State monitoring method and device, computer device, readable storage medium and program product

Abstract

The application relates to a state monitoring method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: obtaining environment data and original control data of a target object in a preset monitoring period; performing disturbance analysis on the original control data according to a disturbance prediction model to obtain disturbance compensation data; performing state prediction on the target object based on the disturbance prediction model, the disturbance compensation data and the original control data to obtain target performance data of the target object; and generating a target threshold interval according to a threshold optimization model, the environment data and the target performance data; and the target threshold interval is used for monitoring the state of the target object in a current monitoring period. The method can improve the accuracy of state monitoring.

Description

Technical Field

[0001] This application relates to the field of aero-engine technology, and in particular to a condition monitoring method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Technology

[0002] With the development of aero-engine technology, it is necessary to monitor the operating status of aero-turboshaft engines in order to ensure their safe operation.

[0003] In traditional technology, the terminal determines the parameter range of the aircraft turboshaft engine under normal operating conditions as a preset range threshold, and monitors the aircraft turboshaft engine according to the preset range threshold to determine whether there is any abnormality in the operating status of the aircraft turboshaft engine and obtain the status monitoring results.

[0004] However, in traditional technologies, because the preset interval thresholds are static and fixed, false alarms and missed alarms are prone to occur in the condition monitoring of aero-turboshaft engines, resulting in poor accuracy of the condition monitoring results. Summary of the Invention

[0005] Therefore, it is necessary to provide a status monitoring method, apparatus, computer equipment, computer-readable storage medium, and computer program product to address the aforementioned technical problems.

[0006] Firstly, this application provides a status monitoring method, including:

[0007] Within a preset monitoring period, acquire environmental data and raw control data of the target object;

[0008] The original control data is subjected to disturbance analysis based on the disturbance prediction model to obtain disturbance compensation data;

[0009] Based on the disturbance prediction model, the disturbance compensation data, and the original control data, the state of the target object is predicted to obtain the target performance data of the target object;

[0010] A target threshold range is generated based on the threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0011] In one embodiment, the step of performing disturbance analysis on the original control data according to the disturbance prediction model to obtain disturbance compensation data includes:

[0012] Based on the disturbance prediction model, the system output data of the target object at the current moment, the system state estimate, and the historical disturbance data, disturbance prediction is performed to obtain the disturbance prediction value at the next moment.

[0013] Based on the predicted disturbance value, a disturbance compensation amount is generated, and based on the disturbance compensation amount, the first performance index, and the first constraint condition, the original control data is subjected to disturbance optimization to obtain disturbance compensation data.

[0014] In one embodiment, the step of predicting the state of the target object based on the disturbance prediction model, the disturbance compensation data, and the original control data to obtain the target performance data of the target object includes:

[0015] Attack compensation is predicted based on the disturbance prediction model and the disturbance compensation data to obtain the attack compensation amount;

[0016] The original control data is compensated and optimized based on the disturbance prediction model, the disturbance compensation amount, and the attack compensation amount to obtain the target control data;

[0017] Based on the disturbance prediction model and the target control data, the state of the target object is predicted to obtain the target performance data.

[0018] In one embodiment, the threshold optimization model includes a deep learning structure and a statistical inference structure;

[0019] The step of generating the target threshold range based on the threshold optimization model, the environmental data, and the target performance data includes:

[0020] The target performance data and the environmental data are preprocessed to obtain preprocessed target performance data and preprocessed environmental data;

[0021] Based on the deep learning structure, feature extraction is performed on the preprocessed target performance data and the preprocessed environmental data to obtain an initial feature matrix, and the initial feature matrix is ​​then subjected to dimensionality reduction to obtain the target feature matrix.

[0022] The target feature matrix is ​​analyzed and processed according to the deep learning structure to obtain an initial threshold;

[0023] The initial threshold is constrained and optimized based on the statistical inference structure to obtain the target threshold range.

[0024] In one embodiment, the preprocessing of the target performance data and the environmental data to obtain preprocessed target performance data and preprocessed environmental data includes:

[0025] The target performance data and the environmental data are denoised using wavelet transform to obtain first target performance data and first environmental data.

[0026] The first target performance data and the first environment data are normalized to obtain the second target performance data and the second environment data.

[0027] The second target performance data and the second environmental data are interpolated and filled to obtain preprocessed target performance data and processed environmental data.

[0028] In one embodiment, the step of constraining and optimizing the initial threshold based on the statistical inference structure to obtain the target threshold interval includes:

[0029] The posterior distribution corresponding to the initial threshold is determined based on the statistical inference structure.

[0030] The initial threshold is constrained and updated based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

[0031] Secondly, this application also provides a status monitoring device, comprising:

[0032] The acquisition module is used to acquire environmental data and raw control data of the target object within a preset monitoring period;

[0033] The disturbance analysis module is used to perform disturbance analysis on the original control data according to the disturbance prediction model to obtain disturbance compensation data;

[0034] The state prediction module is used to predict the state of the target object based on the disturbance prediction model, the disturbance compensation data and the original control data, so as to obtain the target performance data of the target object;

[0035] The threshold generation module is used to generate a target threshold range based on the threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0036] In one embodiment, the disturbance analysis module is specifically used to perform disturbance prediction based on the disturbance prediction model, the system output data of the target object at the current moment, the system state estimate, and historical disturbance data, to obtain the disturbance prediction value at the next moment;

[0037] Based on the predicted disturbance value, a disturbance compensation amount is generated, and based on the disturbance compensation amount, the first performance index, and the first constraint condition, the original control data is subjected to disturbance optimization to obtain disturbance compensation data.

[0038] In one embodiment, the state prediction module is specifically used to perform attack compensation prediction based on the disturbance prediction model and the disturbance compensation data to obtain the attack compensation amount;

[0039] The original control data is compensated and optimized based on the disturbance prediction model, the disturbance compensation amount, and the attack compensation amount to obtain the target control data;

[0040] Based on the disturbance prediction model and the target control data, the state of the target object is predicted to obtain the target performance data.

[0041] In one embodiment, the threshold optimization model includes a deep learning structure and a statistical inference structure;

[0042] The threshold generation module is specifically used to preprocess the target performance data and the environmental data to obtain preprocessed target performance data and preprocessed environmental data.

[0043] Based on the deep learning structure, feature extraction is performed on the preprocessed target performance data and the preprocessed environmental data to obtain an initial feature matrix, and the initial feature matrix is ​​then subjected to dimensionality reduction to obtain the target feature matrix.

[0044] The target feature matrix is ​​analyzed and processed according to the deep learning structure to obtain an initial threshold;

[0045] The initial threshold is constrained and optimized based on the statistical inference structure to obtain the target threshold range.

[0046] In one embodiment, the threshold generation module is specifically used to perform denoising processing on the target performance data and the environmental data according to wavelet transform to obtain first target performance data and first environmental data;

[0047] The first target performance data and the first environment data are normalized to obtain the second target performance data and the second environment data.

[0048] The second target performance data and the second environmental data are interpolated and filled to obtain preprocessed target performance data and processed environmental data.

[0049] In one embodiment, the threshold generation module is specifically used to determine the posterior distribution corresponding to the initial threshold based on the statistical inference structure;

[0050] The initial threshold is constrained and updated based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

[0051] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0052] Within a preset monitoring period, acquire environmental data and raw control data of the target object;

[0053] The original control data is subjected to disturbance analysis based on the disturbance prediction model to obtain disturbance compensation data;

[0054] Based on the disturbance prediction model, the disturbance compensation data, and the original control data, the state of the target object is predicted to obtain the target performance data of the target object;

[0055] A target threshold range is generated based on the threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0056] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0057] Within a preset monitoring period, acquire environmental data and raw control data of the target object;

[0058] The original control data is subjected to disturbance analysis based on the disturbance prediction model to obtain disturbance compensation data;

[0059] Based on the disturbance prediction model, the disturbance compensation data, and the original control data, the state of the target object is predicted to obtain the target performance data of the target object;

[0060] A target threshold range is generated based on the threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0061] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0062] Within a preset monitoring period, acquire environmental data and raw control data of the target object;

[0063] The original control data is subjected to disturbance analysis based on the disturbance prediction model to obtain disturbance compensation data;

[0064] Based on the disturbance prediction model, the disturbance compensation data, and the original control data, the state of the target object is predicted to obtain the target performance data of the target object;

[0065] A target threshold range is generated based on the threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0066] The aforementioned status monitoring method, device, computer equipment, computer-readable storage medium, and computer program product acquire raw control data of the target object according to a preset period; perform disturbance analysis on the raw control data based on a disturbance prediction model to obtain disturbance compensation data; optimize the raw control data based on the disturbance compensation data to obtain target performance data; generate a target threshold range based on a threshold optimization model and the target performance data; and monitor the status of the target object based on the target threshold range and the real-time signal data of the target object. This method uses a disturbance prediction model to perform disturbance analysis on the raw control data and generates a target threshold range based on the disturbance-optimized target performance data using a threshold optimization model. This ensures that the target threshold range maintains high sensitivity under varying conditions. Furthermore, it determines a target threshold range that matches the current environment according to the disturbance-optimized target performance data and environmental data according to a preset period, avoiding false alarms and missed alarms caused by static, fixed preset thresholds that are incompatible with the target object's environment, thereby improving the accuracy of status monitoring results. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a flowchart illustrating a state monitoring method in one embodiment;

[0069] Figure 2 This is a schematic diagram of the process for determining disturbance compensation data in one embodiment;

[0070] Figure 3 This is a flowchart illustrating the process of predicting target performance data in one embodiment;

[0071] Figure 4 This is a schematic diagram of the process for generating a target threshold range in one embodiment;

[0072] Figure 5 This is a schematic diagram of the process for preprocessing target performance data and environmental data in one embodiment;

[0073] Figure 6This is a flowchart illustrating the process of generating a target threshold range based on a statistical inference structure in one embodiment.

[0074] Figure 7 This is a structural block diagram of a status monitoring device in one embodiment;

[0075] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0077] In one embodiment, such as Figure 1 As shown, a status monitoring method is provided. This embodiment illustrates the application of this method to a terminal, which is equipped with an aircraft engine safety protection control system. This system monitors the status of the aircraft engine. It is understood that this method can also be applied to a server, or to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0078] Step 102: Acquire environmental data and raw control data of the target object within the preset monitoring period.

[0079] The target object is an aircraft engine.

[0080] In this embodiment, the preset monitoring period can be set by the terminal according to different flight phases of the aircraft. For example, the preset monitoring period in the terminal is shorter during the takeoff and landing phases, which are characterized by higher risks, while it is longer during the cruise phase, when the aircraft's operating state is relatively stable. For instance, because the aircraft is in a highly dynamic environment during takeoff and landing, with unstable airflow and drastic changes in the external environment, the engine's operating state is easily disturbed, so the terminal can set the preset monitoring period to once every 50 milliseconds; during the cruise phase, the aircraft is in a relatively stable flight state with less environmental disturbance and a more stable engine operating state, so the terminal can set the preset monitoring period to once every 2 seconds.

[0081] During aircraft operation, taking a preset monitoring cycle as an example, the terminal first acquires environmental data and raw control data of the aircraft engine from the aircraft engine safety protection and control system. Environmental data may include, but is not limited to, atmospheric temperature, pressure, humidity, and flight altitude. Raw control data may include fuel flow, intake valve opening, exhaust valve opening, compressor inlet guide vane angle, turbine clearance control, etc. The terminal can acquire environmental data through the aircraft's onboard sensors and raw control data through sensors in the engine control unit. The acquired environmental data and raw control data are then stored in a cache for subsequent steps (disturbance analysis and state compensation, etc.). Furthermore, in each preset monitoring cycle, the terminal determines a target threshold range suitable for the current monitoring cycle based on the environmental data and raw control data, and performs state monitoring of the aircraft engine according to this target threshold range for the current monitoring cycle.

[0082] Step 104: Perform disturbance analysis on the original control data based on the disturbance prediction model to obtain disturbance compensation data.

[0083] In this embodiment of the application, the target object is an aero-turboshaft engine as an example. Aero-turboshaft engines are often subjected to various disturbances in complex flight environments, such as unstable airflow, temperature changes, and mechanical wear. These disturbances may cause the performance of the aero-turboshaft engine to deviate from the predetermined optimal operating state, increasing the risk of failure. Therefore, in adjusting the threshold for monitoring the state of the aero-turboshaft engine, it is necessary to predict, identify, and compensate for disturbances in order to reduce the interference and impact of uncertainties and external disturbances on the monitoring threshold of the aero-turboshaft engine, and ensure that the target threshold range can adapt to the changing environmental conditions and operating states.

[0084] For the prediction, identification, and compensation of the aforementioned disturbances, the aero-engine safety protection and control system includes a disturbance prediction model. First, the terminal inputs the raw control data into the disturbance prediction model. The model then performs disturbance analysis on the raw control data, predicting and identifying disturbances such as airflow instability and temperature changes that the aircraft may encounter in the future. Next, the terminal further analyzes the disturbances based on the disturbance prediction model and the prediction results. The disturbance prediction model maps the raw control data to disturbance compensation data, thus obtaining the disturbance compensation data.

[0085] In a specific embodiment, the disturbance prediction model is LSTM (Long Short-Term Memory) model as an example. During the training process, LSTM learns the long-term dependency relationship between the disturbance and the original control data in the training samples under a certain time series. Then, the terminal inputs the original control data into the LSTM model. The LSTM model predicts and identifies the disturbance at the next time step based on the long-term dependency relationship between the disturbance and the original control data, thereby obtaining the disturbance situation at the next time step. The LSTM model then further analyzes the disturbance situation and finally outputs the disturbance compensation data.

[0086] Step 106: Based on the disturbance prediction model, disturbance compensation data and original control data, perform state prediction on the target object to obtain the target performance data of the target object.

[0087] In this embodiment, the terminal applies disturbance compensation data to the original control data through a disturbance prediction model. This process compensates for the original control data by superimposing the disturbance compensation data onto the original control data. The resulting disturbance-compensated original control data serves as future control data that may be used to control the aero-engine at the next moment, thus determining the aero-engine's actual operating state under disturbance. Furthermore, the terminal predicts the aero-engine's actual operating state based on the disturbance prediction model and uses this actual operating state as target performance data for subsequent generation of target threshold ranges. The target performance data includes key performance indicators or key sensor data of the aero-engine, such as turbine temperature, compressor outlet pressure, and speed, as well as disturbance-compensated control data, such as fuel flow and air flow.

[0088] Step 108: Generate the target threshold range based on the threshold optimization model, environmental data, and target performance data.

[0089] The target threshold range is used to monitor the status of the target object within the current monitoring period.

[0090] In this embodiment of the application, the aero-engine safety protection control system also includes a threshold optimization model. The preset optimization model can dynamically adjust the safety protection threshold used for monitoring the state of the aero-engine based on changes in environmental data and target performance data during the current monitoring period, so that the aero-engine safety protection control system can more accurately respond to the actual operating environment and engine state for monitoring the state of the aero-engine.

[0091] After the terminal completes the prediction, identification and compensation of disturbances in the aero-engine safety protection and control system, the terminal inputs the target performance data and environmental data into the threshold optimization model. The threshold optimization model outputs the target threshold range adapted to the current monitoring period based on the current target performance data and environmental data of the aero-engine. Each monitoring indicator in the aero-engine safety protection and control system contains a target threshold range.

[0092] During the current monitoring cycle, after the aero-engine safety protection and control system obtains the target threshold range from the threshold optimization model, the terminal monitors the real-time status of the aero-engine based on the target threshold range. When a certain operating parameter in the real-time status of the aero-engine exceeds the corresponding target threshold range, an alarm message corresponding to that operating parameter is issued.

[0093] In the aforementioned state monitoring method, the original control data is subjected to disturbance analysis by a disturbance prediction model, and a target threshold range is generated based on the target performance data after disturbance optimization by a threshold optimization model. This allows the target threshold range to maintain high sensitivity under varying conditions. Then, according to a preset period, a target threshold range matching the current environment is determined based on the target performance data and environmental data after disturbance compensation optimization. This avoids false alarms and missed alarms in state monitoring caused by the incompatibility between static and fixed preset thresholds and the target object environment, thereby improving the accuracy of state monitoring results.

[0094] In one exemplary embodiment, such as Figure 2 As shown, step 104 includes steps 202 to 204. Wherein:

[0095] Step 202: Based on the disturbance prediction model, the original control data, the system output data of the target object at the current moment, the system state estimate, and the historical disturbance data, perform disturbance prediction to obtain the disturbance prediction value at the next moment.

[0096] In this embodiment, an LSTM-based disturbance prediction model is used as an example. The terminal uses historical data as sample data for the LSTM-based disturbance prediction model. The historical data consists of time-series aero-engine control data and system state estimates of the aero-engine. The terminal trains the LSTM-based disturbance prediction model using the historical data, adjusts the model parameters, and obtains a trained disturbance prediction model, thereby improving the model's ability to capture complex disturbances that the aero-engine may face.

[0097] When the disturbance prediction model performs disturbance prediction for the aero-engine, the terminal inputs the raw control data and the current system state estimate of the aero-engine into the disturbance prediction model. The input of the disturbance prediction model can be described as follows:

[0098] (1)

[0099] in, The system output data for the target object at the current moment; This refers to the control input, i.e., the raw control data; This is a system state estimate for an aero-engine. This is due to historical disturbances. Specifically, the system output data... For the external performance of an aircraft engine, these are real-time performance indicators that can be measured by sensors, including engine speed, exhaust temperature, etc.; system state estimates. The internal state of the aero-engine is an intermediate parameter in the process of the disturbance prediction model predicting disturbances to the aero-engine, including estimated values ​​of the current state such as combustion chamber pressure and turbine inlet temperature.

[0100] The terminal analyzes and processes the raw control data, the current system output data of the target object, the system state estimate, and historical disturbance data through a disturbance prediction model to predict the disturbance of the aero-engine. The output of the disturbance prediction model is then used as the predicted disturbance value for the next time step. ,Right now and the system state estimate at the next moment. .

[0101] Specifically, in the process of predicting disturbances to aero-engines using an LSTM-based disturbance prediction model, the terminal first sets the initial state of the LSTM-based disturbance prediction model to be disturbance-free, and then at each time step... In the process, the terminal performs the calculations for the forget gate, input gate, and output gate as follows:

[0102] (2)

[0103] in, For the calculation of the forget gate, For the calculation of the input gate, For the calculation of the output gate, For the calculation of candidate memory units, , , , These are the weight matrices used to calculate the linear combination of model inputs in the forget gate, input gate, output gate, and candidate memory units, respectively. , , , Let be the weight matrices used to calculate the hidden state (i.e., the perturbation prediction value) in the forget gate, input gate, output gate, and candidate memory unit, respectively. , , , These represent the bias terms in the forget gate, input gate, output gate, and candidate memory units, respectively. The weight matrix for the linear combination of the model inputs, the weight matrix for the hidden states, and the bias terms are adjusted during the training of the LSTM-based perturbation prediction model.

[0104] Furthermore, the updates to the memory unit state and hidden state are as follows:

[0105] (3)

[0106] (4)

[0107] in, The element-wise multiplication is used to control fine-tuning of the information flow in the LSTM network.

[0108] The terminal obtains the disturbance prediction value through the output of the output gate. (i.e., the predicted value of the disturbance at the next moment) After that, the disturbance prediction value can be further optimized to obtain the final disturbance prediction value. The optimization process is as follows:

[0109] (5)

[0110] (6)

[0111] in, It is a control performance indicator. It is a disturbance prediction model. The state of the aero-engine at the next moment is estimated by the disturbance prediction model. This represents the predicted value of the disturbance.

[0112] Step 204: Generate disturbance compensation amount based on disturbance prediction value, and perform disturbance optimization on original control data based on disturbance compensation amount, first performance index and first constraint condition to obtain disturbance compensation data.

[0113] In this embodiment, the terminal is based on the disturbance prediction value. and compensation gain matrix Generate disturbance compensation amount ,Right now:

[0114] (7)

[0115] Then, the terminal performs disturbance optimization on the original control data based on the disturbance compensation amount, combined with the first performance index and the first constraint condition, to obtain disturbance compensation data that completes the disturbance optimization of the original control data. The process is as follows:

[0116] (8)

[0117] (9)

[0118] (10)

[0119] (11)

[0120] (12)

[0121] in, As the primary performance indicator, , This is the first constraint condition. and These represent the lower and upper limits of the control input, respectively. For aircraft engines in time The reference state below, Here is the weight matrix for the state error. To control the input weight matrix, To predict the time domain, This is the model function for the disturbance prediction model, used to describe the system state estimate of the aero-engine. With the current system state estimate Disturbance compensation data and external disturbances The amount of change. It is a set of external disturbances. The weighted quadratic form representing the state deviation is used to measure the severity of the aero-engine's state deviation from the reference trajectory. This represents a weighted quadratic form of the control input, used to measure the magnitude or energy of the control input.

[0122] Optionally, the disturbance compensation data can not only be used to generate the target threshold range, but the terminal can also transmit the disturbance compensation data determined in the aero-engine safety protection control system to the aero-engine control system. The aero-engine control system then adjusts the control parameters, enabling the aero-engine to better resist disturbances, thereby enhancing its anti-disturbance capability and improving its control accuracy and system stability. Ultimately, by dynamically adjusting the control input, precise compensation and optimized control of aero-engine disturbances are achieved.

[0123] In this embodiment, an LSTM-based disturbance prediction model is used to predict disturbances in the aero-engine, effectively capturing complex dynamic behaviors in the time series and improving the accuracy of future disturbance predictions. Compensation amounts are generated based on the predicted disturbance values. These compensations are then combined with a first performance index and a first constraint condition to optimize the original control data, ensuring the control input is within a reasonable range while minimizing the error between the system state and the target state. This determines the disturbance compensation data for the aero-engine under disturbance conditions, which serves as the generation condition for the target threshold interval. This improves the adaptability of the target threshold interval to the environment of the aero-engine within the current monitoring period, thereby enhancing the accuracy of state monitoring.

[0124] In one exemplary embodiment, as the aircraft engine safety protection control system becomes more intelligent and networked, it may be threatened by cyberattacks or data tampering. Therefore, it is necessary to protect the aircraft engine safety protection control system against cyberattacks and abnormal behavior, such as... Figure 3 As shown, step 106 includes steps 302 to 306. Wherein:

[0125] Step 302: Based on the disturbance prediction model and disturbance compensation data, perform attack compensation prediction to obtain the attack compensation amount.

[0126] In this embodiment, to protect against threats of network attacks and data tampering, the terminal incorporates dynamic network attack detection capabilities. It detects potential network attacks on the current aero-engine safety protection control system by comparing the actual state data measured by sensors with the system state estimate output by a disturbance prediction model. The state estimate is predicted by the disturbance prediction model based on historical disturbances, control inputs, and the system state estimate from the previous moment. Specifically, the terminal analyzes the disturbance compensation data using the disturbance prediction model to obtain the system state estimate. Based on the system state estimate and the actual state data, it performs attack compensation prediction. First, the terminal determines the residual vector. :

[0127] (13)

[0128] in, It is the current estimate of the actual system state provided by the sensor. It is the system state estimate predicted by the disturbance prediction model at the previous time step. The system state estimate is obtained by the disturbance prediction model based on the system state estimate of the previous time step, the original control data of the previous time step, and the disturbance prediction of the previous time step.

[0129] Then, the terminal uses the residual vector Attack compensation amount can be obtained :

[0130] (14)

[0131] in, Estimate the gain matrix for the perturbation attack signal.

[0132] Step 304: Based on the disturbance prediction model, disturbance compensation amount and attack compensation amount, the original control data is compensated and optimized to obtain the target control data.

[0133] In this embodiment, the terminal determines the disturbance compensation amount according to the same principle as steps 202 to 204, and performs joint compensation optimization on the original control data based on the disturbance compensation amount and the attack compensation amount. The impact of network attacks and disturbances is superimposed on the original control data. That is, the original control data is optimized and controlled according to the first performance index and the first constraint condition of the superimposed attack compensation amount to obtain the target control data. ,in:

[0134] (15)

[0135] (16)

[0136] (17)

[0137] (18)

[0138] (19)

[0139] in, The primary performance indicator for superimposed attack compensation is... It is a weight parameter for attack compensation, used to balance the performance of the perturbation prediction model and the degree of compensation for the attack compensation value.

[0140] Step 306: Based on the disturbance prediction model and target control data, perform state prediction on the target object to obtain target performance data.

[0141] In this embodiment, the terminal uses a disturbance prediction model to predict the state of the aero-engine under target control data, and determines the target performance data of the aero-engine under the target control data. Since the target control data is obtained after disturbance compensation and attack compensation are completed, the target performance data can describe the performance state of the aero-engine after completing the counter-disturbance and attack interference. Under this performance state, the normal range of the aero-engine's operating parameters under different environmental conditions can be better determined.

[0142] In this embodiment, the original control data is compensated and optimized by a disturbance prediction model, disturbance compensation amount, and attack compensation amount to obtain target control data. The target performance data of the aero-engine under disturbance and attack can be further obtained based on the target control data. The target performance data can reflect the actual operating state of the aero-engine. The target threshold range can be determined based on the target performance data, which can improve the accuracy of the target threshold range and thus improve the accuracy of monitoring the state of the aero-engine through the target threshold range.

[0143] In one exemplary embodiment, the threshold optimization model includes a deep learning structure and a statistical inference structure; such as Figure 4 As shown, step 108 includes steps 402 to 408. Wherein:

[0144] Step 402: Preprocess the target performance data and environmental data to obtain preprocessed target performance data and preprocessed environmental data.

[0145] In this embodiment, the terminal preprocesses the collected target performance data (e.g., key performance indicators such as engine speed and temperature) and environmental data (e.g., atmospheric temperature, pressure, and humidity). The terminal removes noise and outliers from the target performance and environmental data through preprocessing, and normalizes the data to ensure the data quality and consistency processed by the threshold optimization model. Through preprocessing, the threshold optimization model can obtain more accurate and reliable target performance and environmental data, providing a foundation for subsequent feature extraction and threshold calculation.

[0146] Step 404: Based on the deep learning structure, feature extraction is performed on the preprocessed target performance data and preprocessed environment data to obtain an initial feature matrix. The initial feature matrix is ​​then dimensionality-reduced to obtain the target feature matrix.

[0147] In this embodiment, the terminal utilizes a deep learning structure (e.g., a neural network) to extract features from the preprocessed target performance data and environmental data. The deep learning structure can automatically learn complex features of the input data during training and generate an initial feature matrix. However, this initial feature matrix may contain high-dimensional data, making direct processing of the initial feature matrix computationally complex. Therefore, the terminal further employs a dimensionality reduction method (e.g., Principal Component Analysis, PCA) to reduce the dimensionality of the initial feature matrix, obtaining a low-dimensional target feature matrix. The dimensionality-reduced target feature matrix retains the main features of the original data while reducing computational complexity. The process of dimensionality reduction of the initial feature matrix using Principal Component Analysis is as follows:

[0148] (20)

[0149] in, It is the initial characteristic matrix. It is the eigenvector matrix. It is the feature matrix after dimensionality reduction.

[0150] Step 406: Analyze and process the target feature matrix according to the deep learning structure to obtain the initial threshold.

[0151] In this embodiment, the terminal analyzes and maps the target feature matrix based on the nonlinear relationship between the target performance data, environmental data, and the initial threshold learned during the training process of the deep learning structure, thereby determining the initial threshold corresponding to the current environmental data and target performance data. Specifically, the deep learning structure can be a Convolutional Neural Network (CNN). The terminal receives and processes the target performance data and environmental data through the input layer of the CNN, extracts local features through the convolutional kernels in the convolutional layer, and obtains the initial feature matrix. After obtaining the initial feature matrix, the terminal performs dimensionality reduction processing on the initial feature matrix using the PCA method to obtain the target feature matrix. Finally, the target feature matrix is ​​mapped to a high-dimensional space through the fully connected layer of the CNN, and the initial threshold is obtained through the output layer.

[0152] In one exemplary embodiment, the terminal first defines an initial threshold based on the preprocessed target performance data and the preprocessed environmental data. The function, that is:

[0153] (twenty one)

[0154] For environmental data, Atmospheric temperature, Atmospheric pressure, Atmospheric humidity, For flight altitude; for target performance data, For aircraft engine temperature, For aircraft engine pressure, For runtime, For fuel flow rate, This refers to airflow.

[0155] Step 408: Based on the statistical inference structure, the initial threshold is constrained and optimized to obtain the target threshold range.

[0156] In this embodiment, the statistical inference structure can be a Bayesian inference structure. Taking a Bayesian inference structure as an example, the terminal inputs an initial threshold into the Bayesian inference structure. The Bayesian inference structure constrains and optimizes the initial threshold, and combines the prior knowledge and likelihood function of the Bayesian inference structure to quantify the uncertainty of the initial threshold, ultimately generating a target threshold range. This target threshold range considers multiple factors (including false alarm rate, false negative rate, etc.) and performs trade-offs and optimizations to ensure the adaptability of the target threshold range to the current environmental data and target performance data. Finally, the target threshold range output by the pre-optimized model will be used to dynamically adjust the safety protection threshold of the engine operating parameters, thereby achieving more accurate state monitoring under different environmental and performance conditions.

[0157] In a specific embodiment, the target threshold range finally output by the threshold optimization model can be a multi-threshold selection. For example, in the monitoring of the state of excessive lubricating oil pressure, the minimum threshold of the target threshold range corresponding to the lubricating oil pressure is 0.16 MPa, and the duration is 1.5 seconds. That is, under any environmental data, as long as the duration of the lubricating oil pressure being lower than 0.16 MPa is greater than 1.5 seconds, the aircraft engine safety protection control system in the terminal will issue an alarm message that the lubricating oil pressure is too low.

[0158] The maximum threshold for the target threshold range corresponding to lubricating oil pressure can be selected by the terminal based on environmental data. For example, when the temperature index in the environmental data is greater than or equal to 60℃, the maximum threshold for the target threshold range corresponding to lubricating oil pressure is 0.7MPa, and the duration is set to 1.5 seconds; when the temperature index in the environmental data is less than 60℃, the maximum threshold for the target threshold range corresponding to lubricating oil pressure is 1.4MPa, and the duration is set to 1.5 seconds. Therefore, during lubricating oil pressure status monitoring, the aircraft engine safety protection control system will only issue an over-lubricating oil pressure alarm when the temperature index is less than 60℃ and the lubricating oil pressure is greater than 1.4MPa and the duration is greater than 1.5 seconds; or, when the temperature index is greater than or equal to 60℃ and the lubricating oil pressure is greater than 0.7MPa and the duration is greater than 1.5 seconds.

[0159] In this embodiment, firstly, preprocessing for denoising and normalization ensures the data quality and consistency of target performance and environmental data, providing a reliable foundation for subsequent steps. Then, deep learning is used to automatically extract complex features from high-dimensional data, and dimensionality reduction techniques are employed to obtain a low-dimensional target feature matrix, reducing computational complexity while preserving key information. Furthermore, a deep learning model predicts an initial threshold based on the target feature matrix, reflecting the optimal control point under the current environmental and performance conditions. Finally, the initial threshold is optimized using a statistical inference structure to generate a target threshold range, ensuring the system's safety and reliability under different environmental and performance conditions, improving the accuracy of the target threshold range within the current monitoring cycle, enhancing adaptability to complex dynamic environments, and ultimately achieving precise status monitoring of the aero-engine, thus improving the accuracy of status monitoring.

[0160] In one exemplary embodiment, such as Figure 5 As shown, step 402 includes steps 502 to 506. Wherein:

[0161] Step 502: Denoise the target performance data and environmental data according to wavelet transform to obtain the first target performance data and the first environmental data.

[0162] In this embodiment, the terminal performs denoising processing on the target performance data and environmental data using wavelet transform to obtain first target performance data and first environmental data. Specifically, the wavelet transform processing procedure is as follows:

[0163] (twenty two)

[0164] in, These are wavelet basis functions. These are wavelet coefficients. These are integer variables used to index the wavelet basis functions and wavelet coefficients, respectively. These parameters respectively characterize the number of wavelet transform layers and the resolution of local features.

[0165] Step 504: Normalize the first target performance data and the first environment data to obtain the second target performance data and the second environment data.

[0166] In this embodiment, the terminal normalizes the first target performance data and the first environment data to obtain the second target performance data and the second environment data. Specifically, the normalization process is as follows:

[0167] (twenty three)

[0168] in, These are the minimum and maximum values ​​in the first target performance data or the first environment data, respectively.

[0169] Step 506: Perform interpolation and padding on the second target performance data and the second environmental data to obtain preprocessed target performance data and processed environmental data.

[0170] In this embodiment, if there is missing data in the second target performance data or the second environment data, the terminal uses Lagrange interpolation to interpolate and fill in the second target performance data based on historical target performance data, or uses historical environment data to interpolate and fill in the second environment data, to obtain preprocessed target performance data and processed environment data. Specifically, the interpolation and filling process of Lagrange interpolation is as follows:

[0171] (twenty four)

[0172] in, For the moments when interpolation filling is required, For the first The time corresponding to a known historical performance data or a known historical environmental data. For the first The data value corresponding to a known historical performance data or a known historical environmental data. For the first The time corresponding to a known historical performance data or a known historical environmental data.

[0173] In this embodiment, wavelet transform is used to denoise the target performance data and environmental data, and normalization and Lagrange interpolation are used to fill in missing data, ensuring the data quality and consistency of the target performance data and environmental data. This provides a data basis for the subsequent determination of the target threshold range and improves the accuracy of the target threshold range generation.

[0174] In one exemplary embodiment, such as Figure 6 As shown, step 408 includes steps 602 to 604. Wherein:

[0175] Step 602: Determine the posterior distribution corresponding to the initial threshold based on the statistical inference structure.

[0176] In this embodiment, the statistical inference structure is a Bayesian inference structure. First, the terminal calculates the posterior distribution corresponding to the initial threshold based on the Bayesian inference structure. The initial threshold is as follows:

[0177]

[0178] The Bayesian inference results determine the posterior distribution corresponding to the initial threshold through the prior distribution and the likelihood function. The inference process is as follows:

[0179] (25)

[0180] (26)

[0181] in, It is the prior distribution; It is the likelihood function; It is a posterior distribution used to quantify the uncertainty in threshold prediction. and These represent the mean and standard deviation, respectively.

[0182] Step 604: Constrain and update the initial threshold based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

[0183] In this embodiment, the terminal estimates and analyzes the initial threshold based on the posterior distribution, the second performance index, and the second constraint condition, adjusts the dynamic range of the initial threshold, and constrains the adjustment of the dynamic range of the initial threshold. The constraint process is as follows:

[0184] (27)

[0185] (28)

[0186] in, , These are weighting factors, which respectively measure the impact of false positive rate, false negative rate, and response time. , It is obtained by adjusting the threshold optimization model during training. As the second performance indicator, This represents the dynamic range of the initial threshold. The second performance metric is minimized by adjusting the range of the initial threshold. By analyzing the false alarm rate, false negative rate, and response time under different initial threshold ranges, a balance is achieved among the false alarm rate, false negative rate, and response time to obtain the most suitable initial threshold.

[0187] Then, the terminal iteratively optimizes the initial threshold using the Lagrange multiplier method and updates it using gradient descent:

[0188] (29)

[0189] (30)

[0190] in, and For Lagrange multipliers, For Lagrange functions, As the initial threshold, These are the lower and upper limits of the initial threshold, respectively. For learning rate, For Lagrange functions The partial derivatives of . , and It is obtained by adjusting the training process of the threshold optimization model.

[0191] Finally, the terminal outputs the target threshold range, dynamically adjusted for the current monitoring period, through the threshold optimization model. and its confidence interval:

[0192] (31)

[0193] in, This parameter, used to describe the uncertainty of the threshold estimation, is derived based on the standard deviation of the posterior distribution.

[0194] In this embodiment, by combining Bayesian inference and the Lagrange multiplier method, accurate estimation and optimization of the target threshold interval are achieved. This not only considers the prior distribution and the likelihood function, but also ensures the rationality and robustness of the target threshold interval by weighting the false alarm rate, false negative rate, and response time. Through iterative optimization and gradient descent, the terminal can dynamically adjust the threshold to adapt to constantly changing environmental and performance conditions, thereby achieving precise monitoring and control of the aero-engine's operating status. Furthermore, the adaptive characteristics of the target threshold interval ensure that the aero-engine safety protection control system maintains the accuracy and sensitivity of monitoring under different flight altitudes, temperatures, and pressures, making it suitable for various complex flight conditions (such as long-duration flight missions under high load and extreme environments), significantly improving engine safety and operational efficiency. The dynamic threshold adjustment and fault-tolerant control mechanism reduce the need for human intervention, improve the automation level of the aero-engine safety protection control system, reduce maintenance costs, and extend the service life of the aero-engine.

[0195] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0196] Based on the same inventive concept, this application also provides a state monitoring device for implementing the state monitoring method described above. The solution provided by this device is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more state monitoring device embodiments provided below can be found in the limitations of the state monitoring method described above, and will not be repeated here.

[0197] In one exemplary embodiment, such as Figure 7 As shown, a state monitoring device 700 is provided, including: an acquisition module 701, a disturbance analysis module 702, a state prediction module 703, and a threshold generation module 704, wherein:

[0198] The acquisition module 701 is used to acquire environmental data and raw control data of the target object within a preset monitoring period;

[0199] The disturbance analysis module 702 is used to perform disturbance analysis on the original control data according to the disturbance prediction model to obtain disturbance compensation data;

[0200] The state prediction module 703 is used to predict the state of the target object based on the disturbance prediction model, disturbance compensation data and original control data, and obtain the target performance data of the target object.

[0201] The threshold generation module 704 is used to generate a target threshold range based on the threshold optimization model, environmental data, and target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0202] In one embodiment, the disturbance analysis module 702 is specifically used to perform disturbance prediction based on the disturbance prediction model, the system output data of the target object at the current moment, the system state estimate, and the historical disturbance data, to obtain the disturbance prediction value at the next moment.

[0203] The disturbance compensation amount is generated based on the disturbance prediction value, and the disturbance compensation data is obtained by performing disturbance optimization on the original control data based on the disturbance compensation amount, the first performance index and the first constraint condition.

[0204] In one embodiment, the state prediction module 703 is specifically used to perform attack compensation prediction based on the disturbance prediction model and disturbance compensation data to obtain the attack compensation amount.

[0205] The original control data is optimized by using the disturbance prediction model, disturbance compensation amount, and attack compensation amount to obtain the target control data.

[0206] The target object's state is predicted based on the disturbance prediction model and target control data to obtain target performance data.

[0207] In one embodiment, the threshold optimization model includes a deep learning structure and a statistical inference structure;

[0208] The threshold generation module 704 is specifically used to preprocess the target performance data and environmental data to obtain preprocessed target performance data and preprocessed environmental data.

[0209] Based on the deep learning structure, features are extracted from the preprocessed target performance data and preprocessed environmental data to obtain an initial feature matrix. The initial feature matrix is ​​then reduced in dimensionality to obtain the target feature matrix.

[0210] The target feature matrix is ​​analyzed and processed based on the deep learning structure to obtain the initial threshold;

[0211] The initial threshold is constrained and optimized based on the statistical inference structure to obtain the target threshold range.

[0212] In one embodiment, the threshold generation module 704 is specifically used to perform noise reduction processing on the target performance data and environmental data according to wavelet transform to obtain first target performance data and first environmental data.

[0213] The first target performance data and the first environment data are normalized to obtain the second target performance data and the second environment data.

[0214] Interpolation and padding are performed on the second target performance data and the second environmental data to obtain preprocessed target performance data and processed environmental data.

[0215] In one embodiment, the threshold generation module 704 is specifically used to determine the posterior distribution corresponding to the initial threshold based on the statistical inference structure;

[0216] The initial threshold is constrained and updated based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

[0217] Each module in the aforementioned status monitoring device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0218] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores environmental data, raw control data, and historical disturbances. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a status monitoring method.

[0219] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0220] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0221] Within a preset monitoring period, acquire environmental data and raw control data of the target object;

[0222] The disturbance prediction model is used to perform disturbance analysis on the original control data to obtain disturbance compensation data.

[0223] Based on the disturbance prediction model, disturbance compensation data and original control data, the state of the target object is predicted to obtain the target performance data of the target object;

[0224] The target threshold range is generated based on the threshold optimization model, environmental data, and target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period.

[0225] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0226] Based on the disturbance prediction model, the original control data, the current system output data of the target object, the system state estimate, and historical disturbance data, disturbance prediction is performed to obtain the disturbance prediction value for the next moment.

[0227] The disturbance compensation amount is generated based on the disturbance prediction value, and the disturbance compensation data is obtained by performing disturbance optimization on the original control data based on the disturbance compensation amount, the first performance index and the first constraint condition.

[0228] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0229] Attack compensation is predicted based on the disturbance prediction model and disturbance compensation data to obtain the attack compensation amount;

[0230] The original control data is optimized by using the disturbance prediction model, disturbance compensation amount, and attack compensation amount to obtain the target control data.

[0231] The target object's state is predicted based on the disturbance prediction model and target control data to obtain target performance data.

[0232] In one embodiment, the threshold optimization model includes a deep learning structure and a statistical inference structure; the processor, when executing the computer program, also implements the following steps:

[0233] The target performance data and environmental data are preprocessed to obtain preprocessed target performance data and preprocessed environmental data.

[0234] Based on the deep learning structure, features are extracted from the preprocessed target performance data and preprocessed environmental data to obtain an initial feature matrix. The initial feature matrix is ​​then reduced in dimensionality to obtain the target feature matrix.

[0235] The target feature matrix is ​​analyzed and processed based on the deep learning structure to obtain the initial threshold;

[0236] The initial threshold is constrained and optimized based on the statistical inference structure to obtain the target threshold range.

[0237] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0238] The target performance data and environmental data are denoised by wavelet transform to obtain the first target performance data and the first environmental data.

[0239] The first target performance data and the first environment data are normalized to obtain the second target performance data and the second environment data.

[0240] Interpolation and padding are performed on the second target performance data and the second environmental data to obtain preprocessed target performance data and processed environmental data.

[0241] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0242] The posterior distribution corresponding to the initial threshold is determined based on the statistical inference structure.

[0243] The initial threshold is constrained and updated based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

[0244] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0245] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0246] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0247] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0248] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0249] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A status monitoring method, characterized in that, The method includes: Within a preset monitoring period, acquire environmental data and raw control data of the target object; The original control data is subjected to disturbance analysis based on the disturbance prediction model to obtain disturbance compensation data; Based on the disturbance prediction model, the disturbance compensation data, and the original control data, the state of the target object is predicted to obtain the target performance data of the target object; A target threshold range is generated based on the threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period. The threshold optimization model includes a deep learning structure and a statistical inference structure; the step of generating a target threshold range based on the threshold optimization model, the environmental data, and the target performance data includes: The target performance data and the environmental data are preprocessed to obtain preprocessed target performance data and preprocessed environmental data; Based on the deep learning structure, feature extraction is performed on the preprocessed target performance data and the preprocessed environmental data to obtain an initial feature matrix, and the initial feature matrix is ​​then subjected to dimensionality reduction to obtain the target feature matrix. The target feature matrix is ​​analyzed and processed according to the deep learning structure to obtain an initial threshold; Based on the statistical inference structure, the initial threshold is constrained and optimized to obtain the target threshold range; The step of constraining and optimizing the initial threshold based on the statistical inference structure to obtain the target threshold interval includes: The posterior distribution corresponding to the initial threshold is determined based on the statistical inference structure. The initial threshold is constrained and updated based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

2. The method according to claim 1, characterized in that, The step of performing disturbance analysis on the original control data based on the disturbance prediction model to obtain disturbance compensation data includes: Based on the disturbance prediction model, the system output data of the target object at the current moment, the system state estimate, and the historical disturbance data, disturbance prediction is performed to obtain the disturbance prediction value at the next moment. Based on the predicted disturbance value, a disturbance compensation amount is generated, and based on the disturbance compensation amount, the first performance index, and the first constraint condition, the original control data is subjected to disturbance optimization to obtain disturbance compensation data.

3. The method according to claim 2, characterized in that, The step of predicting the state of the target object based on the disturbance prediction model, the disturbance compensation data, and the original control data to obtain the target performance data of the target object includes: Attack compensation is predicted based on the disturbance prediction model and the disturbance compensation data to obtain the attack compensation amount; The original control data is compensated and optimized based on the disturbance prediction model, the disturbance compensation amount, and the attack compensation amount to obtain the target control data; Based on the disturbance prediction model and the target control data, the state of the target object is predicted to obtain the target performance data.

4. The method according to claim 1, characterized in that, The step of preprocessing the target performance data and the environmental data to obtain preprocessed target performance data and preprocessed environmental data includes: The target performance data and the environmental data are denoised using wavelet transform to obtain first target performance data and first environmental data. The first target performance data and the first environment data are normalized to obtain the second target performance data and the second environment data. The second target performance data and the second environmental data are interpolated and filled to obtain preprocessed target performance data and processed environmental data.

5. A status monitoring device, characterized in that, The device includes: The acquisition module is used to acquire environmental data and raw control data of the target object within a preset monitoring period; The disturbance analysis module is used to perform disturbance analysis on the original control data according to the disturbance prediction model to obtain disturbance compensation data; The state prediction module is used to predict the state of the target object based on the disturbance prediction model, the disturbance compensation data and the original control data, so as to obtain the target performance data of the target object; A threshold generation module is used to generate a target threshold range based on a threshold optimization model, the environmental data, and the target performance data; the target threshold range is used to monitor the status of the target object within the current monitoring period. The threshold optimization model includes a deep learning structure and a statistical inference structure; The threshold generation module is specifically used to preprocess the target performance data and the environmental data to obtain preprocessed target performance data and preprocessed environmental data. Based on the deep learning structure, feature extraction is performed on the preprocessed target performance data and the preprocessed environmental data to obtain an initial feature matrix, and the initial feature matrix is ​​then subjected to dimensionality reduction to obtain the target feature matrix. The target feature matrix is ​​analyzed and processed according to the deep learning structure to obtain an initial threshold; Based on the statistical inference structure, the initial threshold is constrained and optimized to obtain the target threshold range; The threshold generation module is specifically used to determine the posterior distribution corresponding to the initial threshold based on the statistical inference structure. The initial threshold is constrained and updated based on the posterior distribution, the second performance index, and the second constraint to obtain the target threshold range.

6. The apparatus according to claim 5, characterized in that, The disturbance analysis module is specifically used to perform disturbance prediction based on the disturbance prediction model, the system output data of the target object at the current moment, the system state estimate, and historical disturbance data, and to obtain the disturbance prediction value at the next moment. Based on the predicted disturbance value, a disturbance compensation amount is generated, and based on the disturbance compensation amount, the first performance index, and the first constraint condition, the original control data is subjected to disturbance optimization to obtain disturbance compensation data.

7. The apparatus according to claim 6, characterized in that, The state prediction module is specifically used to predict attack compensation based on the disturbance prediction model and the disturbance compensation data, and to obtain the attack compensation amount. The original control data is compensated and optimized based on the disturbance prediction model, the disturbance compensation amount, and the attack compensation amount to obtain the target control data; Based on the disturbance prediction model and the target control data, the state of the target object is predicted to obtain the target performance data.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images