Controller loop self-diagnosis method, device and equipment

Abstract

The invention provides a controller loop self-diagnosis method, device and equipment, and the method comprises the steps: carrying out the multi-level phase space feature extraction of a plurality of sensor signals in a target equipment control system, and constructing a feature time sequence matrix; constructing a function dependence graph of a controller loop based on the characteristic time sequence matrix, calculating a time-varying causal relationship, and analyzing abnormal flow distribution to obtain a masking effect decoupling matrix; constructing a fault interaction potential field in combination with the feature time sequence matrix and the masking effect decoupling matrix, analyzing fault interaction characteristics on different time scales, and generating a fault interaction feature map; and inputting the fault interaction characteristic atlas into a preset working condition perception conditional probability network, and outputting a comprehensive diagnosis result of multiple faults by applying a hybrid reasoning framework. According to the method, the masking effect among multiple faults can be effectively decoupled, multiple fault types can be accurately identified and diagnosed, dangerous and wrong alarm triggering is avoided, and the safety and reliability of equipment operation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment diagnosis, and in particular to a controller loop self-diagnosis method, device and equipment. Background Art

[0002] Controller loop self-diagnosis technology has been widely used in the field of industrial automation, especially in equipment control systems in key industries such as power and chemical industry. Existing controller loop self-diagnosis methods are mainly based on single fault model design, and anomalies in the control loop are identified and located through observer models, spectrum analysis or statistical feature matching. However, in actual industrial environments, the simultaneous occurrence of multiple faults is not uncommon. For example, in a condenser level control system, water level gauge blockage may coexist with control valve failure. When multiple faults occur at the same time, the symptoms they produce may mask or modify each other, causing the traditional diagnosis method based on the single fault hypothesis to fail. Some fault signals are masked by other faults, and the system may be in a dangerous "false health" state, or trigger false fault alarms, affecting the safe operation and production efficiency of the equipment. Summary of the invention

[0003] The main purpose of the present invention is to solve the technical problem that the existing controller loop self-diagnosis method may fail to diagnose or misdiagnose due to the mutual masking effect between faults when facing multiple faults; A first aspect of the present invention provides a controller circuit self-diagnosis method, the controller circuit self-diagnosis method comprising: Perform multi-level phase space feature extraction on multiple sensor signals in the control system of the target device, and construct a feature time series matrix based on the extracted phase space features; Based on the characteristic timing matrix, a functional dependency graph of the controller loop of the target device is constructed and the time-varying causal relationship between the components in the functional dependency graph is calculated, and a masking effect decoupling matrix is ​​obtained by analyzing the abnormal traffic distribution in the time-varying causal relationship; Based on the characteristic timing matrix and the masking effect decoupling matrix, a fault interaction potential field is constructed and the fault interaction characteristics of the fault interaction potential field at different time scales are analyzed to generate a fault interaction characteristic map; The fault interaction feature map is input into a preset working condition perception conditional probability network, and a hybrid reasoning framework is applied to output a comprehensive diagnosis result of multiple faults.

[0004] Optionally, in a first implementation of the first aspect of the present invention, the step of performing multi-level phase space feature extraction on multiple sensor signals in the control system of the target device and constructing a feature time series matrix according to the extracted phase space features includes: Applying a delayed coordinate method to the time series signals of the plurality of sensors to reconstruct the phase space to obtain a high-dimensional phase space trajectory; Calculating nonlinear dynamic indicators for the high-dimensional phase space trajectory, and using adaptive sliding window technology to perform segmented calculations on the nonlinear dynamic indicators to obtain a time-varying nonlinear dynamic indicator sequence; The high-dimensional phase space trajectory, the nonlinear dynamic index and the time-varying nonlinear dynamic index sequence are used as phase space features, and a feature time series matrix is ​​constructed according to the extracted phase space features.

[0005] Optionally, in a second implementation of the first aspect of the present invention, based on the characteristic timing matrix, constructing a functional dependency graph of the controller loop of the target device and calculating the time-varying causal relationship between the components in the functional dependency graph, and obtaining a masking effect decoupling matrix by analyzing the abnormal traffic distribution in the time-varying causal relationship includes: Analyze the phase space feature correlation of each sensor in the feature timing matrix, and construct an initial functional dependency graph by combining the physical structure of the target device and the controller loop schematic diagram; Based on the time-varying sequence of nonlinear dynamic indicators in the characteristic time series matrix, the improved PC algorithm of the conditional mutual information criterion is used to determine the direction and strength of the causal association between components, and the initial functional dependency graph is optimized based on the direction and strength of the causal association; Based on the optimized functional dependency graph, the characteristic time series matrix is ​​divided into multiple time periods, and the time-varying Granger causality analysis method is applied to each time period, a vector autoregression model is established and statistics are calculated to obtain the time-varying causal relationship that characterizes the time-varying changes between the components; The causal strength in the time-varying causal relationship is regarded as the abnormal traffic distribution. According to the abnormal traffic conservation principle, the abnormal absorption rate is calculated for each component to construct the masking effect decoupling matrix.

[0006] Optionally, in a third implementation of the first aspect of the present invention, regarding the causal strength in the time-varying causal relationship as abnormal traffic distribution, calculating the abnormal absorption rate for each component according to the abnormal traffic conservation principle, and constructing the masking effect decoupling matrix includes: A directed weighted graph structure is established for the time-varying causal relationship, wherein nodes represent components of a target device, directions of edges represent directions of causal relationships, and weights of edges represent causal strength, wherein the causal strength is defined as abnormal traffic distribution between components; For each node in the directed weighted graph, the sum of abnormal flow distribution of the inflow node and the sum of abnormal flow distribution of the outflow node are calculated to obtain the abnormal absorption rate of each node; A masking effect decoupling matrix is ​​constructed according to the abnormal absorption rate and the causal strength, rows and columns of the masking effect decoupling matrix correspond to components, and matrix element values ​​are determined according to the abnormal absorption rate of the node and the relative causal strength between two nodes.

[0007] Optionally, in a fourth implementation of the first aspect of the present invention, constructing a fault interaction potential field based on the characteristic timing matrix and the masking effect decoupling matrix and analyzing the fault interaction characteristics of the fault interaction potential field at different time scales to generate a fault interaction feature map includes: Reorganize the feature time series matrix into a third-order tensor according to the three dimensions of the number of sensors, the number of time windows, and the feature dimension, and apply the constrained Tucker decomposition algorithm to the third-order tensor to extract the spatial pattern matrix representing the sensor spatial pattern and the core tensor representing the pattern interaction; Calculating the Hadamard product of the spatial mode matrix and the masking effect decoupling matrix, wherein the Hadamard product is used to adjust the influence weights of each component in the spatial mode matrix, combining the adjusted spatial mode matrix with the core tensor, and defining a fault interaction potential field; Acquire thermodynamic parameter measurement values ​​of the target device, calculate a thermodynamic compensation factor according to a thermodynamic model of the device, and perform compensation adjustment on the fault interaction potential field according to the thermodynamic compensation factor to obtain a compensated fault interaction potential field; The compensated fault interaction potential field is subjected to multi-scale decomposition by applying wavelet transform, fault interaction components of each scale are extracted, and the fault interaction components are mapped into a fault interaction feature map.

[0008] Optionally, in a fifth implementation of the first aspect of the present invention, applying wavelet transform to the compensated fault interaction potential field to perform multi-scale decomposition, extracting fault interaction components of each scale, and mapping the fault interaction components into a fault interaction feature map includes: According to a preset wavelet basis function, a discrete wavelet transform is applied to the compensated fault interaction potential field to decompose the fault interaction potential field into wavelet decomposition coefficients of different scales; Calculating energy distribution of the wavelet decomposition coefficients of different scales, obtaining energy contribution rate of each scale layer, and determining the time scale of occurrence of major fault interactions according to the energy contribution rate; Extracting the phase information of the time scale at which the main fault interaction occurs, and identifying the time delay characteristics and synchronization characteristics of the fault interaction of different components through cross-scale phase coherence analysis; The energy distribution, phase information, time delay characteristics and synchronization characteristics are converted into a graph structure representation as fault interaction components, and a community detection algorithm is applied to the graph structure representation to map it into a fault interaction feature spectrum.

[0009] Optionally, in a sixth implementation of the first aspect of the present invention, inputting the fault interaction feature map into a preset working condition perception conditional probability network, and applying a hybrid reasoning framework to output a comprehensive diagnosis result of multiple faults includes: By monitoring the key operating parameters of the target equipment and combining the fuzzy clustering algorithm to map the key operating parameters to the predefined operating condition space, the current operating condition category is determined; According to the determined current working condition category, the corresponding conditional probability network model is called from the pre-trained conditional probability network library, and the fault interaction feature map is converted into a network input feature vector, which is input into the conditional probability network, and the posterior probability of each fault type and the combination of fault types is calculated through forward reasoning; Applying the Demster-Schaffer evidence theory framework to the posterior probability, defining a basic probability distribution function for each fault type and combination, and calculating an uncertainty factor based on the operating status of the target device; The basic probability distribution function and the uncertainty factor are used to calculate the comprehensive trust through the Demster combination rule, and the comprehensive diagnosis result is generated by combining the interaction characteristics in the fault interaction feature map.

[0010] A second aspect of the present invention provides a controller circuit self-diagnosis device, the controller circuit self-diagnosis device comprising: A feature extraction module is used to collect direct thermal feature parameters, indirect thermal feature parameters and coolant quality feature parameters of the air compressor of the fuel cell in layers to obtain a multi-dimensional thermal data set of the air compressor; A causal analysis module, used for extracting features from the multidimensional thermal data set, inputting the extracted feature data set into a neural network model based on deep transfer learning, and obtaining a corresponding heat dissipation efficiency attenuation prediction model; An interactive identification module, used to establish a multi-objective optimization equation for the air compressor based on the multi-dimensional thermal data set and the heat dissipation efficiency attenuation prediction model, and solve the multi-objective optimization equation to obtain a heat dissipation control strategy set; The fault diagnosis module is used to dynamically partition the heat dissipation system of the air compressor, and coordinately adjust the flow rate and flow direction of the coolant in each area of ​​the dynamic partition according to the heat dissipation control strategy set to achieve the heat dissipation of the air compressor.

[0011] The third aspect of the present invention provides a controller loop self-diagnosis device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via lines; the at least one processor calls the instructions in the memory so that the controller loop self-diagnosis device executes the steps of the above-mentioned controller loop self-diagnosis method.

[0012] The controller loop self-diagnosis method, device and equipment described above construct a characteristic time series matrix by extracting multi-level phase space features of multiple sensor signals in the target device control system; construct a functional dependency graph of the controller loop based on the characteristic time series matrix and calculate the time-varying causal relationship, and obtain a masking effect decoupling matrix by analyzing the abnormal flow distribution; construct a fault interaction potential field by combining the characteristic time series matrix and the masking effect decoupling matrix, analyze the fault interaction characteristics on different time scales, and generate a fault interaction feature map; input the fault interaction feature map into a preset working condition perception conditional probability network, and apply a hybrid reasoning framework to output a comprehensive diagnosis result of multiple faults. The present invention can effectively decouple the masking effects between multiple faults, accurately identify and diagnose the various types of faults that exist, avoid dangerous and erroneous alarm triggering, and improve the safety and reliability of equipment operation.

[0013] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0014] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of a first embodiment of a controller circuit self-diagnosis method in an embodiment of the present invention; Figure 2 A schematic diagram of an embodiment of a controller circuit self-diagnosis device in an embodiment of the present invention; Figure 3 Schematic diagram of an embodiment of a controller circuit self-diagnosis device in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of 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.

[0017] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products, or device ends.

[0018] To facilitate understanding of this embodiment, a controller loop self-diagnosis method disclosed in an embodiment of the present invention is first described in detail. Figure 1 As shown, the method comprises the following steps: 101. Perform multi-level phase space feature extraction on multiple sensor signals in the control system of the target device, and construct a feature time series matrix according to the extracted phase space features; In one embodiment of the present invention, the multi-level phase space feature extraction of multiple sensor signals in the control system of the target device and the construction of a feature time series matrix based on the extracted phase space features include: applying the delayed coordinate method to the time series signals of the multiple sensors to reconstruct the phase space to obtain a high-dimensional phase space trajectory; calculating nonlinear dynamic indicators for the high-dimensional phase space trajectory, and using adaptive sliding window technology to perform segmented calculations on the nonlinear dynamic indicators to obtain a time-varying nonlinear dynamic indicator sequence; using the high-dimensional phase space trajectory, the nonlinear dynamic indicators and the time-varying nonlinear dynamic indicator sequence as phase space features, and constructing a feature time series matrix based on the extracted phase space features.

[0019] Specifically, this embodiment takes the condenser liquid level control system as an example to explain in detail the implementation process of the feature extraction module in the controller loop self-diagnosis method. The condenser liquid level control system typically includes a variety of sensors, such as differential pressure level gauges, capacitive level gauges, ultrasonic level gauges, and temperature, pressure and flow sensors. When processing the time series signals of these sensors, it is first necessary to collect raw data of sufficient length, usually collecting data for 30 minutes to 2 hours at the system sampling frequency to capture the dynamic characteristics of the system under different working conditions. The collected signals are preprocessed, including operations such as removing outliers, filtering and normalization to ensure data quality. Subsequently, the delayed coordinate method is applied to the preprocessed multiple sensor time series signals to reconstruct the phase space. This method is based on the Takens embedding theorem and can reconstruct the high-dimensional phase space of the system from the univariate time series, revealing the intrinsic dynamic characteristics of the system. In the specific implementation, for each sensor time series x(t), a vector [x(t), x(t+τ), x(t+2τ),..., x(t+(m-1)τ)] is constructed, where τ is the time delay and m is the embedding dimension. In the condenser level control system, the time delay τ is usually determined by calculating the first minimum of the mutual information function of the time series, and the embedding dimension m is determined by the pseudo-nearest neighbor analysis method. For the level signal, the typical embedding dimension is 3 to 5. Through the above process, the one-dimensional time series signal is mapped to the m-dimensional phase space to form a high-dimensional phase space trajectory, which can reflect the evolution law and dynamic characteristics of the system state.

[0020] Specifically, the nonlinear dynamic indicators are calculated for the obtained high-dimensional phase space trajectory. These indicators can quantify the complex behavioral characteristics of the system and provide an important basis for fault diagnosis. First, the correlation dimension is calculated. This indicator quantifies the complexity of the trajectory by measuring the distribution characteristics of the distance between point pairs in the phase space. Healthy systems and faulty systems usually show different correlation dimension values. Then the maximum Lyapunov index is calculated. This index measures the separation rate of adjacent trajectories in the phase space, reflects the degree of chaos of the system and the sensitivity to initial conditions. In the liquid level control loop, valve failure usually causes this index to increase significantly. Then, the indicators such as determinism, laminarity and entropy are calculated by recursive graph analysis. The recursive graph is to project the phase space trajectory onto a two-dimensional plane and analyze the proximity between the trajectory points. Determinism measures the predictability of system behavior, laminarity reflects the tendency of the system state to stay in a specific area, and entropy quantifies the complexity of the system dynamics. When calculating these nonlinear dynamic indicators, in order to capture the evolution characteristics of the system state over time, the adaptive sliding window technology is used to calculate the nonlinear dynamic indicators in segments. In the condenser level control system, the window width is adaptively adjusted according to the dynamic characteristics of the system, usually 500 to 2000 data points, corresponding to 1 to 5 minutes of actual time. When it is detected that the system operating conditions change rapidly, the window width is automatically reduced to improve the time resolution; when the system is relatively stable, the window width is increased to improve the statistical reliability of the index calculation. The overlap rate of adjacent windows is set to 50%, which is half of the window width, which not only ensures time continuity but also avoids the waste of computing resources. Through sliding window processing, the static nonlinear dynamic indicators are converted into time-varying sequences to form time-varying nonlinear dynamic indicator sequences. These sequences can reflect the evolution of the system's dynamic behavior, especially the system response characteristics when the condenser load changes or external disturbances exist.

[0021] The above three levels of phase space features, namely, high-dimensional phase space trajectory, nonlinear dynamic index and time-varying nonlinear dynamic index sequence, are organized into a multidimensional feature time series matrix, which serves as the basic data structure for subsequent fault diagnosis. The construction process of the feature time series matrix is ​​as follows: First, the number of rows and columns of the matrix is ​​determined. The number of rows is equal to the number of sliding windows, which represents the time dimension; the number of columns is equal to the total number of phase space features of all sensors, which represents the feature dimension. For each sensor, the key parameters in its high-dimensional phase space trajectory (such as the main components or statistical characteristics of the trajectory), the calculated nonlinear dynamic index (such as the correlation dimension, the maximum Lyapunov exponent, etc.) and the time-varying sequence values ​​of these indicators are selected and arranged in the columns of the matrix in a predefined order. In the actual application of the condenser level control system, a typical feature time series matrix may contain 20 to 50 time windows (rows) and 30 to 100 features (columns). Each element value in the matrix represents the calculation result of a specific feature in a specific time window. For example, the element in the i-th row and j-th column may represent the maximum Lyapunov exponent value of the level sensor signal in the i-th time window. Each column of the matrix is ​​standardized to make the features of different dimensions comparable. The characteristic timing matrix constructed in this way not only retains the dynamic characteristics of the original signal, but also contains the deep nonlinear system behavior characteristics, which can effectively capture the complex system response under multiple fault conditions.

[0022] 102. Based on the characteristic timing matrix, construct a functional dependency graph of the controller loop of the target device and calculate the time-varying causal relationship between the components in the functional dependency graph. By analyzing the abnormal traffic distribution in the time-varying causal relationship, the masking effect decoupling matrix is ​​obtained; In one embodiment of the present invention, based on the characteristic timing matrix, a functional dependency diagram of the controller loop of the target device is constructed and the time-varying causal relationship between the components in the functional dependency diagram is calculated. By analyzing the abnormal flow distribution in the time-varying causal relationship, a masking effect decoupling matrix is ​​obtained, including: analyzing the phase space feature correlation of each sensor in the characteristic timing matrix, combining the physical structure of the target device and the controller loop schematic diagram, to construct an initial functional dependency diagram; based on the time-varying sequence of nonlinear dynamic indicators in the characteristic timing matrix, using the improved PC algorithm of the conditional mutual information criterion, determining the causal correlation direction and strength between components, and optimizing the initial functional dependency diagram based on the causal correlation direction and strength; based on the optimized functional dependency diagram, dividing the characteristic timing matrix into multiple time periods, applying the time-varying Granger causal analysis method to each time period, establishing a vector autoregression model and calculating statistics, and obtaining the time-varying causal relationship that characterizes the time-varying causal relationship between the components; regarding the causal strength in the time-varying causal relationship as the abnormal flow distribution, calculating the abnormal absorption rate for each component according to the abnormal flow conservation principle, and constructing a masking effect decoupling matrix.

[0023] Specifically, after obtaining the characteristic timing matrix, it is necessary to construct the controller loop functional dependency diagram of the target device based on the matrix to reveal the mutual influence relationship between the system components. First, the phase space feature correlation of each sensor in the characteristic timing matrix is ​​analyzed. This process calculates the Pearson correlation coefficient, Spearman rank correlation coefficient and mutual information value for each pair of features in the matrix to form a correlation matrix between features. In the condenser level control system, the phase space characteristics of the level signal are usually highly correlated with parameters such as cooling water flow, turbine load and condenser pressure, which reflects the physical coupling relationship within the system. At the same time, combined with the physical structure of the target device and the control loop schematic diagram, the direct physical connection or signal transmission path between the components in the system is identified, and the functional boundary and interface definition of each component are clarified. For example, in the condenser level control loop, there are physical connections from the level sensor to the signal conditioning circuit, from the conditioning circuit to the controller, from the controller to the actuator (such as the regulating valve), and from the actuator to the controlled object (liquid level). By integrating the results of phase space feature correlation analysis with physical structure knowledge, an initial functional dependency graph is constructed. The graph is represented in the form of a directed graph, where nodes represent system components (such as various sensors, controllers, actuators, controlled objects, etc.), and edges represent physical or signal connections between components. The initial functional dependency graph reflects the expected connection relationship during system design, but has not yet considered the causal relationship revealed by actual operating data, and needs further optimization.

[0024] Based on the time-varying sequence of nonlinear dynamic indicators in the characteristic time series matrix, the improved PC algorithm of the conditional mutual information criterion is used to determine the direction and strength of the causal relationship between components. The PC algorithm was originally a causal graph learning algorithm based on conditional independence test, but it has limitations when processing nonlinear and non-Gaussian distributed industrial process data. Therefore, this embodiment adopts an improved PC algorithm, and changes the conditional independence test from linear correlation coefficient to conditional mutual information, which is more suitable for capturing complex causal relationships in nonlinear systems. The specific implementation process is as follows: First, each column of data in the characteristic time series matrix is ​​regarded as a random variable, representing the phase space characteristics of a component. Then a fully connected undirected graph is constructed, in which each node corresponds to a component. Then, the conditional independence between any two nodes is determined by conditional mutual information calculation. When the conditional mutual information value is less than a preset threshold, the two nodes are considered to be conditionally independent and the corresponding edges are removed. In the condenser system, the threshold is usually set to 0.05 to 0.1, depending on the system noise level and data quality. Subsequently, the direction of the remaining undirected edges is determined to form a directed acyclic graph. The algorithm uses kernel density estimation to calculate high-dimensional mutual information when processing edge cases, avoiding the dimensionality curse problem in traditional methods. Through this process, the direction and strength of the causal relationship between components are obtained, where the direction indicates the transmission direction of the causal influence, and the strength is represented by the numerical value of the conditional mutual information. Based on this causal relationship information, the initial functional dependency graph is optimized, including adding causal links that are not considered in the physical design but are shown to exist in the data, removing links that are physically connected but have no significant causal relationship, and adjusting the weights of the edges to reflect the causal strength. The optimized functional dependency graph more accurately reflects the actual operating status of the system and the mutual influence relationship between components.

[0025] Based on the optimized functional dependency graph, the characteristic time series matrix is ​​divided into multiple time periods to analyze the dynamic characteristics of the system in different operating stages. The division of time periods takes into account the changes in system operating conditions and operating cycles. In the condenser level control system, typical division methods include the startup phase, stable operation phase, load change phase, and disturbance response phase. The time-varying Granger causal analysis method is applied to each time period. This method is based on the prediction principle, that is, if the past value of variable X can help predict the future value of variable Y, then it is considered that X Granger causes Y. In the specific operation, a vector autoregression model is established for each pair of causally connected nodes in the functional dependency graph. Take node Y and all its possible parent nodes For example, we first build a complete model that includes all parent nodes, representing the relationship between the current value of Y and its own and all parent nodes' past values; then we build a model that excludes specific parent nodes. By comparing the prediction errors of the full model with those of the restricted model, the F statistic is calculated, which reflects the contribution of the excluded variable to the prediction of the target variable. The calculation of the F statistic takes into account the model complexity and sample size to ensure statistical significance. This process is performed separately for each time period, and finally the time-varying causal relationship between the components is obtained. The time-varying causal relationship is represented in matrix form, and the matrix elements It represents the causal influence strength of component i on component j at time t. In the condenser liquid level control system, this matrix reveals the dynamic changes of the causal relationship between system components under different operating conditions. For example, when the load changes sharply, the causal influence of the turbine load on the liquid level will be significantly enhanced.

[0026] The causal strength in the time-varying causal relationship is regarded as the abnormal traffic distribution, and the masking effect of multiple faults is analyzed by the abnormal traffic conservation principle. In this step, the system is regarded as an abnormal signal propagation network, and the causal strength represents the flow of abnormal signals between nodes. Ideally, based on the traffic conservation principle, the total amount of abnormal signals flowing into a node should be equal to the total amount of abnormal signals flowing out of the node, unless the node absorbs or amplifies the abnormal signal. For each time point in the time-varying causal relationship matrix C(t), a directed weighted graph is constructed. The nodes represent system components, the direction of the edges represents the causal relationship direction, and the weight of the edges represents the causal strength, that is, the abnormal traffic distribution. For each node i, the sum of the abnormal traffic flowing into the node In(i), that is, the sum of the weights of all edges pointing to node i; at the same time, the sum of the abnormal traffic flowing out of the node Out(i), that is, the sum of the weights of all edges pointing from node i, is calculated. According to the abnormal traffic conservation principle, the abnormal absorption rate of node i is defined as (In(i)-Out(i)) / In(i). A positive value of the abnormal absorption rate indicates that the node absorbs part of the abnormal signal, that is, the fault signal is masked; a negative value indicates that the node amplifies the abnormal signal. In the condenser system, when the water level meter blockage and the control valve failure occur at the same time, the abnormal absorption rate of the control valve node is usually high, indicating that the abnormal signal generated by the water level meter failure is partially masked by the control valve failure. Based on the abnormal absorption rate of each node and the relative causal strength between nodes, the masking effect decoupling matrix M is constructed, and the matrix elements It indicates the degree to which component i masks or amplifies the abnormal signal of component j. Specifically, Values ​​close to 1 indicate a strong masking effect, values ​​close to 0 indicate no significant masking, and negative values ​​indicate an amplification effect.

[0027] Furthermore, the causal strength in the time-varying causal relationship is regarded as the abnormal traffic distribution, and according to the principle of conservation of abnormal traffic, the abnormal absorption rate is calculated for each component, and the masking effect decoupling matrix is ​​constructed, including: establishing a directed weighted graph structure for the time-varying causal relationship, wherein the nodes represent the components of the target device, the directions of the edges represent the causal relationship directions, the weights of the edges represent the causal strength, and the causal strength is defined as the abnormal traffic distribution between components; for each node in the directed weighted graph, the sum of the abnormal traffic distribution flowing into the node and the sum of the abnormal traffic distribution flowing out of the node are calculated to obtain the abnormal absorption rate of each node; a masking effect decoupling matrix is ​​constructed according to the abnormal absorption rate and the causal strength, the rows and columns of the masking effect decoupling matrix correspond to the components, and the matrix element values ​​are determined according to the abnormal absorption rate of the node and the relative causal strength between the two nodes.

[0028] Specifically, in the condenser level control system, the time-varying causal relationship matrix C(t) contains the causal strength information between components at different time points, and this information needs to be mapped into a directed weighted graph. In the specific operation, the nodes of the graph are first determined. Each node represents a physical component or functional unit in the system, such as a level sensor, a pressure sensor, a controller, a control valve, a water pump, etc. Then, the edge connection relationship is established according to the causal relationship matrix. The non-zero elements in the matrix Indicates that there is a causal relationship from component i to component j at time t, and a directed edge is established between the corresponding nodes. The direction of the edge points from i to j, indicating the transmission direction of the causal influence. At the same time, the value of the matrix element is used as the weight of the edge to indicate the strength of the causal relationship. In actual implementation, a threshold θ (such as 0.05) is usually set, and the connection is established only when the causal strength is greater than θ to filter out weak or noise-induced pseudo causal relationships. In addition, for time-varying causal relationships, the causal matrix at a specific time point can be selected for analysis, or the average causal strength over a period of time can be calculated. In the condenser system, a typical directed weighted graph contains 10 to 20 nodes and 30 to 50 edges, forming a complex network structure. The key innovation is to define causal strength as the abnormal flow distribution between components. This definition is based on the following understanding: faults or abnormal states in the system will generate signal disturbances through the causal chain. These disturbances propagate in the system like fluids. The greater the causal strength, the smoother the propagation channel of the abnormal signal, and the larger the corresponding abnormal flow. In this way, the abstract causal relationship is transformed into the concept of abnormal traffic distribution with physical meaning, laying the foundation for the subsequent masking effect analysis.

[0029] Specifically, for the constructed directed weighted graph, it is necessary to calculate the abnormal flow distribution of each node, so as to quantify the absorption or amplification effect of each component on the abnormal signal. For each node i in the graph, first calculate the total abnormal flow distribution In(i) flowing into the node, that is, the sum of the weights of all edges pointing to node i. This calculation process is formally expressed as the accumulation of all weight values ​​w(j,i) that satisfy the existence of edge j→i. In the condenser level control system, for the controller node, the inflow abnormal flow comes from multiple signal sources such as liquid level sensors and pressure sensors; for the control valve node, it mainly comes from the output signal of the controller. Similarly, calculate the total abnormal flow distribution Out(i) flowing out of node i, that is, the sum of the weights of all edges pointing from node i, which represents the total amount of abnormal signal propagation generated by node i to other components. After obtaining the sum of the inflow and outflow abnormal flows, calculate the abnormal absorption rate α(i)=(In(i)-Out(i)) / In(i) of node i. The abnormal absorption rate is a value ranging from [-∞,1]. Its physical meaning is the processing characteristics of the node for the abnormal signal flowing through it: when α(i) is close to 1, it means that node i absorbs almost all the abnormal signals flowing in, that is, a strong masking effect; when α(i) is close to 0, it means that node i transmits the abnormal signal almost losslessly, without significant masking or amplification; when α(i) is a negative value, it means that node i amplifies the abnormal signal and the output exceeds the input. In the condenser system, this situation is common in the control link where self-excited oscillation occurs. It should be noted that when the total inflow abnormal flow In(i) is 0, the abnormal absorption rate cannot be calculated directly. At this time, α(i) is set to 0, indicating that the node does not participate in the processing of abnormal signals. By calculating the abnormal absorption rate of each component in the system, the quantitative representation of the masking or amplification characteristics of each component for the fault signal is obtained, which is the basic data for constructing the masking effect decoupling matrix.

[0030] Specifically, according to the calculated abnormal absorption rate and causal strength, constructing the masking effect decoupling matrix is ​​the key to achieve multiple fault separation. The rows and columns of the matrix M correspond to the components in the system, and the matrix elements It indicates the degree of masking or amplification of the abnormal signal of component j by component i. The specific calculation of the matrix elements takes into account two factors: the abnormal absorption rate of the source node and the relative causal strength between the two nodes. First, set =α(i), that is, the matrix diagonal elements are equal to the abnormal absorption rate of the corresponding node, which represents the processing characteristics of the node for its own fault signal. (i≠j), it is calculated as follows: If there is a direct causal path from i to j, then =α(i)×(w(i,j) / Out(i)), where w(i,j) is the edge weight from i to j. This formula distributes the abnormal absorption rate of node i according to the relative causal strength from i to j. If there is no direct path, but there is an indirect path i→k→j through the intermediate node k, then = × , indicating the composite masking effect along the path; if there is neither direct path nor indirect path, then =0, indicating that i has no masking effect on j. In actual calculations, the maximum length of the indirect path is usually limited (such as 2 or 3) to control the computational complexity and avoid weak effects on too long paths. For the condenser level control system, a typical masking effect decoupling matrix size is 10×10 to 20×20, depending on the number of system components. The masking effect decoupling matrix constructed in this way comprehensively describes the abnormal signal modulation relationship between the components in the system. Larger positive values ​​in the matrix indicate strong masking areas, which require special attention because faults in these areas are easily masked by other faults and not detected; while larger negative values ​​indicate signal amplification areas, where minor faults may be mistakenly judged as serious problems.

[0031] 103. Based on the characteristic timing matrix and the masking effect decoupling matrix, the fault interaction potential field is constructed and the fault interaction characteristics of the fault interaction potential field at different time scales are analyzed to generate the fault interaction characteristic map; In one embodiment of the present invention, the method of constructing a fault interaction potential field based on the characteristic timing matrix and the masking effect decoupling matrix and analyzing the fault interaction characteristics of the fault interaction potential field at different time scales to generate a fault interaction feature map includes: reorganizing the characteristic timing matrix into a third-order tensor according to the three dimensions of the number of sensors, the number of time windows and the characteristic dimension, and applying the constrained Tucker decomposition algorithm to the third-order tensor to extract the spatial mode matrix representing the sensor spatial mode and the core tensor representing the mode interaction; calculating the Hadamard product of the spatial mode matrix and the masking effect decoupling matrix, the Hadamard product is used to adjust the influence weights of each component in the spatial mode matrix, combining the adjusted spatial mode matrix with the core tensor to define the fault interaction potential field; obtaining the measured values ​​of the thermodynamic parameters of the target device, calculating the thermodynamic compensation factor according to the thermodynamic model of the device, compensating and adjusting the fault interaction potential field according to the thermodynamic compensation factor, and obtaining the compensated fault interaction potential field; applying wavelet transform to the compensated fault interaction potential field for multi-scale decomposition, extracting fault interaction components of each scale, and mapping the fault interaction components into a fault interaction feature map.

[0032] Specifically, in the condenser level control system, the characteristic time series matrix contains multiple phase space features of each sensor in different time windows, usually with rows representing time windows and columns representing features. In order to better capture the interaction between sensors, this two-dimensional structure needs to be reorganized into a three-order tensor. In the specific implementation, the three dimensions of the reorganized tensor need to be determined first: the number of sensors, the number of time windows, and the feature dimension of each sensor. For the sensor dimension, the system includes types such as liquid level, pressure, temperature, and flow, with a total of 5 to 10; the number of time windows depends on the duration of data acquisition and the setting of the sliding window, usually 20 to 50; the characteristics of each sensor include key parameters of phase space reconstruction, nonlinear dynamic indicators, etc., usually 5 to 10. When performing dimensional reorganization, the columns of the original matrix are first grouped according to the sensor type, and then each group of data is rearranged according to the time window and feature type. In the specific operation, a data reorganization algorithm is written to read each element in the original matrix in turn, and place it in the corresponding position of the three-order tensor according to the predefined mapping relationship to ensure that different features of the same sensor are stored continuously in the tensor. After the reorganization is completed, the tensor is subjected to constrained Tucker decomposition. The algorithm first decomposes the tensor by alternating least squares method, sets the initial rank to the minimum value that can retain more than 85% of the information, which is usually 3 to 5 in the condenser system, and then iterates and optimizes the decomposition result until convergence. In each iteration, two factor matrices are fixed to solve the third one, and the Lagrange multiplier method is used to handle the constraints, which include the non-negativity and sparsity of the spatial pattern. The iteration stop condition is that the relative error is less than the preset threshold (such as 0.001) or the maximum number of iterations (such as 100) is reached. After the decomposition is completed, the matrix representing the spatial pattern of the sensor and the core tensor representing the interaction of the pattern are extracted. The former describes the synergistic relationship between sensors, and the latter characterizes the interaction between these spatial patterns at different times and features.

[0033] Specifically, calculating the Hadamard product of the spatial pattern matrix and the masking effect decoupling matrix is ​​the core link in constructing the fault interaction potential field. This process integrates the data-driven characteristic pattern with the masking effect based on the physical model. The Hadamard product performs the multiplication operation of the corresponding positions of the elements on two matrices of the same dimension. Each element of the result matrix is ​​the product of the elements of the corresponding positions of the original matrix. When implementing this calculation process, first ensure that the dimensions of the two matrices match. In the condenser system, the number of rows (i.e., the number of sensors) of the spatial pattern matrix is ​​usually 5 to 10, while the masking effect decoupling matrix is ​​a square matrix of the same dimension. When the dimensions do not match, the matrix size is adjusted by row expansion or compression. The specific methods include copying rows, deleting rows, or merging multiple rows into one row. After the adjustment is completed, write a loop to traverse each corresponding element of the two matrices, multiply them and store them in the same position of the result matrix. The computational complexity of this process is the square of the number of sensors. The calculated Hadamard product is actually an adjustment to the original spatial pattern, which corrects the pattern distortion caused by fault masking. Next, the adjusted spatial pattern matrix is ​​combined with the core tensor to define the fault interaction potential field. The tensor-matrix product operation is used to implement the product calculation of the adjusted spatial pattern matrix and the first dimension of the core tensor. The specific algorithm is: initialize an output tensor with the same dimension as the core tensor, and then multiply each slice of the core tensor with the adjusted spatial pattern matrix and store it in the corresponding position of the output tensor. In the code to implement this calculation, a triple nested loop is used to traverse the elements of the tensor, and matrix multiplication is used to calculate the update value of each slice. The computational complexity of this operation depends on the dimension of the tensor and the number of spatial patterns. In the condenser system, thousands to tens of thousands of basic arithmetic operations are usually required. Through the combined operation of Hadamard product and tensor-matrix product, a multidimensional potential field reflecting the fault interaction characteristics between components is successfully constructed.

[0034] Specifically, obtaining the measured values ​​of the thermodynamic parameters of the target equipment and making compensation adjustments according to the thermodynamic model are important links to ensure the accuracy of the fault interaction potential field. This process solves the problem of modulation of the fault characteristics by the change of the system's thermodynamic state. In the condenser level control system, the acquisition of thermodynamic parameters is achieved through sensors distributed throughout the system, including temperature sensors and pressure sensors. The temperature sensor uses PT100 or thermocouples to transmit temperature data through 4-20mA current signals; the pressure sensor uses piezoresistive or capacitive sensors, which are also transmitted through 4-20mA signals. These raw signals are converted into actual physical values ​​after A / D conversion and linear calibration by the data acquisition card. During the data acquisition process, median filtering is used to remove interference signals, and the median of 10 sampling points within one second is taken as the current measurement value to reduce the impact of instantaneous fluctuations. The calculation of the thermodynamic compensation factor is based on the thermodynamic model of the condenser, which is obtained through historical data training when the system is deployed. In the specific implementation, the deviation between the current thermodynamic parameters and the reference state is first calculated. The reference state is the average value of the system design condition or historical normal operation. These deviations are then converted into compensation factors through polynomial fitting or table lookup. This factor is a value between 0 and 1 that represents the similarity between the current thermodynamic state and the reference state. During the calculation process, temperature and pressure have different weights, which are determined based on system characteristics and expert experience and stored in the calculation program as constants. Thermodynamic compensation is implemented using a multiplication operation to multiply each element of the fault interaction potential field by the same compensation factor. This operation is implemented by traversing all elements of the tensor through multiple loops, and the time complexity is proportional to the size of the tensor. In the compensated fault interaction potential field, the abnormal features caused by the change in thermodynamic state are appropriately weakened, while the features caused by the real fault maintain their original strength, thereby enhancing the pertinence of fault identification.

[0035] Specifically, applying wavelet transform to multi-scale decomposition of the compensated fault interaction potential field is an important means to identify the fault interaction characteristics. This process reveals the fault interaction patterns at different time scales. In the specific implementation, the wavelet basis function suitable for industrial process data analysis is first selected. The condenser system usually uses the fourth-order Daubechies wavelet, which has good signal reconstruction characteristics and computational efficiency. The program implementation of the wavelet transform uses the pyramid algorithm to process each one-dimensional signal sequence along the time dimension of the fault interaction potential field. The algorithm first applies a low-pass filter and a high-pass filter to the signal to obtain the approximate coefficient and the detail coefficient respectively, and then doubles the approximate coefficient to form the input of the next layer, and so on recursively until the preset number of decomposition layers is reached. In the implementation, the pre-calculated filter coefficient array is used to obtain the coefficients of each layer through convolution operation. The convolution uses the boundary extension method to process the signal edge to avoid distortion. For the condenser system, 3 to 5 layers of decomposition are usually performed to capture the dynamic characteristics of different time scales. Subsequently, the energy distribution of each layer of detail coefficient is calculated by solving the square sum of the coefficients and dividing it by the total energy to obtain the energy proportion of each scale. The time scale at which the main fault interaction occurs is determined based on the energy distribution. The scale layer with an energy share of more than 10% is usually selected as the focus. Next, the characteristic coefficients are extracted from each important scale layer. The specific method is to retain the part of the coefficient whose absolute value exceeds the threshold while considering its time position and sign. Finally, the extracted fault interaction components of each scale are mapped to a fault interaction feature map. The implementation adopts a graphical representation method, in which each system component is represented as a node in the graph, and the interaction relationship between components is represented as an edge between nodes. The attributes of the node (size, color) encode the severity of the fault, and the attributes of the edge (thickness, color) represent the interaction strength and direction. The map construction algorithm first initializes an empty graph structure, then sets the attributes of the nodes and edges according to the characteristic values ​​of the fault interaction components, and finally converts the graph structure into a visual format output, such as a network diagram or a heat map.

[0036] Furthermore, the application of wavelet transform to the compensated fault interaction potential field for multi-scale decomposition, extraction of fault interaction components of each scale, and mapping of the fault interaction components to a fault interaction feature spectrum includes: applying discrete wavelet transform to the compensated fault interaction potential field according to a preset wavelet basis function, decomposing the fault interaction potential field into wavelet decomposition coefficients of different scales; calculating energy distribution of the wavelet decomposition coefficients of different scales to obtain the energy contribution rate of each scale layer, and determining the time scale of the main fault interaction according to the energy contribution rate; extracting the phase information of the time scale of the main fault interaction, and identifying the time delay characteristics and synchronization characteristics of fault interactions of different components through cross-scale phase coherence analysis; converting the energy distribution, phase information, time delay characteristics and synchronization characteristics as fault interaction components into a graph structure representation, and applying a community detection algorithm to the graph structure representation to map it into a fault interaction feature spectrum.

[0037] Specifically, in the condenser level control system, DB4 (4th-order Daubechies wavelet) is usually selected as the preset basis function. In the specific implementation, the compensated fault interaction potential field is first expanded into a one-dimensional signal sequence according to the time dimension. Then the Mallat fast wavelet decomposition algorithm is executed to create low-pass and high-pass filter coefficient arrays, and convolution operations are performed on each signal sequence and double downsampling is performed to obtain approximate coefficients and detail coefficients. For the approximate coefficients, the above decomposition process is repeated, usually 3 to 5 layers of decomposition. The boundary processing adopts the symmetric extension method to avoid signal edge distortion. After the decomposition is completed, the detail coefficients of each layer and the approximate coefficients of the final layer are obtained. These coefficients represent the characteristics of the fault interaction potential field at different time scales. In the condenser system, the first layer of detail coefficients corresponds to fast fluctuations of 5-10 seconds, the second layer corresponds to fluctuations of 10-20 seconds, and higher layers correspond to longer slow changes.

[0038] Specifically, the energy distribution of wavelet decomposition coefficients of different scales is calculated to identify the dominant scale of fault interaction. In implementation, the energy values ​​of each layer of detail coefficients and the final layer of approximate coefficients are calculated separately by summing the squares of the coefficient sequence. In order to eliminate the influence of signal length differences, the energy of each layer is divided by the total energy to obtain the normalized energy contribution rate. In the specific operation, an array with the same length as the number of decomposition layers is created to store the energy of each layer. The coefficients of each layer are traversed by looping, and their square values ​​are accumulated and finally normalized. According to the calculated energy contribution rate, the threshold method is used to determine the main fault interaction scale, and the scale layer with a contribution rate of more than 15% is usually selected. In the condenser system, when the water level gauge blockage and control valve failure exist at the same time, the energy contribution rate of the medium time scale (10-20 seconds) is often higher; while the cooling water pump failure is reflected in the energy of the longer time scale (30-60 seconds). This energy distribution feature helps to distinguish different types of multiple faults.

[0039] Specifically, the phase information of the main fault interaction time scale is extracted to analyze the timing characteristics of the fault signal. The implementation method is to perform Hilbert transform on the wavelet coefficients of the main fault interaction scale, first perform fast Fourier transform (FFT), set the negative frequency part to zero, multiply the positive frequency part by 2, and then perform inverse FFT to obtain the analytical signal. From the analytical signal, the phase sequence is obtained by calculating arctan (imaginary part / real part), and the phase unwrapping process is performed to obtain continuous phase changes. Subsequently, cross-scale phase coherence analysis is performed, the phase difference is calculated for the phase sequence of any two components, and the mean and standard deviation of the phase difference are analyzed. The mean value of the phase difference is divided by the center frequency to estimate the time delay, and the standard deviation reflects the stability of the phase relationship. In this way, the propagation order and speed of the fault signal between different components are determined, and the causal relationship network of the fault is established, which is crucial to identifying the fault source and understanding the fault propagation mechanism.

[0040] Specifically, the fault interaction components are converted into graph structure representation and mapped into fault interaction feature maps by applying community detection algorithm. The implementation process first creates a graph data structure, such as an adjacency matrix, defines nodes to represent system components, and edges to represent the interaction relationship between components. Set node attributes including ID, type, size (determined by energy proportion) and color (indicating fault type); edge attributes include weight (determined by energy correlation), direction (determined by phase analysis) and delay value (calculated based on phase difference). After the graph is constructed, the Louvain community detection algorithm is applied to identify tightly connected node clusters. The algorithm first assigns each node to an independent community, and then iteratively moves the node to the neighboring community that can maximize the modularity gain until the modularity no longer increases. Finally, the community structure is integrated with the original graph to generate a fault interaction feature map, marking each community with different colors, the thickness of the edge represents the interaction strength, and the direction arrow indicates the fault propagation path. In the condenser system, the map intuitively displays the fault interaction mode between components such as water level gauges, control valves and cooling systems, providing visual support for multiple fault diagnosis.

[0041] 104. The fault interaction feature map is input into the preset working condition perception conditional probability network, and the hybrid reasoning framework is applied to output the comprehensive diagnosis results of multiple faults.

[0042] In one embodiment of the present invention, the fault interaction feature map is input into a preset working condition perception conditional probability network, and a hybrid reasoning framework is applied to output a comprehensive diagnosis result of multiple faults, including: by monitoring the key operating parameters of the target equipment, the key operating parameters are mapped into a predefined working condition space in combination with a fuzzy clustering algorithm to determine the current working condition category; according to the determined current working condition category, the corresponding conditional probability network model is called from a pre-trained conditional probability network library, and the fault interaction feature map is converted into a network input feature vector, which is input into the conditional probability network, and the posterior probability of each fault type and the combination of fault types is calculated through forward reasoning; the posterior probability is processed by applying the Demster-Schaffer evidence theory framework to define a basic probability distribution function for each fault type and combination, and the uncertainty factor is calculated according to the operating state of the target equipment; the basic probability distribution function and the uncertainty factor are used to calculate the comprehensive trust through the Demster combination rule, and the comprehensive diagnosis result is generated in combination with the interactive characteristics in the fault interaction feature map.

[0043] Specifically, by monitoring the key operating parameters of the target equipment and combining the fuzzy clustering algorithm to map the key operating parameters into the predefined operating space, the process of determining the current operating condition category starts with data acquisition. In the condenser level control system, the key operating parameters include turbine load, condenser vacuum, cooling water inlet and outlet temperature, ambient temperature, and system start and stop status. These parameters are collected in real time through a fieldbus network (such as PROFIBUS or Modbus), and the sampling period is usually 100 milliseconds to 1 second. The collected parameter data is preprocessed, including outlier removal, filtering, and standardization. The 3σ criterion is used for outlier removal, and data points that exceed the mean ±3 times the standard deviation are identified as outliers and replaced with the previous valid value. The filtering uses a low-pass filter to eliminate high-frequency noise, and the cutoff frequency is set to 1Hz, which is sufficient to retain the main dynamic characteristics of the system while filtering out measurement noise. The standardization process converts each parameter to the [0,1] interval for subsequent processing. The processed parameter data is used as the input of the fuzzy clustering algorithm, which uses the fuzzy C-means (FCM) method to softly assign data points in the multidimensional parameter space to predefined operating condition categories. When the FCM algorithm is implemented, the cluster center is first initialized, usually based on the selection of typical operating condition points in historical data; then the membership of the data points and the cluster center are iteratively updated until the objective function converges or the maximum number of iterations (such as 100 times) is reached. In the condenser system, the operating space usually includes 5 to 8 categories such as full load operation, partial load operation, startup process, shutdown process and rapid load change. After clustering is completed, the membership of the current parameter and each operating condition center is calculated, and the category with the highest membership is taken as the current operating condition. This operating condition identification method takes into account the fuzziness of the parameters and the transition characteristics between the operating conditions, and is suitable for judging the operating status of industrial systems.

[0044] According to the determined current working condition category, calling the corresponding working condition-specific conditional probability network model from the pre-trained conditional probability network library is the core link to realize working condition perception diagnosis. Before the system is deployed, a dedicated conditional probability network model is trained for each predefined working condition. These models adopt the Bayesian network structure and are learned through historical fault data. The training process includes two stages: structure learning and parameter learning. The structure learning uses a scoring search algorithm to determine the network topology, and the scoring function uses the Bayesian Information Criterion (BIC) to balance the model complexity and fit; the parameter learning uses the maximum likelihood estimation method to determine the value of the conditional probability table. The trained model is stored in the network library. Each model contains a node definition file, a structure file, and a parameter file. When the current working condition is determined, the system loads the corresponding model from the library according to the working condition ID. The loading process includes reading the file, parsing the network structure, and initializing the conditional probability table. At the same time, the fault interaction feature map is converted into a network input feature vector. The conversion method is to extract key features from the map, such as the fault energy of the main node, the interaction strength of the edge, and the community structure characteristics. In the condenser system, the typical feature vector length is 20 to 50, which contains the main information of the fault spectrum. The extracted feature vector is input into the conditional probability network, set as the observation evidence of the corresponding node, and then the forward reasoning of the network is performed. The reasoning process adopts the variable elimination algorithm or the connection tree algorithm, and the calculation process is bottom-up message passing, and finally the posterior probability of each fault type and combination is obtained. This condition-specific diagnosis method takes into account the influence of the system state on the fault characteristics and improves the diagnostic accuracy in the case of multiple faults.

[0045] Applying the Demster-Schaffer evidence theory framework to the posterior probability is a key step in solving diagnostic uncertainty. Demster-Schaffer theory is different from traditional probability theory in that it allows trust to be assigned to a set of propositions rather than a single proposition, and is more suitable for dealing with incomplete and partially conflicting evidence. The implementation process first defines the identification framework Θ, which contains all possible fault types and their combinations. In the condenser system, there are usually 10 to 20 basic fault types, such as level gauge faults and control valve faults. Then the posterior probability output by the conditional probability network is converted into a basic probability assignment function (BPA). The conversion method is to determine the trust of each fault set based on the probability value and threshold. In the specific implementation, a series of thresholds (such as 0.1, 0.3, 0.5, and 0.7) are set to assign different basic probability values ​​to fault types with posterior probabilities falling in different intervals, while retaining a portion of the probability mass to the "unknown" state, indicating the uncertainty of the evidence. Then, the uncertainty factor is calculated based on the operating status of the target equipment. This factor takes into account factors that affect the reliability of diagnosis, such as operating time, load change rate, and environmental interference. In the condenser system, the uncertainty factor is high at the initial startup or when the load changes rapidly, indicating that the reliability of the diagnostic results at this time is low; while in the stable operation stage, the uncertainty factor is low. The uncertainty factor is calculated using the weighted sum method, and the weight of each influencing factor is determined by expert experience or historical data analysis. This processing method based on evidence theory takes into account the uncertainty in the diagnostic process and provides richer information than pure probability output.

[0046] Using the basic probability distribution function and uncertainty factor, the comprehensive trust is calculated by the Demster combination rule, and combined with the interactive characteristics in the fault interaction feature map, generating a comprehensive diagnosis result is the final diagnosis link. The Demster combination rule is used to fuse information from different evidence sources. It is implemented in the system as a pairwise calculation of orthogonal sums and then normalizing the results. In the specific calculation process, the basic probability distribution of multiple feature dimensions in the system (such as energy features, phase features, interaction features, etc.) is regarded as an independent source of evidence, and their combined results are calculated through nested loops. For the 3-5 common evidence sources in the condenser system, the time complexity of the combined calculation is proportional to the number of evidence sources and the size of the identification framework, and can usually be completed in milliseconds on modern processors. The calculated comprehensive trust represents the comprehensive trust level of each fault hypothesis, taking into account the support and conflict of all evidence. Subsequently, the comprehensive trust is combined with the uncertainty factor to appropriately reduce the trust level of the diagnosis results under high uncertainty conditions. Finally, combined with the interactive characteristics in the fault interaction feature map, especially the community structure and propagation path information, the type, location, severity and causal relationship of multiple faults are determined. The generated comprehensive diagnostic results include: identification results of major fault types and combinations (such as "water level meter blockage + control valve failure"); severity assessment of each fault, expressed as 0-100%; confidence interval of the diagnostic results, reflecting the reliability range of the results; prediction of the evolution trend of the fault, estimating the direction and rate of fault development based on time series analysis; and cause-effect relationship analysis of the fault, clarifying which are root faults and which are derivative faults.

[0047] In this embodiment, a characteristic timing matrix is ​​constructed by performing multi-level phase space feature extraction on multiple sensor signals in the target device control system; a functional dependency graph of the controller loop is constructed based on the characteristic timing matrix and the time-varying causal relationship is calculated, and a masking effect decoupling matrix is ​​obtained by analyzing the abnormal flow distribution; a fault interaction potential field is constructed by combining the characteristic timing matrix and the masking effect decoupling matrix, and the fault interaction characteristics on different time scales are analyzed to generate a fault interaction feature map; the fault interaction feature map is input into a preset working condition perception conditional probability network, and a hybrid reasoning framework is applied to output a comprehensive diagnosis result of multiple faults. The present invention can effectively decouple the masking effects between multiple faults, accurately identify and diagnose the various types of faults that exist, avoid dangerous and erroneous alarm triggering, and improve the safety and reliability of equipment operation.

[0048] The controller circuit self-diagnosis method in the embodiment of the present invention is described above. The controller circuit self-diagnosis device in the embodiment of the present invention is described below. Figure 2 , an embodiment of the controller circuit self-diagnosis device in the embodiment of the present invention includes: The feature extraction module 201 is used to collect the direct thermal feature parameters, indirect thermal feature parameters and coolant quality feature parameters of the air compressor of the fuel cell in layers to obtain a multi-dimensional thermal data set of the air compressor; A causal analysis module 202 is used to extract features from the multidimensional thermal data set, and input the extracted feature data set into a neural network model based on deep transfer learning to obtain a corresponding heat dissipation efficiency attenuation prediction model; An interactive identification module 203 is used to establish a multi-objective optimization equation for the air compressor based on the multi-dimensional thermal data set and the heat dissipation efficiency attenuation prediction model, and solve the multi-objective optimization equation to obtain a heat dissipation control strategy set; The fault diagnosis module 204 is used to dynamically partition the heat dissipation system of the air compressor, and coordinately adjust the flow rate and flow direction of the coolant in each area of ​​the dynamic partition according to the heat dissipation control strategy set to achieve heat dissipation of the air compressor.

[0049] In an embodiment of the present invention, the controller loop self-diagnosis device runs the above controller loop self-diagnosis method, and the controller loop self-diagnosis device extracts multi-level phase space features of multiple sensor signals in the target device control system to construct a feature time series matrix; constructs a functional dependency graph of the controller loop based on the feature time series matrix and calculates the time-varying causal relationship, and obtains a masking effect decoupling matrix by analyzing the abnormal flow distribution; combines the feature time series matrix and the masking effect decoupling matrix to construct a fault interaction potential field, analyzes the fault interaction characteristics on different time scales, and generates a fault interaction feature map; inputs the fault interaction feature map into a preset working condition perception conditional probability network, and applies a hybrid reasoning framework to output a comprehensive diagnosis result of multiple faults. The present invention can effectively decouple the masking effects between multiple faults, accurately identify and diagnose the various types of faults that exist, avoid dangerous and erroneous alarm triggering, and improve the safety and reliability of equipment operation.

[0050] above Figure 2 The controller loop self-diagnosis device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The controller loop self-diagnosis device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0051] Figure 3: is a structural diagram of a controller loop self-diagnosis device provided by an embodiment of the present invention. The controller loop self-diagnosis device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage device terminals) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the controller loop self-diagnosis device 300. Furthermore, the processor 310 can be configured to communicate with the storage medium 330, and execute a series of instruction operations in the storage medium 330 on the controller loop self-diagnosis device 300 to implement the steps of the above-mentioned controller loop self-diagnosis method.

[0052] The controller loop self-diagnostic device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 3 The controller loop self-diagnosis device structure shown does not constitute a limitation on the controller loop self-diagnosis device provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0053] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0054] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0055] As described above, 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 aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A controller loop self-diagnosis method, characterized in that: The controller loop self-diagnosis method comprises: Perform multi-level phase space feature extraction on multiple sensor signals in the control system of the target device, and construct a feature time series matrix based on the extracted phase space features; Based on the characteristic timing matrix, a functional dependency graph of the controller loop of the target device is constructed and the time-varying causal relationship between the components in the functional dependency graph is calculated, and a masking effect decoupling matrix is ​​obtained by analyzing the abnormal traffic distribution in the time-varying causal relationship; Based on the characteristic timing matrix and the masking effect decoupling matrix, a fault interaction potential field is constructed and the fault interaction characteristics of the fault interaction potential field at different time scales are analyzed to generate a fault interaction characteristic map; The fault interaction feature map is input into a preset working condition perception conditional probability network, and a hybrid reasoning framework is applied to output a comprehensive diagnosis result of multiple faults.

2. The controller circuit self-diagnosis method according to claim 1, characterized in that: The step of extracting multi-level phase space features from multiple sensor signals in the control system of the target device and constructing a feature time series matrix according to the extracted phase space features includes: Applying a delayed coordinate method to the time series signals of the plurality of sensors to reconstruct the phase space to obtain a high-dimensional phase space trajectory; Calculating nonlinear dynamic indicators for the high-dimensional phase space trajectory, and using adaptive sliding window technology to perform segmented calculations on the nonlinear dynamic indicators to obtain a time-varying nonlinear dynamic indicator sequence; The high-dimensional phase space trajectory, the nonlinear dynamic index and the time-varying nonlinear dynamic index sequence are used as phase space features, and a feature time series matrix is ​​constructed according to the extracted phase space features.

3. The controller circuit self-diagnosis method according to claim 1, characterized in that: Based on the characteristic timing matrix, a functional dependency graph of the controller loop of the target device is constructed and the time-varying causal relationship between the components in the functional dependency graph is calculated. By analyzing the abnormal traffic distribution in the time-varying causal relationship, a masking effect decoupling matrix is ​​obtained, including: Analyze the phase space feature correlation of each sensor in the feature timing matrix, and construct an initial functional dependency graph by combining the physical structure of the target device and the controller loop schematic diagram; Based on the time-varying sequence of nonlinear dynamic indicators in the characteristic time series matrix, the improved PC algorithm of the conditional mutual information criterion is used to determine the direction and strength of the causal association between components, and the initial functional dependency graph is optimized based on the direction and strength of the causal association; Based on the optimized functional dependency graph, the characteristic time series matrix is ​​divided into multiple time periods, and the time-varying Granger causality analysis method is applied to each time period, a vector autoregression model is established and statistics are calculated to obtain the time-varying causal relationship that characterizes the time-varying changes between the components; The causal strength in the time-varying causal relationship is regarded as the abnormal traffic distribution. According to the abnormal traffic conservation principle, the abnormal absorption rate is calculated for each component to construct the masking effect decoupling matrix.

4. The controller circuit self-diagnosis method according to claim 3, characterized in that: The causal strength in the time-varying causal relationship is regarded as abnormal traffic distribution, and according to the abnormal traffic conservation principle, the abnormal absorption rate is calculated for each component to construct the masking effect decoupling matrix, which includes: A directed weighted graph structure is established for the time-varying causal relationship, wherein nodes represent components of a target device, directions of edges represent directions of causal relationships, and weights of edges represent causal strength, wherein the causal strength is defined as abnormal traffic distribution between components; For each node in the directed weighted graph, the sum of abnormal flow distribution of the inflow node and the sum of abnormal flow distribution of the outflow node are calculated to obtain the abnormal absorption rate of each node; A masking effect decoupling matrix is ​​constructed according to the abnormal absorption rate and the causal strength, rows and columns of the masking effect decoupling matrix correspond to components, and matrix element values ​​are determined according to the abnormal absorption rate of the node and the relative causal strength between two nodes.

5. The controller circuit self-diagnosis method according to claim 1, characterized in that: The constructing of a fault interaction potential field based on the characteristic timing matrix and the masking effect decoupling matrix and analyzing the fault interaction characteristics of the fault interaction potential field at different time scales to generate a fault interaction feature map comprises: Reorganize the feature time series matrix into a third-order tensor according to the three dimensions of the number of sensors, the number of time windows, and the feature dimension, and apply the constrained Tucker decomposition algorithm to the third-order tensor to extract the spatial pattern matrix representing the sensor spatial pattern and the core tensor representing the pattern interaction; Calculating the Hadamard product of the spatial mode matrix and the masking effect decoupling matrix, wherein the Hadamard product is used to adjust the influence weights of each component in the spatial mode matrix, combining the adjusted spatial mode matrix with the core tensor, and defining a fault interaction potential field; Acquire thermodynamic parameter measurement values ​​of the target device, calculate a thermodynamic compensation factor according to a thermodynamic model of the device, and perform compensation adjustment on the fault interaction potential field according to the thermodynamic compensation factor to obtain a compensated fault interaction potential field; The compensated fault interaction potential field is subjected to multi-scale decomposition by applying wavelet transform, fault interaction components of each scale are extracted, and the fault interaction components are mapped into a fault interaction feature spectrum.

6. The controller circuit self-diagnosis method according to claim 5, characterized in that: The applying wavelet transform to the compensated fault interaction potential field to perform multi-scale decomposition, extracting fault interaction components of each scale, and mapping the fault interaction components into a fault interaction feature map comprises: According to a preset wavelet basis function, a discrete wavelet transform is applied to the compensated fault interaction potential field to decompose the fault interaction potential field into wavelet decomposition coefficients of different scales; Calculating energy distribution of wavelet decomposition coefficients of different scales to obtain energy contribution rate of each scale layer, and determining the time scale of occurrence of major fault interactions according to the energy contribution rate; Extracting the phase information of the time scale at which the main fault interaction occurs, and identifying the time delay characteristics and synchronization characteristics of the fault interaction of different components through cross-scale phase coherence analysis; The energy distribution, phase information, time delay characteristics and synchronization characteristics are converted into a graph structure representation as fault interaction components, and a community detection algorithm is applied to the graph structure representation to map it into a fault interaction feature spectrum.

7. The controller circuit self-diagnosis method according to claim 1, characterized in that: The inputting of the fault interaction feature map into a preset working condition perception conditional probability network and applying a hybrid reasoning framework to output a comprehensive diagnosis result of multiple faults includes: By monitoring the key operating parameters of the target equipment and combining the fuzzy clustering algorithm to map the key operating parameters to the predefined operating condition space, the current operating condition category is determined; According to the determined current working condition category, the corresponding conditional probability network model is called from the pre-trained conditional probability network library, and the fault interaction feature map is converted into a network input feature vector, which is input into the conditional probability network, and the posterior probability of each fault type and the combination of fault types is calculated through forward reasoning; Applying the Demster-Schaffer evidence theory framework to the posterior probability, defining a basic probability distribution function for each fault type and combination, and calculating an uncertainty factor based on the operating status of the target device; The basic probability distribution function and the uncertainty factor are used to calculate the comprehensive trust through the Demster combination rule, and the comprehensive diagnosis result is generated by combining the interaction characteristics in the fault interaction feature map.

8. A controller circuit self-diagnosis device, characterized in that: The controller loop self-diagnosis device comprises: A feature extraction module is used to collect direct thermal feature parameters, indirect thermal feature parameters and coolant quality feature parameters of the air compressor of the fuel cell in layers to obtain a multi-dimensional thermal data set of the air compressor; A causal analysis module, used for extracting features from the multidimensional thermal data set, inputting the extracted feature data set into a neural network model based on deep transfer learning, and obtaining a corresponding heat dissipation efficiency attenuation prediction model; An interactive identification module, used to establish a multi-objective optimization equation for the air compressor based on the multi-dimensional thermal data set and the heat dissipation efficiency attenuation prediction model, and solve the multi-objective optimization equation to obtain a heat dissipation control strategy set; The fault diagnosis module is used to dynamically partition the heat dissipation system of the air compressor, and coordinately adjust the flow rate and flow direction of the coolant in each area of ​​the dynamic partition according to the heat dissipation control strategy set to achieve the heat dissipation of the air compressor.

9. A controller circuit self-diagnosis device, characterized in that: The controller loop self-diagnosis device comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the controller loop self-diagnosis device to perform the steps of the controller loop self-diagnosis method as described in any one of claims 1-7.

Images