Gas turbine unit fault diagnosis and classification method and system based on LSTM

Abstract

The invention discloses a gas turbine unit fault diagnosis and classification method and system based on LSTM, relates to the technical field of gas turbines, and mainly aims at solving the problem that the accuracy of fault judgment of an existing gas turbine is low. The method mainly comprises the steps that equipment operation parameters of all gas turbines in a gas turbine unit are obtained, and equipment fault tolerance coefficients corresponding to all the gas turbines are called based on the unit network relation of the gas turbine unit; fault diagnosis is carried out on the equipment operation parameters and the equipment fault tolerance coefficient based on a fault diagnosis model after model training is completed, a fault diagnosis result of the gas turbine unit is obtained, and the fault diagnosis model is constructed based on an LSTM network; and if the fault diagnosis result is the unit fault, determining the unit fault level of the gas turbine unit based on the equipment operation parameters.

Description

Technical Field

[0001] The present invention relates to the technical field of gas turbines, and in particular to a gas turbine unit fault diagnosis classification method and system based on LSTM. Background Art

[0002] With the rapid development of artificial intelligence, enterprises in various industries are gradually applying artificial intelligence algorithms to data processing, prediction and other fault diagnosis and analysis. In particular, enterprises that carry out production operations based on gas turbine units usually need to diagnose gas turbine unit faults to ensure the normal operation of gas turbine units.

[0003] At present, when diagnosing the faults of existing gas turbine units, thresholds are usually configured for the operating parameters of the gas turbine units. When the thresholds are exceeded, a fault is determined to have occurred and the corresponding fault classification is determined. However, since the comparison of the thresholds of the operating parameters requires the support of a large amount of debugging data, a large amount of manpower and material resources are wasted. If the collected operating parameters are distorted, it will cause misjudgment of the diagnosis results and reduce the accuracy of the fault diagnosis. Therefore, a gas turbine unit fault diagnosis classification method is urgently needed to solve the above problems. Summary of the invention

[0004] In view of this, the present invention provides a gas turbine unit fault diagnosis classification method and system based on LSTM, the main purpose of which is to solve the problem of low accuracy of existing gas turbine fault judgment.

[0005] According to one aspect of the present invention, a gas turbine unit fault diagnosis classification method based on LSTM is provided, comprising: Acquire equipment operating parameters of each gas turbine in the gas turbine unit, and retrieve equipment fault tolerance coefficients corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; Performing fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit, wherein the fault diagnosis model is constructed based on an LSTM network; If the fault diagnosis result is a unit fault, the unit fault level of the gas turbine unit is determined based on the equipment operating parameters.

[0006] Furthermore, the equipment operating parameters include impeller speed, inlet and outlet temperatures, inlet and outlet pressures, equipment operating time, and network level. The fault diagnosis model based on the completed model training performs fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient. Before obtaining the fault diagnosis result of the gas turbine unit, the method further includes: A fault diagnosis model is constructed based on the LSTM network, and a historical equipment operation sample set is obtained, wherein the fault diagnosis model includes an input gate, a forget gate, a state gate, and an output gate, and the historical equipment operation sample set includes training sample data marked with a fault label, and the weight matrices of the state gate and the forget gate are determined based on the number of layers of the network layer and the number of device nodes; The fault diagnosis model is trained using the historical device operation sample set to obtain the fault diagnosis model that has completed model training.

[0007] Furthermore, the method further comprises: Determining the number of rows and columns of the weight matrix according to the number of devices in the gas turbine group; According to the proportional addition operation relationship and the number of layers and the number of device nodes of the network layer, the element value of each gas turbine is calculated in turn, and a weight matrix is ​​constructed based on the element value.

[0008] Furthermore, before retrieving the equipment fault tolerance coefficient corresponding to each gas turbine based on the unit network relationship of the gas turbine unit, the method further includes: Retrieving the relationship weight corresponding to the network relationship of the unit, and obtaining the single-machine fault threshold of each gas turbine, wherein the single-machine fault threshold includes an impeller speed threshold, an inlet and outlet temperature threshold, an inlet and outlet pressure threshold, and an equipment operation time threshold; A proportional operation is performed based on the relationship weight and the single machine failure threshold to obtain the device failure tolerance coefficient.

[0009] Further, the determining the unit fault level of the gas turbine unit based on the equipment operating parameters includes: Comparing the differences between the equipment operating parameters and the corresponding single-machine failure thresholds one by one; If the obtained difference is greater than the number of thresholds corresponding to the preset fault level classification threshold, the unit fault level corresponding to the fault level classification threshold is determined; If the obtained difference is less than or equal to the number of thresholds corresponding to the preset fault classification threshold, the next unit fault level corresponding to the fault level classification threshold is determined.

[0010] Furthermore, the method further comprises: If the fault diagnosis result is a device fault, a target faulty device is identified, and a device fault level of the target faulty device is determined.

[0011] Further, the identifying the target faulty device and determining the device fault level of the target faulty device includes: Compare the single machine failure threshold with the equipment operating parameters one by one; If the comparison result is greater than the preset unit threshold corresponding to the unit network relationship, determining the target faulty device corresponding to the device operating parameter; The comparison result is divided according to the difference ratio, and the device fault level is determined based on the division result.

[0012] According to another aspect of the present invention, a gas turbine unit fault diagnosis classification system based on LSTM is provided, comprising: An acquisition module, used for acquiring equipment operating parameters of each gas turbine in the gas turbine unit, and retrieving equipment fault tolerance coefficients corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; A diagnosis module, used to perform fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit, wherein the fault diagnosis model is constructed based on an LSTM network; A determination module is used to determine the unit fault level of the gas turbine unit based on the equipment operating parameters if the fault diagnosis result is a unit fault.

[0013] Furthermore, the system further comprises: A construction module is used to construct a fault diagnosis model based on an LSTM network and obtain a historical equipment operation sample set, wherein the fault diagnosis model includes an input gate, a forget gate, a state gate, and an output gate, and the historical equipment operation sample set contains training sample data marked with fault labels, and the weight matrices of the state gate and the forget gate are determined based on the number of layers of the network layer and the number of device nodes; The training module is used to perform model training on the fault diagnosis model through the historical device operation sample set to obtain the fault diagnosis model that has completed model training.

[0014] Furthermore, the system further comprises: The determination module is further used to determine the number of rows and columns of the weight matrix according to the number of devices in the gas turbine group; The calculation module is used to calculate the element value of each gas turbine in turn according to the proportional addition operation relationship and the number of layers and device nodes of the network layer, and construct a weight matrix based on the element value.

[0015] Furthermore, the system further comprises: A calling module, used to call the relationship weight corresponding to the network relationship of the unit, and obtain the single-machine fault threshold of each gas turbine, wherein the single-machine fault threshold includes an impeller speed threshold, an inlet and outlet temperature threshold, an inlet and outlet pressure threshold, and an equipment operation time threshold; The calculation module is used to perform a proportional calculation based on the relationship weight and the single machine failure threshold to obtain the device failure tolerance coefficient.

[0016] Furthermore, the determination module includes: A first comparison unit, used for comparing the differences between the equipment operating parameters and the corresponding single machine failure thresholds one by one; A first determining unit, configured to determine, if the obtained difference is greater than the number of thresholds corresponding to a preset fault level classification threshold, the unit fault level corresponding to the fault level classification threshold; The second determining unit is used to determine the next unit fault level corresponding to the fault level classification threshold if the obtained difference is less than or equal to the number of thresholds corresponding to the preset fault classification threshold.

[0017] Furthermore, the system further comprises: The determination module is further configured to identify a target faulty device and determine a device fault level of the target faulty device if the fault diagnosis result is a device fault.

[0018] Furthermore, the determining module further includes: A second comparison unit, configured to compare the single machine failure threshold with the equipment operation parameters one by one; A third determination unit, configured to determine a target faulty device corresponding to the device operating parameter if the comparison result is greater than a preset unit threshold corresponding to the unit network relationship; The division unit is used to divide the comparison result according to the difference ratio, and determine the equipment fault level based on the division result.

[0019] According to another aspect of the present invention, a storage medium is provided, wherein at least one executable instruction is stored in the storage medium, and the executable instruction enables a processor to perform operations corresponding to the above-mentioned LSTM-based gas turbine unit fault diagnosis and classification method.

[0020] According to another aspect of the present invention, there is provided a terminal, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned LSTM-based gas turbine unit fault diagnosis and classification method.

[0021] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages: The present invention provides a gas turbine unit fault diagnosis classification method and system based on LSTM. Compared with the prior art, the embodiment of the present invention obtains the equipment operation parameters of each gas turbine in the gas turbine unit, and retrieves the equipment fault tolerance coefficient corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; performs fault diagnosis on the equipment operation parameters and the equipment fault tolerance coefficient based on the fault diagnosis model that has completed model training, and obtains the fault diagnosis result of the gas turbine unit, and the fault diagnosis model is constructed based on the LSTM network; if the fault diagnosis result is a unit fault, the unit fault level of the gas turbine unit is determined based on the equipment operation parameters. By configuring the unit network relationship and the equipment fault tolerance coefficient of each gas turbine, the gas turbines in the gas turbine unit are associated, and the single machine fault and unit fault are diagnosed based on the LSTM fault diagnosis model, so that the fault diagnosis result can realize the comprehensive diagnosis of the gas turbine unit from the whole to the single machine, and further subdivide the unit fault level for the unit fault, thereby greatly improving the accuracy and precision of fault diagnosis.

[0022] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 A flow chart of a gas turbine unit fault diagnosis classification method based on LSTM provided by an embodiment of the present invention is shown; Figure 2 Another flow chart of a gas turbine unit fault diagnosis classification method based on LSTM provided by an embodiment of the present invention is shown; Figure 3 A block diagram of another LSTM-based gas turbine unit fault diagnosis and classification system provided by an embodiment of the present invention is shown; Figure 4 A schematic diagram of the structure of a terminal provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0024] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0025] The embodiment of the present invention provides a gas turbine unit fault diagnosis classification method based on LSTM, such as Figure 1 As shown, the method includes: 101. Obtain equipment operating parameters of each gas turbine in a gas turbine unit, and retrieve equipment fault tolerance coefficients corresponding to each gas turbine based on a unit network relationship of the gas turbine unit.

[0026] In an embodiment of the present invention, a gas turbine unit includes a plurality of gas turbines. For example, a three-to-one gas turbine unit includes three gas turbines, three waste heat boilers and one steam turbine. The operating connection relationship between the gas turbines may be a series relationship or a parallel relationship. The series relationship is that the output (gas, energy, etc.) of the upstream gas turbine is used as the input of the downstream gas turbine or provides energy for the operation of the downstream gas turbine; the parallel relationship is that the multiple gas turbines operate independently and input the outputs into the same downstream device (such as a steam turbine). Based on the connection relationship between the gas turbines, a unit network relationship of the gas turbine unit can be constructed. For example, gas turbine A is simultaneously operated in series with gas turbine B and gas turbine C, and gas turbine B is operated in parallel with gas turbine C. Then the relationship between gas turbines A, B, and C is the unit network relationship of the corresponding gas turbine unit, and the unit network relationship is stored in the storage space.

[0027] Since the gas turbine unit is different from a single gas turbine, there is a mutual influence effect between the gas turbines inside the unit, and the allowed equipment operation data limit range of each gas turbine will be smaller than that of a single unit. If the equipment of each gas turbine is still judged according to the single-unit fault threshold, the gas unit may have a fault, but the fault is not identified. Therefore, since the positions of each gas turbine in the unit network of the gas turbine unit are different, when the operation is abnormal, the degree of influence on the operation of the gas turbine unit is also different. Therefore, different equipment fault tolerance coefficients are configured for gas turbine equipment at different positions in the unit network relationship. That is, in order to avoid the equipment operating parameters reaching the corresponding threshold before being identified, which has a serious impact on the gas turbine unit as a whole, the equipment fault tolerance coefficient is used to tighten the fault judgment standard, that is, the equipment fault tolerance coefficient is used to adjust the single-unit fault threshold. The equipment fault tolerance coefficient is proportional to the degree of influence of the equipment on the entire gas turbine unit when the equipment fails, that is, the greater the degree of influence, the lower the tolerance for equipment failure, and the larger the corresponding equipment fault tolerance coefficient value.

[0028] It should be noted that by constructing the unit network relationship of the gas turbine unit and determining the equipment fault tolerance coefficient of different gas turbines according to the unit network relationship, the gas turbines at different positions in the gas turbine unit have corresponding fault judgment limits, thereby achieving targeted fault judgment for gas turbines with different impact levels.

[0029] 102. Perform fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit.

[0030] In an embodiment of the present invention, the fault diagnosis model is constructed based on an LSTM network. Among them, the long short-term memory network (LSTM) is a time recurrent neural network, which is specially designed to solve the long-term dependency problem of recurrent neural networks. Since the equipment operating parameters are collected and there is data that changes continuously in time, and LSTM can effectively save and update historical information, it is more suitable for scenarios where faults are diagnosed and classified based on equipment operating parameters. The fault diagnosis model that has completed model training can classify the operating status of the gas unit according to the equipment operating parameters and the equipment fault tolerance coefficient corresponding to the equipment operating parameters. The classification result (fault diagnosis result) can be one of normal operation (no fault), unit failure and equipment failure.

[0031] It should be noted that by taking the equipment operating parameters and the corresponding equipment fault tolerance coefficient as the input of the model, the association and influence relationship between different gas turbines in the gas turbine unit network relationship can be learned. At the same time, the LSTM network can learn the long-term dependency between equipment operating parameters, thereby improving the accuracy of gas turbine unit fault identification.

[0032] 103. If the fault diagnosis result is a unit fault, determining a unit fault level of the gas turbine unit based on the equipment operating parameters.

[0033] In the embodiment of the present invention, the fault diagnosis results include normal operation, unit failure and equipment failure. Among them, normal operation means that the equipment operation status of the gas turbine unit and each gas turbine unit is normal; unit failure means that the gas turbine unit has an overall failure; equipment failure means that one or more gas turbine units have a failure. When it is determined that the fault diagnosis result is that the gas turbine unit has a failure, in order to output an accurate fault judgment result, the fault level of the gas turbine unit is further identified. For example, the unit fault level may include a first-level fault for characterizing the need for overall shutdown, a second-level fault for characterizing the need for partial shutdown, and a third-level fault for characterizing that adjustment and maintenance can be performed during operation. Of course, other unit fault level division methods can also be customized according to actual application requirements, and the embodiment of the present invention does not make specific limitations. By further identifying the unit fault level, it is possible to achieve a fine division of the fault level on the basis of realizing the identification of the overall fault of the gas turbine unit and the fault of the gas turbine unit, thereby improving the accuracy of gas turbine unit fault judgment.

[0034] In an embodiment of the present invention, for further explanation and limitation, as Figure 2 As shown, the step performs fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, and before obtaining the fault diagnosis result of the gas turbine unit, the method further includes: 201. Build a fault diagnosis model based on the LSTM network and obtain a historical equipment operation sample set.

[0035] 202. Perform model training on a fault diagnosis model using the historical device operation sample set to obtain the fault diagnosis model that has completed model training.

[0036] In the embodiment of the present invention, the equipment operation parameters include impeller speed, inlet and outlet temperature, inlet and outlet pressure, equipment operation time, and network level. Among them, the network level is the level of the corresponding gas turbine in the unit network relationship. The fault diagnosis model includes an input gate, a forget gate, a state gate, and an output gate. Among them, the state gate (memory cell) is the core of the fault diagnosis model, which is responsible for saving important information and passing important information to the subsequent network layer. Its function is to maintain a persistent state. It will be continuously updated in continuous input. Its characteristic is that it can save historical information and apply it to the current prediction. The state gate can be regarded as a simple number. It will be updated at each time step, and then the input gate, forget gate and output gate will control how to update and output this value in each time step. Input gate: determines whether the current input information is written into the memory cell, that is, it can control the impact of the input information on the memory cell. Forget gate: determines whether the information in the memory cell is forgotten, that is, it can control whether the information stored in the memory cell will disappear. Output gate: determines whether the information in the memory cell is output, that is, it can control whether the information stored in the memory cell will affect the subsequent network layer. The above four gates calculate the dot product of the weight matrix and the input signal, and calculate the output value of each gate through the activation function (such as the sigmoid function), and then multiply it by the value of the memory cell for the final calculation to obtain the classification result of the fault diagnosis, that is, the probability values ​​of the three categories of normal operation, unit failure and equipment failure are obtained, and the classification with the largest probability value is taken as the fault diagnosis result. The historical equipment operation sample set contains training sample data marked with fault labels, wherein the training sample data includes sample data marked with normal operation sample labels, unit failure sample labels, and equipment failure sample labels.

[0037] It should be noted that the weight matrices of the state gate and the forget gate of the fault diagnosis model are determined based on the number of layers and the number of device nodes of the network layer. Among them, the forget gate and the state gate can use the same set of weight matrices, that is, a weight matrix is ​​calculated based on the number of layers and the number of device nodes for the forget gate and the state gate to share. It is also possible to use different calculation methods according to the number of layers and the number of device nodes to obtain two different weight matrices for the forget gate and the state gate respectively. By constructing the weight matrix by the number of layers and the number of device nodes of the network layer, the forget gate and the state gate can fully learn the degree of influence of different gas turbines on the gas turbine unit, thereby improving the accuracy of the model output, that is, improving the accuracy of the fault diagnosis results.

[0038] In an embodiment of the present invention, for further explanation and limitation, the method further includes: Determining the number of rows and columns of the weight matrix according to the number of devices in the gas turbine group; According to the proportional addition operation relationship and the number of layers and the number of device nodes of the network layer, the element value of each gas turbine is calculated in turn, and a weight matrix is ​​constructed based on the element value.

[0039] In an embodiment of the present invention, the weight matrix can be constructed with the number of layers of the network layer as the number of rows of the weight matrix, and the number of device nodes (total number of gas turbines) as the number of columns of the weight matrix; or the number of layers of the network layer as the number of columns of the weight matrix, and the total number of gas turbines as the number of rows of the weight matrix. The embodiment of the present invention does not make specific restrictions. The element value of each gas turbine is calculated, and the element value is configured in the position corresponding to the network relationship of the gas turbine unit in the weight matrix to obtain the weight matrix. For example, the gas turbine unit includes 3 gas turbines, which are divided into two network layers. The first network layer contains one gas turbine, and the second network layer contains two gas turbines. The number of layers is the number of rows, and the number of device nodes is the number of columns. Then the weight matrix is ​​a 2×3 matrix. The first row of the matrix can contain the element values ​​of the gas turbines of the first network layer, and the second row of the matrix can contain the element values ​​of the gas turbines of the second network layer. Other vacancies can be filled with 0. The calculation of element values ​​can use the proportional method or the sum method. For example, if the number of levels is 2 and the number of equipment nodes is 3, according to the proportional method, the element value of each gas turbine can be 1 / 3; according to the sum method, the element value of the first-level gas turbine can be 0.4, and the element values ​​of the second-level gas turbine can be 0.3 and 0.3. The proportional sum method can also be used. For example, if the number of levels is 2, each level accounts for 1 / 2, and the sum of the weights of the gas turbines in each level is 1 / 2. The element value of the gas turbine in the level containing one gas turbine is 0.5, and the sum of the element values ​​of each gas turbine in the level containing two gas turbines is 0.5. By calculating the element value of each gas turbine and constructing a weight matrix based on the element value, the fault diagnosis model can fully learn the importance of each gas turbine in the gas turbine unit and the relevance of the equipment operation parameters to the fault diagnosis classification results, thereby improving the learning ability of the fault diagnosis model and ensuring the accuracy of fault diagnosis.

[0040] In one embodiment of the present invention, for further explanation and limitation, before the step of retrieving the equipment fault tolerance coefficient corresponding to each of the gas turbines based on the unit network relationship of the gas turbine unit, the method further includes: Retrieving the relationship weight corresponding to the network relationship of the unit, and obtaining the single-machine failure threshold of each of the gas turbines; A proportional operation is performed based on the relationship weight and the single machine failure threshold to obtain the device failure tolerance coefficient.

[0041] In the embodiment of the present invention, the equipment fault tolerance coefficient is calculated based on the relationship weight corresponding to the unit network relationship and the single-machine fault threshold. Among them, the single-machine fault threshold includes the impeller speed threshold, the inlet and outlet temperature threshold, the inlet and outlet pressure threshold, and the equipment operation time threshold. The single-machine fault threshold represents the limit value configured according to the equipment performance limit or historical fault experience under the condition of single operation of the gas turbine, and may include a maximum value and / or a minimum value. For example, the equipment operation time threshold includes the equipment operation time maximum value; the inlet pressure threshold includes the inlet pressure maximum value and the inlet pressure minimum value; the outlet temperature threshold includes the outlet temperature minimum value. The relationship weight is used to characterize the influence or contribution of each gas turbine on the overall operation of the gas turbine unit. It can be determined based on the weight matrix extracted by the fault diagnosis model, or it can be determined based on the operation importance of the gas turbine in the gas turbine unit. The size of the relationship weight can be configured to be proportional to the importance of the gas turbine. The equipment fault tolerance coefficient can be determined according to the product of the single machine fault threshold and the relationship weight. For example, if the relationship weight is 1 / 3 and the pressure fault threshold is 300P, the equipment fault tolerance coefficient is 100P multiplied by the adjustment coefficient. The adjustment coefficient is used to scale all the relationship weights in equal proportion. For example, the adjustment coefficient is 10%. The unit fault threshold of the gas turbine in the unit can also be calculated according to the relationship weight and the importance of the gas turbine, and then the difference between the unit fault threshold and the single machine fault threshold is used as the equipment fault tolerance coefficient. The present invention does not make specific restrictions. In the method for calculating the unit fault threshold, for the single machine fault minimum value, the single machine fault minimum value is set to A, and the relationship weight is W. Then the calculation formula of the unit fault minimum value a is: a=A / (1-W×adjustment coefficient) (1). Since the relationship weight is less than 1, the minimum value and the relationship weight are divided to obtain a limit value greater than the single machine fault minimum value, which realizes the reduction of the single machine fault minimum value. The larger the relationship weight, the larger the equipment fault tolerance coefficient, and the larger the reduction of the original single machine fault minimum value. For the maximum value B in the single-machine fault threshold, the calculation formula for the unit fault maximum value b is: b=B×(1-W×adjustment coefficient) (2); the product of the maximum value and the relationship weight can obtain a limit value that is smaller than the fault maximum value, thereby achieving the limitation of the single-machine fault maximum value. The larger the relationship weight, the larger the equipment fault tolerance coefficient, and the greater the limitation of the original single-machine fault minimum value.

[0042] In one embodiment of the present invention, for further explanation and limitation, the step of determining the unit fault level of the gas turbine unit based on the equipment operating parameters includes: Comparing the differences between the equipment operating parameters and the corresponding single-machine failure thresholds one by one; If the obtained difference is greater than the number of thresholds corresponding to the preset fault level classification threshold, the unit fault level corresponding to the preset fault level classification threshold is determined; If the obtained difference is less than or equal to the number of thresholds corresponding to the preset fault classification threshold, the next unit fault level corresponding to the preset fault level classification threshold is determined.

[0043] In the embodiment of the present invention, the difference between each equipment operating parameter and the corresponding single-machine fault threshold is calculated one by one. And the size of each difference is compared with the preset fault level classification threshold one by one, and the number of differences greater than the preset fault level classification threshold is counted, and the unit fault level is determined based on this number. For example, the threshold number 1 corresponds to a level 3 unit failure, the threshold number 3 corresponds to a level 2 unit failure, and the threshold number 5 corresponds to a level 1 unit failure; if the inlet temperature is 40°C and the corresponding inlet temperature threshold is 60°C, the difference is 20°C, and the preset fault level classification threshold for the corresponding inlet temperature is 15°C; the outlet temperature is 230°C and the corresponding outlet temperature threshold is 260°C, then the difference is 30°C, and the preset fault level classification threshold for the corresponding outlet temperature is 25°C; the differences of other equipment operating parameters are all less than the corresponding preset fault level classification threshold, then the difference is greater than The number of preset fault level classification thresholds is 2, which is greater than the threshold number 1 corresponding to the third-level unit fault, and less than the threshold number 3 corresponding to the second-level unit fault. The unit fault level is determined to be a smaller level, that is, a third-level unit fault. Similarly, if the difference is greater than the number of preset fault level classification thresholds of 5, which is equal to the threshold number corresponding to the first-level unit fault, the unit fault level is determined to be a second-level unit fault; if the difference is greater than the number of preset fault level classification thresholds of 6, which is greater than the threshold number corresponding to the first-level unit fault, the unit fault level is determined to be a first-level unit fault.

[0044] By calculating the difference between each equipment operating parameter and the corresponding single-machine equipment threshold, the abnormality degree of the equipment operating parameters can be further accurately obtained on the basis of determining that the equipment operating parameters are abnormal. Further, based on the number of differences greater than the preset fault level classification threshold, the number of equipment operating parameters with more serious abnormalities can be determined, thereby achieving an assessment of the overall abnormality degree of the unit equipment, thereby achieving accurate classification of the unit fault level.

[0045] In an embodiment of the present invention, for further explanation and limitation, the method further includes: If the fault diagnosis result is a device fault, a target faulty device is identified, and a device fault level of the target faulty device is determined.

[0046] In the embodiment of the present invention, when the fault diagnosis result of the fault diagnosis model is an equipment fault, it indicates that there is a fault in an individual gas turbine in the current gas turbine group. However, which gas turbine is faulty needs to be further identified based on the equipment operating parameters of each gas turbine, and each equipment operating parameter can be determined based on the single-machine fault threshold. For example, the outlet temperature fault threshold in the single-machine fault threshold is 250°C, and the outlet temperature of the corresponding gas turbine is verified according to 250°C. If the outlet temperature of only one gas turbine among the three gas turbines exceeds 250°C, the gas turbine is determined to be the target fault device. It can be further determined based on the number of equipment operating parameters of the target fault device that exceed the corresponding threshold. For example, the target fault device includes 6 equipment operating parameters, namely, impeller speed, inlet and outlet temperature, inlet and outlet pressure, and equipment operation time. If 5 of them exceed the corresponding threshold, the equipment fault level can be level 1, if 3 of them exceed the corresponding threshold, the equipment fault level can be level 2, and if 1 of them exceeds the corresponding threshold, the equipment fault level can be level 3. The equipment fault level can also be determined by the difference between the abnormal equipment operating parameters and the corresponding threshold. For example, the minimum outlet temperature is 200°C. If the outlet temperature of the target faulty equipment is 150°C, the equipment fault level can be level one; if the outlet temperature of the target faulty equipment is 180°C, the equipment fault level can be level two; if the outlet temperature of the target faulty equipment is 195°C, the equipment fault level can be level three.

[0047] In one embodiment of the present invention, for further explanation and limitation, the step of identifying a target faulty device and determining a device fault level of the target faulty device includes: Compare the single machine failure threshold with the equipment operating parameters one by one; If the comparison result is greater than the preset unit threshold corresponding to the unit network relationship, determining the target faulty device corresponding to the device operating parameter; The comparison result is divided according to the difference ratio, and the device fault level is determined based on the division result.

[0048] In the embodiment of the present invention, the method of comparing the single-machine fault threshold with the equipment operating parameters one by one can be a difference comparison or a ratio comparison, that is, calculating the absolute value of the difference between the single-machine fault threshold and the corresponding equipment operating parameter, or calculating the ratio of the single-machine fault threshold and the corresponding equipment operating parameter. Correspondingly, the preset unit thresholds corresponding to the two comparison methods are also different. The preset unit threshold corresponding to the difference comparison is the maximum limit of the absolute value of the difference, and the preset unit threshold corresponding to the ratio comparison is the maximum or minimum limit of the ratio. The comparison results of each gas turbine are calculated according to the above method, and the gas turbine whose comparison result is greater than the corresponding preset unit threshold is determined as the target fault device, and the number of target fault devices is not limited to one. If the comparison result is a difference, the comparison result is matched according to different pre-divided difference intervals, and the equipment fault level corresponding to the difference interval matching the comparison result is determined as the equipment fault level of the corresponding gas turbine. If the comparison result is a ratio, the comparison result is matched according to different pre-divided ratio intervals, and the equipment fault level corresponding to the ratio interval matching the comparison result is determined as the equipment fault level of the corresponding gas turbine.

[0049] The present invention provides a gas turbine unit fault diagnosis classification method based on LSTM. Compared with the prior art, the embodiment of the present invention obtains the equipment operation parameters of each gas turbine in the gas turbine unit, and retrieves the equipment fault tolerance coefficient corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; performs fault diagnosis on the equipment operation parameters and the equipment fault tolerance coefficient based on the fault diagnosis model that has completed model training, and obtains the fault diagnosis result of the gas turbine unit, and the fault diagnosis model is constructed based on the LSTM network; if the fault diagnosis result is a unit fault, the unit fault level of the gas turbine unit is determined based on the equipment operation parameters. By configuring the unit network relationship and the equipment fault tolerance coefficient of each gas turbine, the gas turbines in the gas turbine unit are associated, and the single machine fault and unit fault are diagnosed based on the LSTM fault diagnosis model, so that the fault diagnosis result can realize the comprehensive diagnosis of the gas turbine unit from the whole to the single machine, and further subdivide the unit fault level according to the unit fault, thereby greatly improving the accuracy and precision of fault diagnosis.

[0050] Furthermore, as a response to the above Figure 1 The implementation of the method shown in the figure, the embodiment of the present invention provides a gas turbine unit fault diagnosis classification system based on LSTM, such as Figure 3 As shown, the system includes: An acquisition module 31 is used to acquire the equipment operating parameters of each gas turbine in the gas turbine unit, and retrieve the equipment fault tolerance coefficient corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; A diagnosis module 32, configured to perform fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit, wherein the fault diagnosis model is constructed based on an LSTM network; The determination module 33 is used to determine the unit fault level of the gas turbine unit based on the equipment operating parameters if the fault diagnosis result is a unit fault.

[0051] Furthermore, the system further comprises: A construction module is used to construct a fault diagnosis model based on an LSTM network and obtain a historical equipment operation sample set, wherein the fault diagnosis model includes an input gate, a forget gate, a state gate, and an output gate, and the historical equipment operation sample set contains training sample data marked with fault labels, and the weight matrices of the state gate and the forget gate are determined based on the number of layers of the network layer and the number of device nodes; The training module is used to perform model training on the fault diagnosis model through the historical device operation sample set to obtain the fault diagnosis model that has completed model training.

[0052] Furthermore, the system further comprises: The determination module is further used to determine the number of rows and columns of the weight matrix according to the number of devices in the gas turbine group; The calculation module is used to calculate the element value of each gas turbine in turn according to the proportional addition operation relationship and the number of layers and device nodes of the network layer, and construct a weight matrix based on the element value.

[0053] Furthermore, the system further comprises: A calling module, used to call the relationship weight corresponding to the network relationship of the unit, and obtain the single-machine fault threshold of each gas turbine, wherein the single-machine fault threshold includes an impeller speed threshold, an inlet and outlet temperature threshold, an inlet and outlet pressure threshold, and an equipment operation time threshold; The calculation module is used to perform a proportional calculation based on the relationship weight and the single machine failure threshold to obtain the device failure tolerance coefficient.

[0054] Furthermore, the determination module 33 includes: A first comparison unit, used for comparing the differences between the equipment operating parameters and the corresponding single machine failure thresholds one by one; A first determining unit, configured to determine, if the obtained difference is greater than the number of thresholds corresponding to a preset fault level classification threshold, the unit fault level corresponding to the fault level classification threshold; The second determining unit is used to determine the next unit fault level corresponding to the fault level classification threshold if the obtained difference is less than or equal to the number of thresholds corresponding to the preset fault classification threshold.

[0055] Furthermore, the system further comprises: The determination module 33 is further configured to identify a target faulty device and determine a device fault level of the target faulty device if the fault diagnosis result is a device fault.

[0056] Furthermore, the determining module further includes: A second comparison unit, configured to compare the single machine failure threshold with the equipment operation parameters one by one; A third determination unit, configured to determine a target faulty device corresponding to the device operating parameter if the comparison result is greater than a preset unit threshold corresponding to the unit network relationship; The division unit is used to divide the comparison result according to the difference ratio, and determine the equipment fault level based on the division result.

[0057] The present invention provides a gas turbine unit fault diagnosis classification system based on LSTM. Compared with the prior art, the embodiment of the present invention obtains the equipment operation parameters of each gas turbine in the gas turbine unit, and retrieves the equipment fault tolerance coefficient corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; performs fault diagnosis on the equipment operation parameters and the equipment fault tolerance coefficient based on the fault diagnosis model that has completed model training, and obtains the fault diagnosis result of the gas turbine unit, and the fault diagnosis model is constructed based on the LSTM network; if the fault diagnosis result is a unit fault, the unit fault level of the gas turbine unit is determined based on the equipment operation parameters. By configuring the unit network relationship and the equipment fault tolerance coefficient of each gas turbine, the gas turbines in the gas turbine unit are associated, and the single machine fault and unit fault are diagnosed based on the LSTM fault diagnosis model, so that the fault diagnosis result can realize the comprehensive diagnosis of the gas turbine unit from the whole to the single machine, and further subdivide the unit fault level according to the unit fault, thereby greatly improving the accuracy and precision of fault diagnosis.

[0058] According to one embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the LSTM-based gas turbine unit fault diagnosis and classification method in any of the above method embodiments.

[0059] Figure 4 A schematic diagram of the structure of a terminal provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the terminal.

[0060] like Figure 4 As shown, the terminal may include: a processor (processor) 402 , a communication interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

[0061] The processor 402 , the communication interface 404 , and the memory 406 communicate with each other via a communication bus 408 .

[0062] The communication interface 404 is used to communicate with other devices such as clients or other servers.

[0063] The processor 402 is used to execute the program 410, and specifically can execute the relevant steps in the above-mentioned LSTM-based gas turbine unit fault diagnosis and classification method embodiment.

[0064] Specifically, the program 410 may include program codes, which include computer operation instructions.

[0065] The processor 402 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiment of the present invention. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0066] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0067] The program 410 may be specifically configured to enable the processor 402 to perform the following operations: Acquire equipment operating parameters of each gas turbine in the gas turbine unit, and retrieve equipment fault tolerance coefficients corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; Performing fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit, wherein the fault diagnosis model is constructed based on an LSTM network; If the fault diagnosis result is a unit fault, the unit fault level of the gas turbine unit is determined based on the equipment operating parameters.

[0068] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing system, they can be concentrated on a single computing system, or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by a program code executable by a computing system, so that they can be stored in a storage system and executed by the computing system, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0069] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A gas turbine unit fault diagnosis and classification method based on LSTM, characterized in that: include: Acquire equipment operating parameters of each gas turbine in the gas turbine unit, and retrieve equipment fault tolerance coefficients corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; Performing fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit, wherein the fault diagnosis model is constructed based on an LSTM network; If the fault diagnosis result is a unit fault, the unit fault level of the gas turbine unit is determined based on the equipment operating parameters.

2. The method according to claim 1, characterized in that: The equipment operation parameters include impeller speed, inlet and outlet temperatures, inlet and outlet pressures, equipment operation time, and network level. The fault diagnosis model based on the completed model training performs fault diagnosis on the equipment operation parameters and the equipment fault tolerance coefficient. Before obtaining the fault diagnosis result of the gas turbine unit, the method further includes: A fault diagnosis model is constructed based on the LSTM network, and a historical equipment operation sample set is obtained, wherein the fault diagnosis model includes an input gate, a forget gate, a state gate, and an output gate, and the historical equipment operation sample set includes training sample data marked with fault labels, and the weight matrices of the state gate and the forget gate are determined based on the number of layers of the network layer and the number of equipment nodes; The fault diagnosis model is trained using the historical device operation sample set to obtain the fault diagnosis model that has completed model training.

3. The method according to claim 2, characterized in that The method further comprises: Determining the number of rows and columns of the weight matrix according to the number of devices in the gas turbine group; According to the proportional addition operation relationship and the number of layers and the number of device nodes of the network layer, the element value of each gas turbine is calculated in turn, and a weight matrix is ​​constructed based on the element value.

4. The method according to claim 1, characterized in that: Before retrieving the equipment fault tolerance coefficient corresponding to each of the gas turbines based on the unit network relationship of the gas turbine unit, the method further includes: Retrieving the relationship weight corresponding to the network relationship of the unit, and obtaining the single-machine fault threshold of each gas turbine, wherein the single-machine fault threshold includes an impeller speed threshold, an inlet and outlet temperature threshold, an inlet and outlet pressure threshold, and an equipment operation time threshold; A proportional operation is performed based on the relationship weight and the single machine failure threshold to obtain the device failure tolerance coefficient.

5. The method according to claim 4, characterized in that Determining the unit fault level of the gas turbine unit based on the equipment operating parameters includes: Comparing the differences between the equipment operating parameters and the corresponding single-machine failure thresholds one by one; If the obtained difference is greater than the number of thresholds corresponding to the preset fault level classification threshold, the unit fault level corresponding to the fault level classification threshold is determined; If the obtained difference is less than or equal to the number of thresholds corresponding to the preset fault classification threshold, the next unit fault level corresponding to the fault level classification threshold is determined.

6. The method according to claim 1, characterized in that The method further comprises: If the fault diagnosis result is a device fault, a target faulty device is identified, and a device fault level of the target faulty device is determined.

7. The method according to any one of claims 1 to 6, characterized in that: The step of identifying a target faulty device and determining a device fault level of the target faulty device includes: Compare the single machine failure threshold with the equipment operating parameters one by one; If the comparison result is greater than the preset unit threshold corresponding to the unit network relationship, determining the target faulty device corresponding to the device operating parameter; The comparison result is divided according to the difference ratio, and the device fault level is determined based on the division result.

8. A gas turbine unit fault diagnosis and classification system based on LSTM, characterized in that: The system comprises: An acquisition module, used for acquiring equipment operating parameters of each gas turbine in the gas turbine unit, and retrieving equipment fault tolerance coefficients corresponding to each gas turbine based on the unit network relationship of the gas turbine unit; A diagnosis module, used to perform fault diagnosis on the equipment operating parameters and the equipment fault tolerance coefficient based on the fault diagnosis model for which model training has been completed, to obtain a fault diagnosis result of the gas turbine unit, wherein the fault diagnosis model is constructed based on an LSTM network; A determination module is used to determine the unit fault level of the gas turbine unit based on the equipment operating parameters if the fault diagnosis result is a unit fault.

9. A storage medium, wherein at least one executable instruction is stored in the storage medium, and the executable instruction enables a processor to perform operations corresponding to the LSTM-based gas turbine unit fault diagnosis and classification method as described in any one of claims 1 to 7.

10. A terminal, comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the LSTM-based gas turbine unit fault diagnosis and classification method as described in any one of claims 1-7.

Images