, the first 100 sampling points, etc.), analyze The changing trend of the state in a time step. The specific backtracking process is as follows:

[0046] First, define the lookback window The state change function within , used to record the incremental change of state at each time step:

[0047]

[0048] The above formula indicates that from time To the current time , the incremental change of the system state. By calculating the incremental change, the fluctuation of the system state within the retrospective time window can be observed.

[0049] Define the consistency coefficient of state change , which is used to measure the continuity of system state changes in the entire backtracking window. The consistency coefficient is expressed by calculating the mean square error of the state increment change:

[0050]

[0051] The above formula is used to evaluate the fluctuation range of the system state within the retrospective time window. If it exceeds the set standard, it means that the system's state changes have been unstable over the past period of time and there are abnormal trends; otherwise, it means that the state is relatively stable and there are only short-term fluctuations.

[0052] If the full performance integrated test system is within the traceback window The state change trend is consistent, that is, the state continues to rise or fall, which represents a gradually accumulated abnormality. It can be judged by calculating the cumulative trend of state changes:

[0053]

[0054] in, Indicates the trend analysis results within the lookback window. is the sign function of the state increment, which is used to indicate whether the state is rising or falling. Continued accumulation in a certain direction indicates that the abnormality is gradually worsening or accumulating.

[0055] By combining the consistency coefficient and trend analysis results To determine the continuity of the anomaly, if the consistency coefficient Exceeding the set standard and trend analysis results If it indicates that the state changes continuously in a certain direction, the existence of the anomaly is confirmed and marked as a "continuous anomaly".

[0056] S2. After confirming that the anomaly is continuous, use multivariate collaborative reasoning to perform in-depth analysis, capture the complex interactions between variables through nonlinear convolution, build a comprehensive state representation, and combine the current state to determine the root cause of the anomaly through collaborative reasoning functions, design a feedback anomaly handling algorithm, calculate the optimal recovery path, and gradually adjust the state to restore the system to normal.

[0057] After determining that the anomaly is persistent, it is still necessary to further analyze the complex relationships between multiple variables in order to gain a deeper understanding of the nature and potential causes of the anomaly and conduct multivariate collaborative reasoning. The introduction of multivariate collaborative reasoning does not involve repeated judgments on anomalies, but rather a multi-dimensional and refined analysis of confirmed anomalies.

[0058] Assuming that the various variables in the full-performance integrated test system (such as ambient temperature and humidity, mechanical status, equipment feedback, etc.) are interrelated, the abnormality of a single variable may be related to changes in other variables. By reasoning about the interactive relationship between multiple variables, the real cause behind the abnormality can be found.

[0059] Use nonlinear convolution operations to capture the interactive relationship between different variables. Through convolution operations, we can discover the abnormal linkage effects hidden in different variables, especially when there are complex nonlinear relationships between variables. Specifically, the process of collaborative reasoning is as follows:

[0060] First, a comprehensive state representation is obtained by convolution operation on all variables , the comprehensive state represents the integration of the mutual influence relationship of multiple variables, and the formula is as follows:

[0061]

[0062] in, and Respectively indicate the system status the i-th and j-th variables in (e.g., temperature and humidity); is the number of variables of the system state; is the convolution kernel function of the i-th and j-th variables, used to couple the interaction between variables; It is a nonlinear activation function, and common functions such as Sigmoid or ReLU are selected. The purpose of convolution is to extract long-term dependencies between variables in order to identify whether anomalies are related to the linkage of multiple variables.

[0063] In order to further improve the accuracy of abnormal reasoning, the comprehensive state representation System status Combined with the collaborative reasoning function, the final judgment is made to ensure that the system can consider the complex relationship between multiple variables during the reasoning process while maintaining sensitivity to the original data. The formula of the collaborative reasoning function is:

[0064]

[0065] in, is the collaborative reasoning function; and is a weight matrix, which is used to adjust the contribution of the comprehensive state representation and the system state to the final reasoning result; is a nonlinear mapping function; the logarithmic term The bi-norm of the system state is introduced to ensure that the system has good control over the scale of the data. The purpose of collaborative reasoning is not to determine whether there is an anomaly again, but to determine the specific nature and severity of the anomaly by deeply analyzing the relationship between multiple variables, more accurately locate the root cause of the anomaly, and then formulate targeted processing strategies.

[0066] If the collaborative reasoning results The bi-norm of exceeds the preset abnormal threshold , the system will record the abnormal flag , and enter the exception handling stage. The recording formula of the exception flag is:

[0067]

[0068] When an anomaly is confirmed, effective feedback measures must be taken to prevent the anomaly from further deteriorating. Therefore, a feedback anomaly handling algorithm is designed to ensure that the full performance integrated test system is restored to normal at the minimum cost by calculating the optimal path for anomaly recovery. The feedback function is defined as:

[0069]

[0070] in, is a feedback function, which indicates the recovery path after an anomaly; It is the operation of solving the minimum value, which means finding the state path that can minimize the objective function ; It is an integral operation used to calculate the time from arrive Integrate the state change of is the cost function of abnormal recovery, which indicates the cost of the recovery process; the gradient term Control the smoothness of the recovery path to prevent drastic fluctuations in the system state during the recovery process; It is an adjustment parameter that controls the smoothness of the recovery path.

[0071] By feeding back the exception handling algorithm, a smooth and efficient recovery path is found, and finally the current system status is adjusted To achieve self-repair of the system, the current state is gradually adjusted according to the feedback path. The adjustment formula is:

[0072]

[0073] in, is the adjusted system state, is the step size parameter, which determines the amplitude of each state adjustment. Through gradual feedback adjustment, the system can gradually return to the normal state and complete abnormal self-repair.

[0074] In summary, the abnormality monitoring method of the full performance integrated test system based on the electric energy meter was completed.

[0075] The experimental process of the abnormal monitoring method of the full performance integrated test system of the present invention is as follows:

[0076] A: Historical state matrix construction

[0077] In the full-performance integrated test system, multi-dimensional data is first collected in real time from the mechanical structure, test equipment, and software operation to establish a historical state matrix.

[0078] Experimental scenario:

[0079] The mechanical equipment is an industrial robot. The test device includes a signal source, a withstand voltage tester and a standard meter. The software layer monitors the operation log of the scheduling system. The collected data is as follows (taking 10 time points as an example):

[0080]

[0081] Based on these data, the historical state matrix is ​​constructed using the formula Assuming the time decay factor :

[0082]

[0083] Pick ,and As the random disturbance term, we get:

[0084]

[0085] Assuming that the system assigns equal weight to each variable, the historical state matrix is ​​calculated:

[0086] Robot position information , ,...,

[0087] Withstand voltage ,...,

[0088] Calculate the value of the historical state matrix manipulator position information:

[0089]

[0090] The historical status matrix of the voltage of the withstand voltage tester:

[0091]

[0092] B: Difference function calculation

[0093] Compare current status With the historical state matrix The difference between .

[0094]

[0095] Assume that the current position of the robot is 80mm, the voltage of the withstand voltage tester is 2450V, and the standard deviation is Adjust the parameters to 15 (manipulator position) and 250 (voltage of the withstand voltage tester). . Calculate the difference between the manipulator and the pressure tester respectively:

[0096] Robot position differences:

[0097]

[0098] Voltage difference of withstand voltage tester:

[0099]

[0100] The change rate is calculated by numerically differentiating the acquired time series data, assuming that the change rate is 2 in 10 time steps (in units of mm / s and V / s). Therefore:

[0101]

[0102]

[0103] If the difference function value exceeds the preset threshold (such as 10), the system will determine that there is an abnormality and trigger the backtracking mechanism.

[0104] C: Backtracking analysis

[0105] When the difference function exceeds the threshold, the backtracking analysis is triggered. Assuming the backtracking window For 5 time steps, calculate the incremental change:

[0106]

[0107] For example, the incremental changes in the robot position data are as follows:

[0108]

[0109]

[0110]

[0111]

[0112]

[0113] Calculate the coefficient of consistency , which is used to evaluate the fluctuation range of the system state within the retrospective time window:

[0114]

[0115] First calculate the average incremental change:

[0116]

[0117] Then calculate the mean square error:

[0118]

[0119] like If it exceeds the set standard (for example, 1.5), it indicates that the system state changes are unstable and there is an abnormal trend.

[0120] D: Feedback exception handling

[0121] After confirming the anomaly, enter the feedback anomaly handling stage. Use the feedback anomaly handling formula to calculate the optimal path for system recovery. :

[0122]

[0123] Assume that the recovery cost function Degree of deviation from the state is proportional to, and the gradient term Indicates the smoothness of the recovery path. Select , and assume that the initial moment is 0. The optimization result of the recovery path is calculated as:

[0124] (robot position adjustment speed) and (Voltage adjustment speed of withstand voltage tester)

[0125] According to the feedback path, the system state is gradually adjusted, and the adjustment formula is:

[0126]

[0127] Set step length ,get:

[0128] Robot position adjustment:

[0129] Withstand voltage tester voltage adjustment:

[0130] The full-performance integration test gradually returned to normal and completed the abnormal self-repair.

[0131] Through complete experimental data calculation and process display, the abnormality monitoring method of the full-performance integrated test system based on the electric energy meter of the present invention has obtained clear numerical results in multiple steps such as data acquisition, historical state matrix construction, difference function calculation, backtracking analysis and feedback abnormality processing.

[0132] The above technical solution has the following beneficial effects:

[0133] 1. By constructing a historical state matrix and using a difference function for anomaly detection, the present invention can effectively capture the difference between the current moment and the historical data of the full-performance integrated test system; compared with the anomaly detection method of single-point data, the historical state matrix combines information from multiple time dimensions, and through retrospective analysis of historical states, it helps to improve the accuracy of anomaly detection and avoid false alarms caused by accidental noise or instantaneous disturbances; in addition, the introduction of the state change rate in the difference function ensures that the system can identify potential rapid change trends, providing additional judgment basis for anomaly judgment.

[0134] 2. The present invention does not immediately determine it as an abnormality when the difference function exceeds the threshold, but further analyzes the historical state changes through the backtracking mechanism; the state change function and consistency coefficient in the backtracking window are used to monitor the incremental changes of the system state at multiple times, and the persistence and severity of the abnormality are judged by calculating the state trend and fluctuation amplitude; this mechanism can not only capture the time point when the abnormality occurs, but also judge the trend change of the abnormality, providing more comprehensive information for system maintenance and repair.

[0135] 3. The various variables in the system (such as ambient temperature and humidity, mechanical status, etc.) do not work independently, but are interrelated; by capturing the complex interactive relationship between multiple variables through nonlinear convolution operations, the present invention can discover the abnormal linkage effects hidden in different variables; the collaborative reasoning function combines the comprehensive state and the original state of multivariate data, thereby improving the accuracy of abnormal reasoning; this analysis method based on multi-dimensional reasoning can more deeply understand the nature of the anomaly and accurately locate the root cause of the anomaly.

[0136] 4. The present invention can not only detect and infer anomalies, but also calculate the optimal path for anomaly recovery through feedback of the anomaly handling algorithm. Through the feedback function, the system can adaptively adjust the current state, gradually repair the anomaly and return to normal state; this adaptive repair mechanism ensures that the system can resume work in a timely and smooth manner after the anomaly occurs, avoiding the continued expansion of the anomaly or causing more serious system failures; compared with traditional exception handling methods, the present invention greatly improves the system's autonomous repair capability and reduces the need for manual intervention.

[0137] See attached Figure 2 , which shows an architecture diagram of a full performance integrated test device based on an electric energy meter of the present invention, which is used to implement the above-mentioned full performance integrated test method, and includes:

[0138] Abnormal judgment unit: constructs a historical state matrix, records past state changes, uses time decay and disturbance correction to compare the current state with the historical state, calculates the difference function, determines whether there is an abnormality, and analyzes the abnormal trend through the backtracking mechanism to confirm the continuity of the abnormality;

[0139] Abnormal recovery unit: After confirming that the abnormality is continuous, it uses multivariable collaborative reasoning to perform in-depth analysis, captures the complex interactions between variables through nonlinear convolution, builds a comprehensive state representation, and combines the current state to determine the root cause of the abnormality through collaborative reasoning functions, designs feedback abnormality handling algorithms, calculates the optimal recovery path, and gradually adjusts the state to restore the system to normal.

[0140] It should be noted that each unit in the above-mentioned full-performance integrated test device based on an electric energy meter can be fully or partially implemented by software, hardware and a combination thereof. The above-mentioned units can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules. For the specific definition of a full-performance integrated test device based on an electric energy meter, please refer to the definition of a full-performance integrated test method based on an electric energy meter above. The two have the same functions and effects, which will not be repeated here.

[0141] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such 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 invention, and should be included in the protection scope of the present invention.

Claims

1. A method for abnormal monitoring of a full performance integrated test system based on an electric energy meter, characterized in that: Includes steps: S1. Construct a historical state matrix to record past state changes, use time decay and disturbance correction to compare the current state with the historical state, calculate the difference function, determine whether there is an anomaly, and analyze the anomaly trend through the backtracking mechanism to confirm the continuity of the anomaly; S2. After confirming that the anomaly is continuous, use multivariate collaborative reasoning to perform in-depth analysis, capture the complex interactions between variables through nonlinear convolution, build a comprehensive state representation, and combine the current state to determine the root cause of the anomaly through collaborative reasoning functions, design a feedback anomaly processing algorithm, calculate the optimal recovery path, and gradually adjust the state to restore the system to normal; In S2, the comprehensive state representation is combined with the system state, and a collaborative reasoning function is used to make a final judgment. The formula of the collaborative reasoning function is: in, is the collaborative reasoning function; and is a weight matrix, which is used to adjust the comprehensive state representation System status Contribution to the final reasoning result; is a nonlinear mapping function; The feedback exception handling algorithm uses the following feedback function: in, is a feedback function, which indicates the recovery path after an anomaly; It is the operation of solving the minimum value, which means finding the state path that can minimize the objective function ; It is an integral operation used to calculate the time from arrive Integrate the state change of is the cost function of abnormal recovery, which indicates the cost of the recovery process; the gradient term Control the smoothness of the recovery path to prevent drastic fluctuations in the system state during the recovery process; It is an adjustment parameter that controls the smoothness of the recovery path.

2. The method for abnormality monitoring of a full performance integrated test system according to claim 1, characterized in that: In S1, a difference function is introduced to compare the difference between the current state of the full-performance integrated test system and the historical state matrix. Assuming that the full-performance integrated test system has been in a normal state, the difference between the current state of the full-performance integrated test system and the historical state will not exceed a preset threshold; if the current state of the full-performance integrated test system deviates from the historical state by more than a preset threshold, it means that an abnormality has occurred in the full-performance integrated test system.

3. The method for abnormality monitoring of a full performance integrated test system according to claim 2, characterized in that: In S1, when the difference function exceeds the set threshold, the backtracking mechanism is triggered to analyze the state changes and trends between multiple moments, try to trace the possible origin of the anomaly, and determine whether the anomaly is continuous; the concept of the backtracking window is introduced to analyze the changing trend of the state in the backtracking window.

4. The method for abnormality monitoring of a full performance integrated test system according to claim 3, characterized in that: In S1, the consistency coefficient of state change is defined to measure the continuity of the state change of the full performance integrated test system in the entire backtracking window; if the state change trend of the full performance integrated test system in the backtracking window is consistent, that is, the state continues to rise or fall, it represents a gradually accumulated abnormality, which is judged by calculating the cumulative trend of the state change.

5. The method for monitoring abnormality of a full performance integrated test system according to claim 4, characterized in that: In S1, the continuity of the anomaly is judged by combining the consistency coefficient and the trend analysis result. If the consistency coefficient exceeds the set standard and the trend analysis result shows that the full performance integration test state continues to change in a certain direction, the existence of the anomaly is confirmed and the anomaly is marked as a "continuity anomaly".

6. The method for abnormality monitoring of a full performance integrated test system according to claim 1, characterized in that: In S2, after determining that the anomaly is persistent, multivariate collaborative reasoning is performed; nonlinear convolution operations are used to capture the interactive relationship between different variables. Through the convolution operation, a comprehensive state representation is obtained to discover the abnormal linkage effects hidden in different variables, especially when there are complex nonlinear relationships between the variables.

7. The method for abnormality monitoring of a full performance integrated test system according to claim 1, characterized in that: In S2, if the binary norm of the collaborative reasoning result exceeds the preset abnormal threshold, the system will record the abnormal flag and enter the abnormality handling stage; when the abnormality is confirmed, the system must take effective feedback measures to prevent the abnormality from further deteriorating.

8. The method for abnormality monitoring of a full performance integrated test system according to claim 7, characterized in that: In S2, a smooth and efficient recovery path is found by feeding back the exception handling algorithm, and finally the self-repair of the system is achieved by adjusting the current system state, and the current state is gradually adjusted according to the feedback path.

9. An abnormality monitoring device for a full performance integrated test system based on an electric energy meter, used to implement the abnormality monitoring method for a full performance integrated test system according to any one of claims 1 to 8, characterized in that: Includes steps: Abnormal judgment unit: constructs a historical state matrix, records past state changes, uses time decay and disturbance correction to compare the current state with the historical state, calculates the difference function, determines whether there is an abnormality, and analyzes the abnormal trend through the backtracking mechanism to confirm the continuity of the abnormality; Abnormal recovery unit: After confirming that the abnormality is continuous, it uses multivariable collaborative reasoning to perform in-depth analysis, captures the complex interactions between variables through nonlinear convolution, builds a comprehensive state representation, and combines the current state to determine the root cause of the abnormality through collaborative reasoning functions, designs feedback abnormality handling algorithms, calculates the optimal recovery path, and gradually adjusts the state to restore the system to normal.

Images