Primary helium fan health state evaluation system and method based on state estimation

Abstract

The invention discloses a primary helium fan health state evaluation system and method based on state estimation, and relates to the technical field of nuclear power station equipment state monitoring and health management, and the system comprises a data acquisition unit, a feature extraction unit, a state estimation unit and a health evaluation unit. According to the system, strategies of multi-sensor data fusion, mechanism data hybrid modeling, active incremental learning and the like are adopted, and an end-to-end health state evaluation model is constructed; and the block chain technology is introduced to realize data security sharing and trusted computing. According to the system, multi-source heterogeneous data can be comprehensively utilized, the health state of the main helium fan can be automatically and accurately evaluated in real time, and a reliable basis is provided for intelligent operation and maintenance decision making. The safety and economical efficiency of the nuclear power station are improved, and intelligent development of nuclear power equipment is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of nuclear power plant equipment status monitoring and health management, and in particular to a main helium blower health status assessment system and method based on status estimation. Background Art

[0002] The main helium blower is a key device in a nuclear power plant. Its function is to transfer the heat generated by the reactor core to the steam turbine through the helium circulation, driving the steam turbine to generate electricity. The safe and reliable operation of the main helium blower directly affects the power generation efficiency and economy of the nuclear power plant. In order to ensure the normal operation of the main helium blower, it is necessary to monitor and evaluate its status in real time, timely discover and warn potential fault hazards, and take corresponding operation and maintenance measures.

[0003] Traditional methods for monitoring and evaluating the condition of main helium blowers mainly rely on a single type of sensor data, such as vibration, temperature, etc. These methods usually use simple analysis methods such as threshold comparison or empirical models, which are difficult to fully reflect the health status of the main helium blowers. With the development of sensor technology and data acquisition systems, the monitoring data of main helium blowers is becoming increasingly rich and diverse. How to make full use of these heterogeneous data, extract effective health degradation features, and build accurate and reliable state estimation and evaluation models has become a key issue that needs to be solved urgently.

[0004] In addition, most of the existing main helium blower status monitoring systems adopt a centralized data management model, which poses risks such as data security and privacy leakage, and is not conducive to multi-party data sharing and collaboration. How to achieve transparency, traceability and verifiability of the status assessment process while ensuring data security and reliability is also a practical problem that needs to be solved urgently. In response to the above technical problems, there is an urgent need for a new main helium blower health status assessment system that can fully tap the value of multi-source heterogeneous monitoring data, accurately assess the health status of equipment, and avoid the drawbacks of centralized data management, providing reliable support for intelligent operation and maintenance decisions. Summary of the invention

[0005] In view of the above existing problems, the present invention adopts key technologies such as multi-sensor data fusion, combination of mechanism modeling and data drive, active learning and incremental learning, blockchain to ensure data security, etc., which can comprehensively, dynamically, intelligently and safely monitor the health status of the main helium blower, timely discover and warn abnormalities, and provide a reliable basis for status maintenance decisions. This system has good real-time, accuracy, adaptability and robustness, can significantly improve the safety and economy of nuclear power plants, and has broad application prospects.

[0006] In order to solve the above technical problems, a main helium blower health status assessment system based on state estimation is proposed, which is characterized by comprising a data acquisition unit, a feature extraction unit, a state estimation unit and a health assessment unit.

[0007] The data acquisition unit collects comprehensive status monitoring data and performs blockchain data management by installing various types of sensors at key locations of the main helium blower. The status monitoring data includes vibration data, temperature data, pressure data, flow data, and current data.

[0008] The feature extraction unit uses signal processing and machine learning algorithms to extract features reflecting the health status of the main helium blower from the status monitoring data.

[0009] The state estimation unit comprehensively utilizes the characteristics reflecting the health state of the main helium blower to construct a health state estimation model for the main helium blower and obtain a plurality of state quantities of the main helium blower.

[0010] The health assessment unit calculates the health index of the main helium blower according to the state quantity and generates a health assessment result.

[0011] As a preferred solution of the main helium blower health status assessment system based on state estimation described in the present invention, the data acquisition unit includes digital and networked sensors to achieve real-time data transmission and remote monitoring through industrial Ethernet.

[0012] The feature extraction unit includes: extracting the features of the vibration data using a wavelet packet decomposition algorithm, an empirical mode decomposition algorithm and a singular value decomposition algorithm; extracting the features of the temperature data using a Kalman filter algorithm; and extracting the features of the current data using a Hilbert transform algorithm.

[0013] Blockchain technology is used to securely store and control access to monitoring data, cryptographic algorithms are used to encrypt data, and smart contract mechanisms are used to perform fine-grained control over data access rights for different roles.

[0014] As a preferred solution of the main helium blower health status assessment system based on state estimation described in the present invention, wherein: the state estimation unit includes constructing a health status estimation model by combining physical mechanism modeling and machine learning.

[0015] The physical mechanism modeling is to establish a mathematical model according to the structural parameters and working principles of the main helium blower, and use experimental data to perform parameter identification and model verification. The machine learning is to use support vector machine algorithm, Bayesian network algorithm, fuzzy reasoning algorithm and integrated learning algorithm to input multi-source heterogeneous features into the model to obtain the state quantity of the main helium blower.

[0016] The state estimation unit adopts an incremental learning mechanism and a hybrid active learning mechanism. When new monitoring data is input, the state estimation model is updated and fine-tuned online using a local memory storage mechanism to adapt to changes in data distribution. The hybrid active learning mechanism actively selects the most representative and informative sample points for experts to label based on the uncertainty of the model output, and adds the training set to iteratively update the model to improve the generalization ability of the model.

[0017] The health assessment unit includes, for main helium blowers of different models and working conditions, using expert knowledge and data analysis to determine an adaptive health index system, including vibration index, temperature index and efficiency index, and assigning different weight coefficients, and using weighted average and geometric average methods to calculate the health index. When the health index is lower than a preset threshold, the health assessment unit triggers an early warning, prompting the operation and maintenance personnel to take maintenance measures, and displaying the real-time working condition and health status of the main helium blower through a human-computer interaction interface.

[0018] Another object of the present invention is to provide a method for evaluating the health status of a main helium blower based on state estimation.

[0019] As a preferred solution of the main helium blower health status assessment method based on state estimation described in the present invention, the signal processing and machine learning algorithm includes: extracting the features of the vibration data using a wavelet packet decomposition algorithm, an empirical mode decomposition algorithm and a singular value decomposition algorithm: optimizing the selection of wavelet packet basis functions and the number of decomposition layers:

[0020]

[0021] Among them, WP(x,j,k) is the wavelet packet coefficient, x is the vibration signal, j is the decomposition scale, k is the frequency band index, g j,k,n is the adaptively selected wavelet packet basis function, φ j,n (x) is the scaling function.

[0022] Add noise-assisted technology to decompose non-stationary signals:

[0023]

[0024] Where x(t) is the original vibration signal, c i (t) is the eigenmode function, r N (t) is the residual signal, ò(t) is the introduced white noise, and N is the number of decomposition layers determined adaptively.

[0025] Add sliding window technology to capture local features:

[0026]

[0027] Where X is the sliding window trajectory matrix of the vibration signal, U and V are the left and right singular vector matrices, Σ is the singular value diagonal matrix, σ i are singular values ​​and r is the rank.

[0028] Finally, multi-scale feature fusion is performed, and a feature weight adaptive adjustment method based on information entropy is proposed:

[0029]

[0030] Among them, F is the fused feature vector, f i is the feature subset extracted by various methods, w i is the corresponding weight, H(f i ) is the information entropy of the feature subset, and M is the number of feature subsets.

[0031] As a preferred solution of the main helium blower health status assessment method based on state estimation described in the present invention, the signal processing and machine learning algorithm also includes extracting the characteristics of temperature data and improving the state space model through an improved adaptive Kalman filter algorithm:

[0032]

[0033] Among them, x k is the temperature state vector at time k, y k is the observed value, f(·) and h(·) are the nonlinear state transfer function and observation function respectively, w k and v k are process noise and observation noise, respectively.

[0034] Estimate adaptive noise:

[0035]

[0036] Among them, Q k and R k are the process noise covariance and observation noise covariance of the adaptive estimation, λ is the forgetting factor, K k is the Kalman gain, P k is the state estimation error covariance, F k and H k are the Jacobian matrices of the state transfer matrix and the observation matrix respectively, N is the sliding window size, is the state estimation vector at time i.

[0037] Improve the Kalman filter equation for prediction:

[0038]

[0039] Update based on forecast results:

[0040]

[0041] in, and are the prior and posterior state estimates, respectively, and P k are the prior and a posteriori estimated error covariances, respectively.

[0042] Perform multi-model fusion on the updated results:

[0043]

[0044] in, is the final fused state estimate, is the state estimate of the ith model, μ i is the corresponding weight, e i is the innovation sequence, S i is the innovation covariance, and M is the number of models.

[0045] Extract the characteristics of current data, use Hilbert transform to extract instantaneous amplitude and phase information, calculate the analytical signal of the current signal, estimate the instantaneous amplitude A(t) and phase φ(t), and reflect the energy change and frequency characteristics of the current signal:

[0046]

[0047] Among them, i(t) is the original current signal, is its Hilbert transform.

[0048] After all features are extracted, the multi-source heterogeneous features of vibration, temperature, and current are aggregated to form a high-dimensional feature vector as the input of the state estimation model. Principal component analysis and independent component analysis are used to eliminate the redundancy and correlation between features and extract the essential feature subset.

[0049] As a preferred solution of the main helium blower health status assessment method based on state estimation described in the present invention, wherein: the health status estimation model includes constructing a health status estimation model by combining physical mechanism modeling and machine learning.

[0050] The physical mechanism modeling includes: the state estimation unit adopts an innovative hybrid modeling paradigm, integrates physical mechanism modeling and data-driven modeling, increases multi-physical field coupling effects, and establishes a more comprehensive main helium fan dynamics model:

[0051]

[0052] Where q is the generalized displacement vector, M, C, G, K are the mass matrix, damping matrix, gyro matrix and stiffness matrix respectively, Ω is the rotation speed, F a 、F m 、F t They are aerodynamic force, electromagnetic force and thermal stress respectively.

[0053] Adopting an improved hybrid deep learning model:

[0054]

[0055] i t =σ(W xi x t +W hi h t-1 +b i )

[0056] f t =σ(W xf x t +W hf h t-1 +b f )

[0057] o t =σ(W xo x t +W ho h t-1 +b o )

[0058] c t =f t ⊙c t-1 +i t ⊙tanh(W xc x t +W hc h t-1 +b c )

[0059] h t =o t ⊙tanh(c t )

[0060] Among them, H (l) is the l-th layer feature matrix of GCN, To enhance the adjacency matrix, is the degree matrix, W (l) is the weight matrix, i t 、f t , o t 、c t 、h tThey are the input gate, forget gate, output gate, unit state and hidden state of LSTM respectively, and W and b are the corresponding weights and biases.

[0061] Introducing physical constraints:

[0062] L total =L data +λL physics

[0063] Among them, L total is the total loss function, L data is the data-driven loss, L physics is the physical constraint loss, and λ is the trade-off factor.

[0064] A multi-model fusion strategy based on dynamic weights is proposed:

[0065]

[0066] in, is the fused state estimate, is the estimated result of the ith model, w i (t) is the dynamic weight, e i (t) is the estimation error, α is the sensitivity parameter, and N is the number of models.

[0067] Through the domain adaptation strategy, the model can be quickly migrated between different working conditions:

[0068] L transfer =L task +βMMD(S,T)

[0069] Among them, L transfer is the transfer learning loss function, L task is the target task loss, MMD(S,T) is the maximum mean difference between the source and target domains, and β is the balance parameter.

[0070] Aiming at the complex dynamic characteristics of the main helium blower, a multi-physics coupling model is proposed, combining rotor dynamics, fluid dynamics and electromagnetic thermal coupling effects:

[0071]

[0072] Where q is the generalized displacement vector, including the displacement of the rotor, bearings and stator. M is the mass matrix, C is the damping matrix, G is the gyro matrix, K is the stiffness matrix, Ω is the speed, K b (δ) is the nonlinear stiffness term considering the bearing clearance δ, K m (i) is the electromagnetic stiffness caused by current i, F is the aerodynamic force, taking into account the influence of speed and rotation speed. m(i,δ) is the electromagnetic force, taking into account the influence of current and gap. F t (T) is thermal stress, taking into account the influence of temperature distribution. The newly introduced nonlinear force terms capture the complex nonlinear dynamic behaviors.

[0073] The machine learning includes establishing a data-driven model using a machine learning algorithm for the multi-source monitoring features extracted by the feature extraction unit. The forward calculation formula of DBN is:

[0074] h (1) =f(W (1) x+b (1) )h (l)

[0075] h (1) =f(W (l) h (l-1) +b (l) ),l=2,…,L-1

[0076]

[0077] Among them, x is the input feature, h (l) is the hidden variable of the lth layer, f(·) is the activation function, W (l) and b (l) are weight and bias parameters, To predict the state quantity, during the training process, a gradient-based optimization algorithm is used to minimize the reconstruction error and prediction error.

[0078] As a preferred solution of the main helium blower health status assessment method based on state estimation described in the present invention, the incremental learning mechanism and the hybrid active learning mechanism include: the adaptive incremental learning strategy introduces an adaptive hierarchical structure and a dynamic weight adjustment mechanism by improving the traditional DBN model:

[0079]

[0080] in, is the output of the i-th hidden unit in layer I, is the output of the jth unit in the previous layer, is the weight, is the bias, f(·) is the activation function, η(t) is the adaptive learning rate, η 0 is the initial learning rate, λ is the decay coefficient, η min is the minimum learning rate.

[0081] Local memory storage via experience replay strategy based on importance sampling:

[0082]

[0083] p i =|δ i |+ò

[0084] Among them, p(i) is the probability that sample i is selected, p i is the priority of the sample, α controls the degree of sampling deviation, δ i is the timing difference error, and ò is a small positive number.

[0085] The hybrid active learning mechanism combines uncertainty sampling and diversity sampling to propose a new sample selection criterion:

[0086]

[0087] D(x,L)=min x′∈L d(x,x′)

[0088] Among them, S is the selected sample set, U is the unlabeled sample pool, L is the labeled sample set, H(y|x,θ) is the predicted entropy of the model for sample x, reflecting uncertainty, D(x,L) is the minimum distance between sample x and the labeled sample set, reflecting diversity, λ is the balance factor, and d(·,·) is the distance measurement function.

[0089] Introducing fuzzy cognitive graph to integrate expert knowledge:

[0090]

[0091] Among them, A i (t) is the activation value of concept node i at time t, w ji is the causal weight from concept j to i, f(·) is the sigmoid function, and λ is the steepness parameter.

[0092] Propose an adaptive update mechanism based on performance degradation detection:

[0093]

[0094] τ(t)=τ 0 ·(1-e -γt )

[0095] Among them, Δ is the average prediction error, y i and are the true value and predicted value respectively, N is the number of samples, τ(t) is the dynamic threshold, τ 0 is the initial threshold, γ is the adjustment coefficient, and the model update is triggered when Δ exceeds τ(t).

[0096] As a preferred solution of the main helium fan health status assessment method based on state estimation described in the present invention, wherein: the calculation of the health index of the main helium fan includes constructing an adaptive health index system, considering the importance and influence of vibration, temperature, and efficiency state quantities, using the basic probability assignment method in evidence theory, converting the state quantity output by the state estimation unit into a confidence distribution of health, and using the Dempster combination rule to fuse evidence for the BPA of different health indicators to obtain the confidence distribution of the overall health status of the main helium fan. Suppose the BPA of the i-th indicator is m i , the focal element is A i1 ,A i2 ,...,A ik , then the Dempster combination rule is:

[0097]

[0098] in, represents the Dempster combination, A is any focal element after fusion, A i is the focal element of each indicator, Represents the empty set.

[0099] Using the fuzzy logic reasoning method, the fused confidence distribution is converted into a language value description, and the weighted sum of the membership centers of all fuzzy levels is taken to obtain a health index between 0 and 100. When the health index is lower than 60, the system automatically triggers an early warning, prompting the operation and maintenance personnel to take corresponding measures.

[0100] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of a method for evaluating the health status of a main helium blower based on state estimation are implemented.

[0101] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of a method for evaluating the health status of a main helium blower based on state estimation are implemented.

[0102] The beneficial effects of the present invention are as follows: a multi-sensor data fusion strategy is adopted to make full use of multi-source heterogeneous monitoring data such as vibration, temperature, and current to comprehensively characterize the health status of the main helium fan, overcome the limitations of the traditional single data source, and improve the accuracy and reliability of state assessment. Advanced signal processing and feature extraction algorithms are introduced to extract sensitive health degradation features from multiple angles such as time-frequency domain, energy domain, and modulation domain, and a high-dimensional, high-quality feature set is constructed, which lays a good foundation for subsequent modeling. An innovative hybrid modeling paradigm is proposed, which combines physical mechanism modeling with data-driven modeling, which not only makes full use of existing physical knowledge and experience, but also learns and discovers new laws from data, realizes the perfect unity of the mechanism explanation and data adaptability of state estimation, and greatly improves the robustness and generalization of the model. Incremental learning and active learning mechanisms are introduced in the state estimation model, giving the model the ability to continuously evolve and self-optimize. Overall, the main helium fan health status assessment system constructed by the present invention realizes the seamless integration and process closed loop of functions such as state monitoring, feature extraction, state estimation, health assessment, and data management, forming a complete end-to-end solution. The system can automatically, in real time, accurately and safely assess the health status of the main helium blower, promptly warn of abnormal risks, and provide visual health portraits and operation and maintenance suggestions, greatly improving the safety, reliability and economy of nuclear power operations, reducing the risk of unplanned downtime, and has significant social benefits and market prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0104] Figure 1 A structural principle block diagram of a main helium blower health status assessment system based on state estimation provided by an embodiment of the present invention.

[0105] Figure 2 A schematic diagram of a data acquisition unit of a main helium blower health status assessment system based on state estimation provided by an embodiment of the present invention.

[0106] Figure 3 A schematic diagram of a feature extraction unit of a main helium blower health status assessment system based on state estimation provided by an embodiment of the present invention.

[0107] Figure 4 A schematic diagram of a state estimation unit of a main helium blower health status assessment system based on state estimation is provided in accordance with an embodiment of the present invention.

[0108] Figure 5 A schematic diagram of a health assessment unit of a main helium blower health status assessment system based on state estimation provided by an embodiment of the present invention.

[0109] Figure 6 A schematic diagram of blockchain data management of a main helium blower health status assessment system based on state estimation is provided for one embodiment of the present invention. DETAILED DESCRIPTION

[0110] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0111] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0112] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is mutually exclusive with other embodiments, either individually or selectively.

[0113] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0114] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0115] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0116] Example 1, reference Figure 1-Figure 6 , which is the first embodiment of the present invention, provides a main helium blower health status assessment system based on state estimation, including a data acquisition unit, a feature extraction unit, a state estimation unit and a health assessment unit.

[0117] The data acquisition unit collects comprehensive status monitoring data and performs blockchain data management by installing various types of sensors at key locations of the main helium blower. The status monitoring data includes vibration data, temperature data, pressure data, flow data, and current data.

[0118] The feature extraction unit uses signal processing and machine learning algorithms to extract features reflecting the health status of the main helium blower from the status monitoring data.

[0119] The state estimation unit comprehensively utilizes the characteristics reflecting the health state of the main helium blower to construct a health state estimation model for the main helium blower and obtain the state quantity of the main helium blower.

[0120] The health assessment unit calculates the health index of the main helium blower according to the state quantity and generates a health assessment result.

[0121] Furthermore, the data acquisition unit adopts digital and networked sensors to achieve real-time data transmission and remote monitoring through industrial Ethernet.

[0122] In the feature extraction unit, wavelet packet decomposition algorithm, empirical mode decomposition algorithm and singular value decomposition algorithm are used to extract the features of the vibration data, Kalman filter algorithm is used to extract the features of the temperature data, and Hilbert transform algorithm is used to extract the features of the current data.

[0123] Furthermore, blockchain technology is used to securely store and control access to monitoring data, cryptographic algorithms are used to encrypt data, and smart contract mechanisms are used to perform fine-grained control over data access rights of different roles to ensure the confidentiality and integrity of the data.

[0124] It should be noted that the state estimation unit constructs the state estimation model by combining physical mechanism modeling and machine learning. The physical mechanism modeling establishes a mathematical model according to the structural parameters and working principle of the main helium blower, and uses experimental data for parameter identification and model verification; the machine learning adopts support vector machine algorithm, Bayesian network algorithm, fuzzy inference algorithm and integrated learning algorithm, and inputs multi-source heterogeneous features into the model to obtain the state quantity of the main helium blower.

[0125] Furthermore, the state estimation unit adopts an incremental learning mechanism. When new monitoring data arrives, the state estimation model is updated and fine-tuned online using a local memory storage mechanism to adapt to changes in data distribution. The state estimation unit also adopts an active learning mechanism. According to the uncertainty of the model output, it actively selects the most representative and informative sample points for experts to label, and adds the training set to iteratively update the model to improve the generalization ability of the model.

[0126] It should also be noted that the health assessment unit uses expert knowledge and data analysis to determine an adaptive health index system for main helium blowers of different models and operating conditions, including vibration indicators, temperature indicators and efficiency indicators, and assigns different weight coefficients, and uses weighted average or geometric average methods to calculate the health index; when the health index is lower than a preset threshold, the health assessment unit triggers an early warning to prompt the operation and maintenance personnel to take maintenance measures; at the same time, the real-time operating conditions and health status of the main helium blower are displayed in the form of dashboards, trend charts and radar charts through the human-computer interaction interface.

[0127] Furthermore, the health assessment unit adopts a service-oriented architecture, realizes the integration and calling of status assessment functions through a Web service interface, and supports interconnection with other systems.

[0128] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0129] Embodiment 2, the third embodiment of the present invention, is different from the first two embodiments in that:

[0130] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the 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 for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0131] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0132] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0133] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0134] Example 3, reference Figure 1-Figure 6 , which is the fourth embodiment of the present invention, provides a main helium blower health status assessment method based on state estimation, comprising:

[0135] like Figure 1 As shown, this embodiment provides a main helium blower health status assessment system based on state estimation, including a data acquisition unit, a feature extraction unit, a state estimation unit, a health assessment unit and a human-computer interaction interface.

[0136] Furthermore, Figure 2 As shown in the figure, the data acquisition unit uses digital and networked sensors, which are arranged in key parts of the main helium blower, such as bearings, stator windings, etc., to collect comprehensive status monitoring data. Among them, vibration data is collected by acceleration sensor with a frequency range of 0.1Hz~10kHz; temperature data is collected by platinum resistance thermometer with a range of -200~500℃; pressure data is collected by strain gauge pressure sensor with a range of 0~10MPa; flow data is collected by vortex flowmeter with a range of 0~1000m 3 / h; the current data is collected by Hall current sensor with a range of 0 to 1000A. Each sensor transmits the collected digital signal to the database server of the status assessment system in real time through industrial Ethernet to achieve remote monitoring.

[0137] It should be noted that if Figure 3 As shown, the feature extraction unit uses a variety of signal processing and machine learning algorithms to extract sensitive features reflecting the health status of the main helium blower from the collected original monitoring data.

[0138] For the analysis of vibration signals, the present invention proposes an improved multi-scale fusion feature extraction method, which combines the advantages of wavelet packet decomposition, empirical mode decomposition and singular value decomposition to achieve more comprehensive and precise feature extraction of vibration signals.

[0139] The characteristics of the vibration data are extracted by using wavelet packet decomposition algorithm, empirical mode decomposition algorithm and singular value decomposition algorithm, and the selection of wavelet packet basis function and decomposition layer number is optimized:

[0140]

[0141] Among them, WP(x,j,k) is the wavelet packet coefficient, x is the vibration signal, j is the decomposition scale, k is the frequency band index, g j,k,n is the adaptively selected wavelet packet basis function, φ j,n (x) is the scaling function.

[0142] Add noise-assisted technology to decompose non-stationary signals:

[0143]

[0144] Where x(t) is the original vibration signal, c i (t) is the eigenmode function, r N (t) is the residual signal, ò(t) is the introduced white noise, and N is the number of decomposition layers determined adaptively.

[0145] Add sliding window technology to capture local features:

[0146]

[0147] Where X is the sliding window trajectory matrix of the vibration signal, U and V are the left and right singular vector matrices, Σ is the singular value diagonal matrix, σ i are singular values ​​and r is the rank.

[0148] Finally, multi-scale feature fusion is performed, and a feature weight adaptive adjustment method based on information entropy is proposed:

[0149]

[0150] Among them, F is the fused feature vector, f i is the feature subset extracted by various methods, w i is the corresponding weight, H(f i ) is the information entropy of the feature subset, and M is the number of feature subsets.

[0151] Through the above-mentioned improved mathematical model, the present invention realizes more accurate and comprehensive feature extraction of vibration signals, can better capture the health status information of the equipment, and provide more reliable input for subsequent state estimation and health assessment.

[0152] For the analysis of temperature signals, this paper proposes an improved adaptive Kalman filter algorithm to achieve more accurate dynamic estimation and prediction, introduces adaptive noise estimation and multi-model fusion strategy, which can better cope with the nonlinear and slowly changing characteristics of temperature signals and improve the state space model:

[0153]

[0154] Among them, x k is the temperature state vector at time k, y k is the observed value, f(·) and h(·) are the nonlinear state transfer function and observation function respectively, w k and v k are process noise and observation noise, respectively.

[0155] Estimate adaptive noise:

[0156]

[0157] Among them, Q k and R k are the process noise covariance and observation noise covariance of the adaptive estimation, λ is the forgetting factor, K k is the Kalman gain, P k is the state estimation error covariance, F k and H k are the Jacobian matrices of the state transfer matrix and the observation matrix respectively, N is the sliding window size, is the state estimation vector at time i.

[0158] Improve the Kalman filter equation for prediction:

[0159]

[0160] Update based on forecast results:

[0161]

[0162] in, and are the prior and posterior state estimates, respectively, and P k are the prior and a posteriori estimated error covariances, respectively.

[0163] Perform multi-model fusion on the updated results:

[0164]

[0165] in, is the final fused state estimate, is the state estimate of the ith model, μi is the corresponding weight, e i is the innovation sequence, S i is the innovation covariance, and M is the number of models.

[0166] Through the above improved mathematical model, the present invention realizes more accurate and robust dynamic estimation and prediction of temperature signals. Adaptive noise estimation can better adapt to the dynamic changes of the system, and the multi-model fusion strategy can effectively handle the nonlinear characteristics of temperature signals, thereby improving the dynamic responsiveness and estimation accuracy of temperature characteristics, and providing more reliable temperature characteristic input for the health status assessment of the main helium blower.

[0167] Extract the characteristics of current data, use Hilbert transform to extract instantaneous amplitude and phase information, calculate the analytical signal of the current signal, estimate the instantaneous amplitude A(t) and phase φ(t), and reflect the energy change and frequency characteristics of the current signal:

[0168]

[0169] Among them, i(t) is the original current signal, is its Hilbert transform.

[0170] After all features are extracted, the multi-source heterogeneous features of vibration, temperature, and current are aggregated to form a high-dimensional feature vector as the input of the state estimation model. Dimensionality reduction methods such as principal component analysis and independent component analysis are used to eliminate redundancy and correlation between features, extract the most essential feature subset, and reduce the complexity of subsequent modeling.

[0171] In summary, the feature extraction unit of the present invention integrates a variety of advanced signal processing and feature extraction methods, and can comprehensively mine the state degradation information contained in multi-source monitoring data. In particular, the introduction of innovative algorithms such as pitch detection and Hilbert transform greatly enriches the types and dimensions of the main helium fan fault characteristics, laying a solid data foundation for subsequent state estimation and health assessment.

[0172] It should be noted that if Figure 4 As shown in the figure, a health status estimation model is constructed by combining physical mechanism modeling and machine learning.

[0173] The physical mechanism modeling includes: the state estimation unit adopts an innovative hybrid modeling paradigm, integrates physical mechanism modeling and data-driven modeling, increases multi-physical field coupling effects, and establishes a more comprehensive main helium fan dynamics model:

[0174]

[0175] Where q is the generalized displacement vector, M, C, G, K are the mass matrix, damping matrix, gyro matrix and stiffness matrix respectively, Ω is the rotation speed, F a 、F m 、F t They are aerodynamic force, electromagnetic force and thermal stress respectively.

[0176] Adopting an improved hybrid deep learning model:

[0177]

[0178] i t =σ(W xi x t +W hi h t-1 +b i )

[0179] f t =σ(W xf x t +W hf h t-1 +b f )

[0180] o t =σ(W xo x t +W ho h t-1 +b o )

[0181] c t =f t ⊙c t-1 +i t ⊙tanh(W xc x t +W hc h t-1 +b c )

[0182] h t =o t ⊙tanh(c t )

[0183] Among them, H (l) is the l-th layer feature matrix of GCN, To enhance the adjacency matrix, is the degree matrix, W (l) is the weight matrix, i t 、f t , o t 、c t 、h tThey are the input gate, forget gate, output gate, unit state and hidden state of LSTM respectively, and W and b are the corresponding weights and biases.

[0184] Introducing physical constraints:

[0185] L total =L data +λL physics

[0186] Among them, L total is the total loss function, L data is the data-driven loss, L physics is the physical constraint loss, and λ is the trade-off factor.

[0187] A multi-model fusion strategy based on dynamic weights is proposed:

[0188]

[0189] in, is the fused state estimate, is the estimated result of the ith model, w i (t) is the dynamic weight, e i (t) is the estimation error, α is the sensitivity parameter, and N is the number of models.

[0190] Through the domain adaptation strategy, the model can be quickly migrated between different working conditions:

[0191] L transfer =L task +βMMD(S,T)

[0192] Among them, L transfer is the transfer learning loss function, L task is the target task loss, MMD(S,T) is the maximum mean difference between the source and target domains, and β is the balance parameter.

[0193] Through the above-mentioned improved hybrid modeling paradigm, the present invention achieves a more accurate, robust and interpretable estimation of the health status of the main helium blower. The deep learning model guided by physical knowledge not only retains the interpretability of the physical model, but also has the adaptability of the data-driven model. The adaptive multi-model fusion strategy can dynamically adjust the weights of different models, improving the accuracy and reliability of state estimation. The online transfer learning mechanism enables the model to quickly adapt to different working conditions and equipment, greatly improving the generalization ability and practicality of the system.

[0194] Aiming at the complex dynamic characteristics of the main helium blower, this paper proposes an innovative multi-physics field coupling model, combining rotor dynamics, fluid dynamics and electromagnetic thermal coupling effects:

[0195]

[0196] Where q is the generalized displacement vector, including the displacement of the rotor, bearings and stator. M is the mass matrix, C is the damping matrix, G is the gyro matrix, K is the stiffness matrix, Ω is the speed, K b (δ) is the nonlinear stiffness term considering the bearing clearance δ, K m (i) is the electromagnetic stiffness caused by current i, F is the aerodynamic force, taking into account the influence of speed and rotation speed. m (i,δ) is the electromagnetic force, taking into account the influence of current and gap. F t (T) is thermal stress, taking into account the influence of temperature distribution. The newly introduced nonlinear force terms capture the complex nonlinear dynamic behaviors.

[0197] Compared with the traditional model, this improved model introduces K b (δ) and K m Item (i) considers the nonlinear effects of bearing clearance and electromagnetic force on system stiffness respectively. These nonlinear terms can more accurately describe the dynamic characteristics of the main helium blower under different operating conditions, especially its behavior under high speed and heavy load conditions.

[0198] Aerodynamic force F a , electromagnetic force F m and thermal stress F t It is modeled as a more complex function that takes into account the coupling effects of multiple variables. For example, the aerodynamic force is not only related to the rotational speed, but also to the displacement velocity, which can better reflect the fluid-structure interaction.

[0199] New nonlinear force terms It can capture complex dynamic behaviors that cannot be described by traditional linear models, such as chaos and bifurcation phenomena. This is crucial for predicting the behavior of the main helium blower under extreme conditions.

[0200] The model parameters are optimized through advanced parameter identification algorithms and a large amount of test data to improve the accuracy and reliability of the model. An adaptive parameter identification method based on a combination of particle swarm optimization and Kalman filtering is adopted, which can update the model parameters in real time and adapt to the dynamic changes of the equipment. A multi-scale modeling method is adopted to combine the microscopic material behavior with the macroscopic system response to improve the physical accuracy of the model. This method can consider the impact of the microstructural changes of the material on the overall system performance, especially in terms of material degradation caused by long-term operation. The finite element analysis method is introduced to perform refined modeling of complex geometric structures, which improves the accuracy of the model in predicting local stress and deformation. Combined with fault tree analysis, a dynamic fault propagation model is established, which can simulate multiple failure modes and their interactions, providing more comprehensive information for health status assessment.

[0201] Through this improved physical mechanism model, the present invention achieves a high-precision description of the dynamic characteristics of the main helium blower. The model can not only accurately predict key state quantities such as vibration, temperature and efficiency under normal operating conditions, but also simulate the system response under various abnormal and fault conditions. This lays a solid theoretical foundation for subsequent health status assessment and improves the accuracy and reliability of status monitoring and fault diagnosis.

[0202] At the same time, a large amount of test data is used to continuously identify parameters and verify the model of the physical model, and an adaptive optimization mechanism is established. This enables the model to continuously learn and update, adapt to changes in the equipment throughout its life cycle, and maintain high prediction accuracy. This hybrid method of "physical model + data-driven" greatly improves the generalization ability and practicality of the model, and provides strong support for the intelligent operation and maintenance of the main helium blower.

[0203] The machine learning includes proposing a multi-task learning model based on a deep belief network (DBN). A machine learning algorithm is used to establish a data-driven model for the multi-source monitoring features extracted by the feature extraction unit. The forward calculation formula of DBN is:

[0204] h (1) =f(W (1) x+b (1) )h (l)

[0205] h (1) =f(W (l) h (l-1) +b (l) ),l=2,…,L-1

[0206]

[0207] Among them, x is the input feature, h(l) is the first layer hidden variable, f(·) is the activation function, W (l) and b (l) are weight and bias parameters, In order to predict the state quantity, during the training process, a gradient-based optimization algorithm is used to minimize the reconstruction error and prediction error, and L1 regularization and Dropout technology are introduced to improve the sparsity and generalization of the model.

[0208] Furthermore, to improve the accuracy and real-time performance of state estimation, the present invention introduces two innovative strategies: adaptive incremental learning and hybrid active learning. These strategies not only improve the real-time performance and accuracy of the model, but also enhance its adaptability and interpretability.

[0209] The adaptive incremental learning strategy improves the traditional DBN model by introducing an adaptive hierarchical structure and a dynamic weight adjustment mechanism:

[0210]

[0211] η(t)=η 0 exp(-λt)+η min

[0212] in, is the output of the ith hidden unit in the lth layer, is the output of the jth unit in the previous layer, is the weight, is the bias, f(·) is the activation function, η(t) is the adaptive learning rate, η 0 is the initial learning rate, λ is the decay coefficient, η min is the minimum learning rate.

[0213] Local memory storage via experience replay strategy based on importance sampling:

[0214]

[0215] p i =|δ i |+ò

[0216] Among them, p(i) is the probability that sample i is selected, p i is the priority of the sample, α controls the degree of sampling deviation, δ i is the timing difference error, and ò is a small positive number.

[0217] The hybrid active learning mechanism combines uncertainty sampling and diversity sampling to propose a new sample selection criterion:

[0218]

[0219] D(x,L)=min x′∈L d(x,x′)

[0220] Among them, S is the selected sample set, U is the unlabeled sample pool, L is the labeled sample set, H(y|x,θ) is the predicted entropy of the model for sample x, reflecting uncertainty, D(x,L) is the minimum distance between sample x and the labeled sample set, reflecting diversity, λ is the balance factor, and d(·,·) is the distance measurement function.

[0221] Introducing fuzzy cognitive graph to integrate expert knowledge:

[0222]

[0223] Among them, A i (t) is the activation value of concept node i at time t, w ji is the causal weight from concept j to i, f(·) is the sigmoid function, and λ is the steepness parameter.

[0224] Propose an adaptive update mechanism based on performance degradation detection:

[0225]

[0226] τ(t)=τ 0 ·(1-e -γt )

[0227] Among them, Δ is the average prediction error, y i and are the true value and predicted value respectively, N is the number of samples, τ(t) is the dynamic threshold, τ 0 is the initial threshold, γ is the adjustment coefficient, and the model update is triggered when Δ exceeds τ(t).

[0228] Through the above-mentioned improved incremental learning and active learning strategies, the present invention realizes the continuous optimization and adaptive updating of the state estimation model. The adaptive incremental learning strategy can adjust the model structure and parameters according to the dynamic changes of the data distribution to maintain the optimal prediction performance. The hybrid active learning strategy effectively selects the most informative samples for labeling by balancing uncertainty and diversity. The expert knowledge fusion mechanism converts qualitative expert experience into quantitative model input, improving the interpretability of the model. The dynamic model update strategy ensures real-time monitoring and timely optimization of model performance. This "expert + machine" interactive learning paradigm greatly improves the accuracy, real-time and adaptability of state estimation, and provides strong support for the health status assessment of the main helium blower.

[0229] In summary, the state estimation unit adopts a hybrid modeling paradigm that combines physical mechanism modeling with data-driven modeling, and promotes incremental learning with active learning, to build a comprehensive, dynamic, and intelligent health state estimation model for the main helium blower. This model makes full use of multi-source heterogeneous data and prior knowledge, overcomes the shortcomings of traditional single modeling methods, and achieves significant improvements in accuracy, real-time performance, and interpretability, providing a reliable basis for subsequent health assessment and maintenance decisions. The innovations of this unit are mainly reflected in: the multi-task learning framework of deep belief networks, the online update strategy of incremental learning, the interactive optimization method of active learning, etc., which have strong theoretical significance and practical value.

[0230] It should also be noted that if Figure 5 As shown in the figure, the purpose of the health assessment unit is to comprehensively judge the overall health level of the main helium blower based on the multiple state quantities output by the state estimation unit, and generate intuitive and explainable health assessment results to provide decision support for operation and maintenance personnel. This unit adopts a multi-attribute decision-making method based on evidence theory and fuzzy logic, which overcomes the subjectivity and uncertainty of the traditional weighted average method and can evaluate the health status of the equipment more comprehensively, objectively and reasonably.

[0231] First, for main helium blowers of different models and working conditions, a set of adaptive health index system is constructed by combining analytic hierarchy process and expert experience. This index system fully considers the importance and influence of multiple state quantities such as vibration, temperature, efficiency, etc., and performs reasonable classification and quantification. For example, the vibration state can be divided into five levels: excellent, good, general, warning, and critical, and the health score is represented by a value between 0 and 100. Similarly, the fuzzy membership function can be used to describe the fuzzy levels of normal, high, and too high temperature states.

[0232] Secondly, for each health indicator, the Basic Probability Assignment (BPA) method in evidence theory is used to convert the state quantity output by the state estimation unit into a confidence distribution of health. For example, for the vibration state, five confidence functions can be defined, which respectively represent the probability of being excellent, good, general, warning, and critical, satisfying the normalization condition. This BPA construction method not only takes into account the numerical size of the state quantity, but also incorporates the uncertainty and conflict of the monitoring data, and can more reasonably describe the health status of the equipment. The present invention creatively proposes an adaptive confidence rule, which automatically optimizes the parameters of BPA based on historical operating data and expert knowledge, and improves the accuracy of its confidence distribution.

[0233] For BPA of different health indicators, the Dempster combination rule is used to fuse the evidence and obtain the confidence distribution of the overall health status of the main helium blower.

[0234] Let the BPA of the i-th indicator be m i , the focal element is A i1 ,A i2 ,...,A ik , then the Dempster combination rule is:

[0235]

[0236] in, represents the Dempster combination, A is any focal element after fusion, A i is the focal element of each indicator, Represents an empty set. Through evidence fusion, multi-source information can be comprehensively utilized to reduce the uncertainty of a single indicator and improve the credibility of health assessment.

[0237] In order to intuitively display the health assessment results, this unit uses fuzzy logic reasoning to convert the fused confidence distribution into a language value description and health index. For example, a set of fuzzy rules can be defined to infer language descriptions such as "the health status of the main helium fan is good, but there are slight vibration anomalies, and monitoring needs to be strengthened" based on the membership size of the confidence distribution at each fuzzy level. At the same time, the weighted sum of the membership centers of each fuzzy level can be used to obtain a health index between 0 and 100 to quantify the overall health level of the equipment. When the health index is lower than 60, the system automatically triggers an early warning, prompting the operation and maintenance personnel to take corresponding measures.

[0238] In summary, the health assessment unit adopts a multi-attribute decision-making method that combines evidence theory and fuzzy logic to construct an adaptive health indicator system and evaluation model. This model makes full use of the multi-source heterogeneous information of state estimation, overcomes the subjectivity and uncertainty of traditional methods, and can more comprehensively, objectively and reasonably judge the health status of the main helium fan, and generate intuitive and explainable evaluation results, providing strong support for operation and maintenance decisions. The innovations of this unit are mainly reflected in: the BPA construction method of adaptive confidence rules, the evidence fusion strategy of Dempster combination rules, the language description method of fuzzy logic reasoning, etc., which greatly improves the intelligence level and practical value of health assessment, and has broad promotion prospects in engineering applications.

[0239] It should also be noted that if Figure 6 As shown in the figure, in order to ensure the security of monitoring data, this system uses blockchain technology to encrypt and store data and perform fine-grained access control. After the data is uploaded to the chain, it is encrypted using cryptographic algorithms, and the access rights of different roles are managed using smart contract mechanisms, which not only ensures the confidentiality and integrity of the data, but also facilitates multi-party data sharing and collaboration.

[0240] In the main helium blower health status assessment system of the present invention, monitoring data is a key asset, and its security and privacy are of vital importance. The traditional centralized data management method has risks such as single point failure, data tampering, and privacy leakage, and it is difficult to meet the needs of data sharing and collaboration. In order to overcome these problems, this system innovatively introduces blockchain technology, using its decentralized, tamper-proof, traceable, and verifiable characteristics to achieve encrypted data storage and fine-grained access control.

[0241] Specifically, this system adopts a consortium chain architecture, which is composed of power plants, equipment manufacturers, regulatory agencies and other parties, and jointly maintains a distributed ledger. Each node stores a complete copy of the data, and the consistency and credibility of the data are ensured through a consensus mechanism.

[0242] When new monitoring data is generated, it is first encrypted, then packaged into a transaction and submitted to the blockchain network. After consensus verification, the transaction is recorded in a new block and linked to the previous block, forming an unalterable chain structure. This distributed storage method can effectively prevent data from being forged or tampered with, ensuring the integrity and traceability of the data.

[0243] In terms of data encryption, this system adopts attribute-based encryption (ABE) scheme, which can achieve fine-grained access control. ABE generates encryption keys and access policies based on user attributes: identity, role, permissions, etc. Only users who meet the policy requirements can decrypt data. For example, an access policy "power plant administrator AND equipment manufacturer engineer" can be defined, and only users with both attributes can access the relevant data. The encryption process of ABE can be expressed as:

[0244] CT=Encrypt(PK,M,A)

[0245] Among them, CT is ciphertext, PK is public key, M is plaintext data, and A is access policy.

[0246] The decryption process can be expressed as:

[0247] M=Decrypt(SK,CT)

[0248] Among them, SK is the user's private key that satisfies access policy A.

[0249] Through ABE, the system can achieve fine-grained authorized access to data of different types and sensitivity levels, greatly improving the flexibility and security of data sharing.

[0250] In addition, to further protect user privacy, this system also uses zero-knowledge proof (ZKP) technology. ZKP allows the prover to prove to the verifier that a statement is true without revealing any information other than that the statement is true. This is very useful for identity authentication and data verification without leaking private data. For example, a device manufacturer can prove that the monitoring data it provides is authentic without disclosing technical details.

[0251] The ZKP process can be expressed as:

[0252] PK{(x):y=g x}

[0253] Among them, PK represents the proof process, x is private knowledge (such as device parameters), y is public knowledge (such as monitoring data), and g is the generator. Through ZKP, this system can maximize the privacy and security of all parties while ensuring data credibility.

[0254] In summary, this system uses blockchain technology and encryption algorithms to achieve distributed storage, encryption protection, fine-grained authorized access and privacy protection of monitoring data, effectively improving the security, reliability and flexibility of data management, and providing a trusted environment for multi-party collaboration and data sharing. It is an important innovative application in the field of industrial Internet and intelligent manufacturing.

[0255] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A main helium blower health status assessment system based on state estimation, characterized by: It includes a data acquisition unit, a feature extraction unit, a state estimation unit and a health assessment unit; The data acquisition unit collects comprehensive status monitoring data and performs blockchain data management by installing various types of sensors at key locations of the main helium blower. The status monitoring data includes vibration data, temperature data, pressure data, flow data, and current data. The feature extraction unit extracts features reflecting the health status of the main helium blower from the status monitoring data using signal processing and machine learning algorithms; The state estimation unit comprehensively utilizes the characteristics reflecting the health state of the main helium blower to construct a health state estimation model of the main helium blower and obtain the state quantity of the main helium blower; The health assessment unit calculates the health index of the main helium blower according to the state quantity and generates a health assessment result.

2. A main helium blower health status assessment system based on state estimation according to claim 1, characterized in that: The data acquisition unit includes a digital and networked sensor to achieve real-time data transmission and remote monitoring through industrial Ethernet; The feature extraction unit includes: extracting the features of the vibration data by using a wavelet packet decomposition algorithm, an empirical mode decomposition algorithm and a singular value decomposition algorithm; extracting the features of the temperature data by using a Kalman filter algorithm; and extracting the features of the current data by using a Hilbert transform algorithm; Blockchain technology is used to securely store and control access to monitoring data, cryptographic algorithms are used to encrypt data, and smart contract mechanisms are used to perform fine-grained control over data access rights for different roles.

3. A main helium blower health status assessment system based on state estimation as claimed in claim 2, characterized in that: The state estimation unit includes constructing a health state estimation model by combining physical mechanism modeling and machine learning; The physical mechanism modeling is to establish a mathematical model according to the structural parameters and working principle of the main helium blower, and use test data to perform parameter identification and model verification. The machine learning is to use support vector machine algorithm, Bayesian network algorithm, fuzzy reasoning algorithm and integrated learning algorithm to input multi-source heterogeneous features into the model to obtain the state quantity of the main helium blower; The state estimation unit adopts an incremental learning mechanism and a hybrid active learning mechanism. When new monitoring data is input, the state estimation model is updated and fine-tuned online using a local memory storage mechanism to adapt to changes in data distribution. The hybrid active learning mechanism actively selects the most representative and informative sample points for experts to annotate based on the uncertainty of the model output, and adds the training set to iteratively update the model to improve the generalization ability of the model. The health assessment unit includes, for main helium blowers of different models and working conditions, using expert knowledge and data analysis to determine an adaptive health index system, including vibration index, temperature index and efficiency index, and assigning different weight coefficients, and using weighted average and geometric average methods to calculate the health index. When the health index is lower than a preset threshold, the health assessment unit triggers an early warning, prompting the operation and maintenance personnel to take maintenance measures, and displaying the real-time working condition and health status of the main helium blower through a human-computer interaction interface.

4. A method for using a main helium blower health status assessment system based on state estimation as claimed in any one of claims 1 to 3, characterized in that: The signal processing and machine learning algorithms include: extracting the features of the vibration data using a wavelet packet decomposition algorithm, an empirical mode decomposition algorithm, and a singular value decomposition algorithm; optimizing the selection of wavelet packet basis functions and the number of decomposition layers; Among them, WP(x,j,k) is the wavelet packet coefficient, x is the vibration signal, j is the decomposition scale, k is the frequency band index, g j,k,n is the adaptively selected wavelet packet basis function, φ j,n (x) is the scaling function; Add noise-assisted technology to decompose non-stationary signals: Where x(t) is the original vibration signal, c i (t) is the eigenmode function, r N (t) is the residual signal, ò(t) is the introduced white noise, and N is the number of decomposition layers determined adaptively; Add sliding window technology to capture local features: Where X is the sliding window trajectory matrix of the vibration signal, U and V are the left and right singular vector matrices, Σ is the singular value diagonal matrix, σ i is a singular value, r is the rank; Finally, multi-scale feature fusion is performed, and a feature weight adaptive adjustment method based on information entropy is proposed: Among them, F is the fused feature vector, f i is the feature subset extracted by various methods, w i is the corresponding weight, H(f i ) is the information entropy of the feature subset, and M is the number of feature subsets.

5. A main helium blower health status assessment system based on state estimation as claimed in claim 4, characterized in that: The signal processing and machine learning algorithm also includes extracting features of the temperature data and improving the state space model through an improved adaptive Kalman filter algorithm: Among them, x k is the temperature state vector at time k, y k is the observed value, f(·) and h(·) are the nonlinear state transfer function and observation function respectively, w k and v k are process noise and observation noise respectively; Estimate adaptive noise: Among them, Q k and R k are the process noise covariance and observation noise covariance of the adaptive estimation, λ is the forgetting factor, K k is the Kalman gain, P k is the state estimation error covariance, F k and H k are the Jacobian matrices of the state transfer matrix and the observation matrix respectively, N is the sliding window size, is the state estimation vector at time i; Improve the Kalman filter equation for prediction: Update based on forecast results: in, and are the prior and posterior state estimates, respectively, and P k are the prior and a posteriori estimated error covariances, respectively; Perform multi-model fusion on the updated results: in, is the final fused state estimate, is the state estimate of the ith model, μ i is the corresponding weight, e i is the innovation sequence, S i is the innovation covariance, M is the number of models; Extract the characteristics of current data, use Hilbert transform to extract instantaneous amplitude and phase information, calculate the analytical signal of the current signal, estimate the instantaneous amplitude A(t) and phase φ(t), and reflect the energy change and frequency characteristics of the current signal: Among them, i(t) is the original current signal, for its Hilbert transform; After all features are extracted, the multi-source heterogeneous features of vibration, temperature, and current are aggregated to form a high-dimensional feature vector as the input of the state estimation model. Principal component analysis and independent component analysis are used to eliminate the redundancy and correlation between features and extract the essential feature subset.

6. A method for evaluating the health status of a main helium blower based on state estimation according to claim 5, characterized in that: The health status estimation model includes constructing a health status estimation model by combining physical mechanism modeling and machine learning; The physical mechanism modeling includes: the state estimation unit adopts an innovative hybrid modeling paradigm, integrates physical mechanism modeling and data-driven modeling, increases multi-physical field coupling effects, and establishes a more comprehensive main helium fan dynamics model: Where q is the generalized displacement vector, M, C, G, K are the mass matrix, damping matrix, gyro matrix and stiffness matrix respectively, Ω is the rotation speed, F a 、F m 、F t They are aerodynamic force, electromagnetic force and thermal stress respectively; Adopting an improved hybrid deep learning model: i t =σ(W xi x t +W hi h t-1 +b i ) f t =σ(W xf x t +W hf h t-1 +b f ) o t =σ(W xo x t +W ho h t-1 +b o ) c t =f t ⊙c t-1 +i t ⊙tanh(W xc x t +W hc h t-1 +b c h t =o t ⊙tanh(c t ) Among them, H (l) is the l-th layer feature matrix of GCN, To enhance the adjacency matrix, is the degree matrix, W (l) is the weight matrix, i t 、f t , o t 、c t 、h t are the input gate, forget gate, output gate, unit state and hidden state of LSTM respectively, W and b are the corresponding weights and biases; Introducing physical constraints: THE total =L data +λL physics Among them, L total is the total loss function, L data is the data-driven loss, L physics is the physical constraint loss, λ is the trade-off factor; A multi-model fusion strategy based on dynamic weights is proposed: in, is the fused state estimate, is the estimated result of the ith model, w i (t) is the dynamic weight, e i (t) is the estimation error, α is the sensitivity parameter, and N is the number of models; Through domain adaptation strategy, the model can be quickly migrated between different working conditions: L transfer =L task +βMMD(S,T) Among them, L transfer is the transfer learning loss function, L task is the target task loss, MMD(S,T) is the maximum mean difference between the source domain and the target domain, and β is the balance parameter; Aiming at the complex dynamic characteristics of the main helium blower, a multi-physics coupling model is proposed, combining rotor dynamics, fluid dynamics and electromagnetic thermal coupling effects: Where q is the generalized displacement vector, including the displacement of the rotor, bearing and stator; M is the mass matrix, C is the damping matrix, G is the gyro matrix, K is the stiffness matrix, Ω is the speed, K b (δ) is the nonlinear stiffness term considering the bearing clearance δ, K m (i) is the electromagnetic stiffness caused by current i, is the aerodynamic force, taking into account the influence of speed and rotation speed; F m (i,δ) is the electromagnetic force, considering the influence of current and gap; F t (T) is thermal stress, considering the influence of temperature distribution; To capture complex nonlinear dynamic behaviors for the newly introduced nonlinear force terms; The machine learning includes establishing a data-driven model using a machine learning algorithm for the multi-source monitoring features extracted by the feature extraction unit. The forward calculation formula of DBN is: h (1) =f(W (1) x+b (1) )h (l) h (1) =f(W (l) h (l-1) +b (l) ),l=2,…,L-1 Among them, x is the input feature, h (l) is the hidden variable of layer I, f(·) is the activation function, W (l) and b (l) are weight and bias parameters, To predict the state quantity, during the training process, a gradient-based optimization algorithm is used to minimize the reconstruction error and prediction error.

7. A method for evaluating the health status of a main helium blower based on state estimation according to claim 6, characterized in that: The incremental learning mechanism and hybrid active learning mechanism include: the adaptive incremental learning strategy improves the traditional DBN model and introduces an adaptive hierarchical structure and a dynamic weight adjustment mechanism: η(t)=η0·exp(-λt)+η min in, is the output of the i-th hidden unit in the first layer, is the output of the jth unit in the previous layer, is the weight, is the bias, f(·) is the activation function, η(t) is the adaptive learning rate, η0 is the initial learning rate, λ is the decay coefficient, η min is the minimum learning rate; Local memory storage via experience replay strategy based on importance sampling: Among them, p(i) is the probability that sample i is selected, p i is the priority of the sample, α controls the degree of sampling deviation, δ i is the timing difference error, is a small positive number; The hybrid active learning mechanism combines uncertainty sampling and diversity sampling to propose a new sample selection criterion: D(x,L)=min x′∈L d(x,x′) Where S is the selected sample set, U is the unlabeled sample pool, L is the labeled sample set, H(y|x,θ) is the model’s predicted entropy for sample x, reflecting uncertainty, D(x,L) is the minimum distance between sample x and the labeled sample set, reflecting diversity, λ is the balance factor, and d(·,·) is the distance metric function; Introducing fuzzy cognitive graph to integrate expert knowledge: Among them, A i (t) is the activation value of concept node i at time t, w ji is the causal weight from concept j to i, f(·) is the sigmoid function, and λ is the steepness parameter; Propose an adaptive update mechanism based on performance degradation detection: τ(t)=τ0·(1-e -γt ) Among them, Δ is the average prediction error, y i and are the true value and predicted value respectively, N is the number of samples, τ(t) is the dynamic threshold, τ0 is the initial threshold, γ is the adjustment coefficient, and the model update is triggered when Δ exceeds τ(t).

8. A method for evaluating the health status of a main helium blower based on state estimation according to claim 7, characterized in that: The calculation of the health index of the main helium blower includes: constructing an adaptive health index system, considering the importance and influence of vibration, temperature, and efficiency state quantities, using the basic probability assignment method in the evidence theory, converting the state quantity output by the state estimation unit into the confidence distribution of health, and using the Dempster combination rule to fuse the BPA of different health indicators to obtain the confidence distribution of the overall health status of the main helium blower. Suppose the BPA of the i-th indicator is m i , the focal element is A i1 ,A i2 ,...,A ik , then the Dempster combination rule is: in, represents the Dempster combination, A is any focal element after fusion, A i is the focal element of each indicator, represents the empty set; Using the fuzzy logic reasoning method, the fused confidence distribution is converted into a language value description, and the weighted sum of the membership centers of all fuzzy levels is taken to obtain a health index between 0 and 100. When the health index is lower than 60, the system automatically triggers an early warning, prompting the operation and maintenance personnel to take corresponding measures.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a main helium blower health status assessment method based on state estimation as described in any one of claims 4 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a main helium blower health status assessment method based on state estimation according to any one of claims 4 to 8 are implemented.

Images