Multi-task risk assessment method for offshore wind power grid-connected system

Abstract

The embodiment of the invention discloses a multi-task risk assessment method for an offshore wind power grid-connected system, and relates to the technical field of power grids, and the method comprises the steps: obtaining an operation time sequence data sequence of the offshore wind power grid-connected system, carrying out the preprocessing of the operation time sequence data sequence, and adding various symbols to obtain a to-be-coded sample data sequence; performing digital coding, constructing a sample data set based on a digital sequence, training the multi-task processing model, and outputting the trained multi-task processing model; and processing the to-be-processed operation time sequence data sequence based on the trained multi-task processing model, predicting the future state probability of the offshore wind power grid-connected system according to an output result, and outputting a risk assessment result. According to the method, multiple tasks are integrated into one multi-task processing model, so that the state management of the offshore wind power grid-connected system can be unified and simplified, and the efficiency and the reliability of the system can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of electric power grids, and in particular to a multi-task risk assessment method for offshore wind power grid-connected systems. Background Art

[0002] Offshore wind power grid-connected systems are playing an increasingly important role in the global renewable energy sector, and have been widely used, especially in marine areas rich in wind resources. However, the operating environment of offshore wind power grid-connected systems is susceptible to some factors, including extreme meteorological conditions and the impact of the marine environment, such as extreme meteorological conditions such as high temperature, high humidity, strong salt spray corrosion, thunderstorms and typhoons, as well as marine environmental factors such as ocean currents and waves. These factors not only accelerate equipment aging, but may also trigger complex fault chains, leading to instability of the entire system. Therefore, offshore wind power grid-connected systems and their equipment need to adapt to the continuously changing marine environment. In order to ensure that offshore wind power grid-connected systems can operate stably and safely, higher requirements are placed on related systems, and real-time monitoring of the operating status of offshore wind power grid-connected systems and their equipment is required.

[0003] At present, in the process of monitoring offshore wind power grid-connected systems and their equipment, due to the low accessibility of offshore wind power equipment, the maintenance and fault repair costs are significantly higher than those of onshore wind power systems. Therefore, it is particularly important to develop a method that can accurately predict the potential risks and performance changes of the grid-connected system. This prediction method needs to take into account the uncertainty and variability of the marine environment, be able to assess risks in real time and provide reliable data support to optimize the operation and maintenance strategy of the system. Current research mainly focuses on the internal factors of the equipment and the analysis of failure modes in the onshore environment, while the understanding of the equipment status and fault development mechanism in the special offshore environment is still limited. In addition, the fault diagnosis and remote monitoring technology of offshore wind power systems has not been fully developed, various tasks have independent models, and the overall system redundancy is complex.

[0004] Therefore, there is currently a lack of a method that can accurately predict the performance and safety risks of offshore wind power systems. Summary of the invention

[0005] The embodiment of the present application provides a multi-task risk assessment method for an offshore wind power grid-connected system, which is used to solve the defects of the above-mentioned related technologies. The technical solution is as follows:

[0006] In a first aspect, an embodiment of the present application provides a multi-task risk assessment method for an offshore wind power grid-connected system, comprising:

[0007] Acquire an operation time series data sequence of an offshore wind power grid-connected system, and preprocess the operation time series data sequence;

[0008] Adding a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded;

[0009] Digitally encode the data sequence to be encoded, and construct a sample data set based on the digital sequence after the digital encoding;

[0010] Training the multi-task processing model based on the sample data set, and outputting the trained multi-task processing model;

[0011] The to-be-processed operating time sequence data sequence of the offshore wind power grid-connected system is processed based on the trained multi-task processing model, the future state probability of the offshore wind power grid-connected system is predicted according to the output result of the trained multi-task processing model, and a risk assessment result is output.

[0012] In an optional solution of the first aspect, the acquiring of the operation time series data sequence of the offshore wind power grid-connected system and preprocessing of the operation time series data sequence includes:

[0013] Get the operating sequence data sequence X of the offshore wind power grid-connected system t =[x1,x2,…,x t ];

[0014] Perform anomaly detection on the data in the runtime data sequence, and if there is abnormal data, use a symbol corresponding to the abnormal type to replace the original abnormal data; if there is missing data, use a symbol corresponding to the missing data to replace the original missing data;

[0015] Discretize the running time series data sequence, determine the discrete series, discretize the running time series data sequence into a time series discrete data sequence, and apply the formula:

[0016]

[0017] Among them, x t is an element in the runtime data sequence, x d is an element in the time-series discrete data sequence, x max is the maximum value in the runtime data sequence, x min is the minimum value in the runtime data sequence, It is a downward integer operation, N is the discrete series of time series continuous data, and ∈ is a parameter.

[0018] In an optional solution of the first aspect, adding a task header symbol, a start symbol, a separator symbol, and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded includes:

[0019] Cutting the discrete time series data sequence to obtain sub-data sequences of preset length;

[0020] For each sub-data sequence, if there is abnormal data, obtain the result data sequence after the abnormal data is processed by each task type, and pair the original data sequence before the sub-data sequence is processed by the abnormality detection and the result data sequence of each task type respectively;

[0021] Based on the paired original data sequence and result data sequence, add a task header symbol corresponding to the task type at the beginning of the original data sequence, add a start symbol after the task header symbol, add a separator symbol at the end of the original data sequence, add a result data sequence corresponding to the task type after the separator symbol, and add an end symbol at the end of the result data sequence;

[0022] Output the sample data sequence to be encoded.

[0023] In an optional solution of the first aspect, digitally encoding the to-be-encoded data sequence and constructing a sample data set based on the digitally encoded digital sequence includes:

[0024] Taking each data in the sample data sequence to be encoded as a subword, digitally encoding each data according to a preset encoding method to obtain a digital sequence after digital encoding;

[0025] Based on the original data sequence and the result data sequence after the original data sequence is processed by the corresponding task type, the digital sequence is displaced to generate input sample data corresponding to the original data sequence and output sample data corresponding to the result data sequence;

[0026] The sample data set is constructed based on all the input sample data and the output sample data.

[0027] In an optional solution of the first aspect, the training of the multi-task processing model based on the sample data set and outputting the trained multi-task processing model includes:

[0028] Training the multi-task processing model in batches based on the sample data set, and optimizing the network parameters of the multi-task processing model by gradient descent method;

[0029] For each batch of sample data, calculate the value of the cross entropy loss function;

[0030] Determining whether the multi-task processing model has converged according to the value of the cross entropy loss function of each batch;

[0031] If converged, the trained multi-task processing model is output.

[0032] In an optional solution of the first aspect, the processing of the to-be-processed operating time series data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model includes:

[0033] Acquire a sequence of operation time series data to be processed of an offshore wind power grid-connected system, and preprocess the sequence of operation time series data to be processed;

[0034] Adding a task header symbol, a start symbol and a separator symbol to the preprocessed sequence of runtime data to be processed, to obtain a sequence of data to be encoded;

[0035] Digitally encoding the data sequence to be encoded, and outputting a digital sequence to be processed;

[0036] Inputting the digital sequence to be processed into a trained multi-task processing model;

[0037] Execute the task corresponding to the task header symbol through the trained multi-task processing model, generate a predicted value, add the predicted value after the separator symbol, and output an updated digital sequence to be processed;

[0038] Determining whether the updated digital sequence to be processed meets a preset condition, and if the preset condition is met, performing a risk assessment of the offshore wind power grid-connected system based on the updated digital sequence to be processed;

[0039] Otherwise, the updated digital sequence to be processed is input into the trained multi-task processing model, and the process proceeds to the step of executing the task corresponding to the task header symbol through the trained multi-task processing model.

[0040] In an optional solution of the first aspect, the determining whether the updated digital sequence to be processed meets a preset condition includes:

[0041] If, after the trained multi-task processing model outputs the updated digital sequence to be processed, an end symbol is generated at the end of the updated digital sequence to be processed, or the number of digits of the updated digital sequence to be processed exceeds a preset digit threshold, then the updated digital sequence to be processed meets the preset condition;

[0042] Otherwise, the updated digital sequence to be processed does not meet the preset condition.

[0043] In a second aspect, the embodiment of the present application further provides a multi-task risk assessment device for an offshore wind power grid-connected system, comprising:

[0044] A data processing unit, used to obtain an operation time sequence data sequence of the offshore wind power grid-connected system and pre-process the operation time sequence data sequence;

[0045] A symbol processing unit, used for adding a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded;

[0046] A sample construction unit, used for digitally encoding the to-be-encoded data sequence, and constructing a sample data set based on the digitally encoded digital sequence;

[0047] A model training unit, used to train the multi-task processing model based on the sample data set and output the trained multi-task processing model;

[0048] The risk assessment unit is used to process the to-be-processed operating time sequence data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model, predict the future state probability of the offshore wind power grid-connected system according to the output result of the trained multi-task processing model, and output the risk assessment result.

[0049] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method provided in the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect is implemented.

[0050] In a fourth aspect, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided by the first aspect of the embodiment of the present application or any one of the implementations of the first aspect.

[0051] The beneficial effects brought about by the technical solutions provided by some embodiments of the present application include at least:

[0052] The embodiment of the present application provides a multi-task risk assessment method for an offshore wind power grid-connected system. By collecting the operating timing data of the offshore wind power grid-connected system, a symbol system is constructed to discretize multiple tasks and data. By constructing the symbol system and generating a corresponding multi-task processing sample set, the multi-task processing model is trained based on the sample set, thereby obtaining a multi-task processing model that can be applied to multiple tasks such as probability prediction, anomaly detection, and data filling.

[0053] By integrating multiple tasks into a multi-task processing model, the state management of the offshore wind power grid-connected system can be unified and simplified, which is conducive to improving the efficiency and reliability of the system.

[0054] The data output by the multi-task processing model can provide reliable data support for the risk assessment of the offshore wind power grid-connected system and can predict the status of the offshore wind power grid-connected system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 It is a flow chart of a multi-task risk assessment method for an offshore wind power grid-connected system provided in an embodiment of the present application;

[0057] Figure 2 It is a flow chart of a multi-task risk assessment method for an offshore wind power grid-connected system provided in an embodiment of the present application;

[0058] Figure 3 It is a structural schematic diagram of a multi-task risk assessment device for an offshore wind power grid-connected system provided in an embodiment of the present application;

[0059] Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0061] The terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules that are not listed, or may optionally include other steps or modules that are inherent to these processes, methods, products or devices.

[0062] It should be noted that the terms "first\second" involved in the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that "first\second" can be interchanged with a specific order or sequence where permitted. It should be understood that the objects distinguished by "first\second" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those described or illustrated herein.

[0063] The present application is described in detail below with reference to specific embodiments.

[0064] Next, combine Figure 1 , introduces a multi-task risk assessment method for an offshore wind power grid-connected system provided by an embodiment of the present application. For details, please refer to Figure 1 , Figure 1 The flowchart of a multi-task risk assessment method for an offshore wind power grid-connected system provided by an embodiment of the present application is shown. Figure 1 As shown, the method comprises the following steps:

[0065] S101, obtaining an operation time series data sequence of an offshore wind power grid-connected system, and preprocessing the operation time series data sequence;

[0066] S102, adding a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded;

[0067] S103, digitally encoding the to-be-encoded data sequence, and constructing a sample data set based on the digitally encoded digital sequence;

[0068] S104, training the multi-task processing model based on the sample data set, and outputting the trained multi-task processing model;

[0069] S105, processing the to-be-processed operating time sequence data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model, predicting the future state probability of the offshore wind power grid-connected system according to the output result of the trained multi-task processing model, and outputting a risk assessment result.

[0070] In some embodiments, in S101, the obtaining of the operating time series data sequence of the offshore wind power grid-connected system and preprocessing of the operating time series data sequence includes:

[0071] Get the operating sequence data sequence X of the offshore wind power grid-connected system t =[x1,x2,…,x t ];

[0072] Perform anomaly detection on the data in the runtime data sequence, and if there is abnormal data, use a symbol corresponding to the abnormal type to replace the original abnormal data; if there is missing data, use a symbol corresponding to the missing data to replace the original missing data;

[0073] Discretize the running time series data sequence, determine the discrete series, discretize the running time series data sequence into a time series discrete data sequence, and apply the formula:

[0074]

[0075] Among them, x t is an element in the runtime data sequence, x d is an element in the time series discrete data sequence, x max is the maximum value in the runtime data sequence, x min is the minimum value in the runtime data sequence, It is a downward integer operation, N is the discrete series of time series continuous data, and ∈ is a parameter.

[0076] Specifically, the monitoring system of the offshore wind farm can be used to collect operation time series data of various data of the offshore wind farm, such as power time series data.

[0077] Specifically, anomaly detection can be performed by manual inspection, and data that is obviously deviated from the normal range can be identified as abnormal data. <error>Replace the original data; find the missing data caused by various measurement reasons, and use the symbol <none>Replace the original data.

[0078] In some embodiments, in S102, a task header symbol, a start symbol, a separator symbol, and an end symbol are added to the preprocessed runtime data sequence to obtain a sample data sequence to be encoded, including:

[0079] Cutting the discrete time series data sequence to obtain sub-data sequences of preset length;

[0080] For each sub-data sequence, if there is abnormal data, obtain the result data sequence after the abnormal data is processed by each task type, and pair the original data sequence before the sub-data sequence is processed by the abnormality detection and the result data sequence of each task type respectively;

[0081] Based on the paired original data sequence and result data sequence, add a task header symbol corresponding to the task type at the beginning of the original data sequence, add a start symbol after the task header symbol, add a separator symbol at the end of the original data sequence, add a result data sequence corresponding to the task type after the separator symbol, and add an end symbol at the end of the result data sequence;

[0082] Output the sample data sequence to be encoded.

[0083] For example, for a prediction task, the discrete data X can be cut into d =[x1,x2,…,x t ], get the sub-data sequence X of preset length dnm =[x n ,x n+1 ,…,x m ], by adding the task header <predict>and the start symbol <s>, end symbol <\s> and other symbols, and get the form <predict> <s>x n ,x n+1 ,…,x m Sample of <\s>.

[0084] For example, a sub-data sequence that has been manually labeled to contain abnormal data may be selected to determine an original data sequence that has not been detected for abnormalities and corresponds to the sub-data sequence, for example, data sequence 1: <0> <1> <8> <none><4> ; After the anomaly detection task, it is determined that the data corresponding to 8 is abnormal data, and the sub-data sequence example is data sequence 2: <0> <1> <error> <none><4> , data sequence 2 is the result data sequence of the anomaly detection task; after performing the data filling task on the abnormal data, data sequence 3 is obtained: <0> <1> <2> <3> <4> , data sequence 3 is the result data sequence of the data filling task.

[0085] Data sequence 1 and data sequence 2 can be paired as samples corresponding to the anomaly detection task, and data sequence 2 and data sequence 3 can be paired as samples for the data filling task.

[0086] Specifically, you can use the task header <e-detect>and <e-repair>Respectively represent the anomaly detection task and the data filling task, using As a separator, use <s>Indicates the start symbol, and <\s> indicates the end symbol.

[0087] For example, the task header symbol of the anomaly detection task can be added at the beginning of data sequence 1. <e-detect>, and add a start symbol after the task header symbol, add a separator symbol at the end of data sequence 1, add the result data sequence of the anomaly detection task after the separator symbol, that is, data sequence 2, and add an end symbol at the end of data sequence 2 to obtain the sample data sequence to be encoded for the anomaly detection task: <e-detect> <s><0><1><8> <n> <4> <0><1> <e> <n><4> <\s>.

[0088] For example, similarly, the task header symbol of the data filling task can be added at the beginning of the data sequence 2. <e-repair>, and add a start symbol after the task header symbol, add a separator symbol at the end of data sequence 2, add the result data sequence of the data filling task after the separator symbol, that is, data sequence 3, and add an end symbol at the end of data sequence 3 to obtain the sample data sequence to be encoded for the data filling task: <e-repair> <s><0><1><8> <n> <4> <0> <1> <2> <3> <4> <\s>.

[0089] It can be understood that during the training process, samples can be constructed according to different task types, and the samples include original data and corresponding result data. During the application of the model, there is only the collected data to be processed, so it is only necessary to add the task header symbol and the start symbol at the beginning of the data to be processed, and add the separator symbol at the end of the data to be processed.

[0090] In some embodiments, in S103, digitally encoding the data sequence to be encoded and constructing a sample data set based on the digital sequence after digital encoding includes:

[0091] Taking each data in the sample data sequence to be encoded as a subword, digitally encoding each data according to a preset encoding method to obtain a digital sequence after digital encoding;

[0092] Based on the original data sequence and the result data sequence after the original data sequence is processed by the corresponding task type, the digital sequence is displaced to generate input sample data corresponding to the original data sequence and output sample data corresponding to the result data sequence;

[0093] The sample data set is constructed based on all the input sample data and the output sample data.

[0094] Specifically, for the data input into the model, each discretization level or a special character can be regarded as a subword, and each subword can correspond to a unique digital code. For time series data, first use the discretizer to discretize, use its discrete level as the code, and its code range can be 0 to N-1, and then encode the required special characters. For example <predict> 、 <e-detect>and <e-repair>Can be encoded as N, N+1, N+2 respectively. <s>、<\s>、 Can be encoded as N+3, N+4, N+5, <error>and <none>Can be encoded as N+6, N+7 respectively.

[0095] For example, the sample data sequence to be encoded: <predict> <s><0> <1> <2> <3> <4> <\s>, after digital encoding, the corresponding digital sequence can be obtained as N N+3 0 1 2 3 4N+4.

[0096] Exemplarily, a sample data set for model training can be constructed using a digital sequence corresponding to the sample data sequence to be encoded, and a multi-task training data set can be constructed using the digital sequences corresponding to anomaly detection and data filling, respectively. The original data and the processed data are staggered and used as input and output, respectively. For example, the prediction sequence N, N+3, 0, 1, 2, 3, 4, N+4 is shifted and aligned, with N N+3 0 1 2 3 4 as input and N+3 0 1 2 3 4N+4 as output.

[0097] In some embodiments, in S104, the training of the multi-task processing model based on the sample data set and outputting the trained multi-task processing model includes:

[0098] Training the multi-task processing model in batches based on the sample data set, and optimizing the network parameters of the multi-task processing model by gradient descent method;

[0099] For each batch of sample data, calculate the value of the cross entropy loss function;

[0100] Determining whether the multi-task processing model has converged according to the value of the cross entropy loss function of each batch;

[0101] If converged, the trained multi-task processing model is output.

[0102] In some embodiments, the process of constructing the multi-tasking model specifically includes:

[0103] First, an embedding layer is constructed to convert the discrete input representation into a continuous vector representation. The input vocabulary size is N+7, the embedding dimension is d, and a weight matrix W(N+7x d) can be used to represent the embedding layer.

[0104] For a given input x, whose index in the vocabulary is i, the embedding vector e corresponding to the input x is x , can be extracted from the embedding matrix W through a table lookup operation and is calculated as follows:

[0105] e x =W[i];

[0106] Furthermore, we build a causal self-attention module, including:

[0107] A single causal attention module mainly consists of two components. The first half is the multi-head attention mechanism. The three basic computational elements in multi-head attention are questions, keys, and values, represented by Q, K, and V, respectively, which have different actual fillings in different scenarios.

[0108] Q, K, and V can be mapped to multiple different semantic representation subspaces, namely:

[0109]

[0110] Where W i Q , W i K , W i V They represent the parameter matrices used to map Q, K, and V to the i-th subspace. The attention mechanism in each subspace is defined as:

[0111]

[0112] Among them, mask is a lower triangular matrix used to implement causality and ensure that only the current and previous time steps are considered when calculating attention. :

[0113] Then, the outputs of multiple subspaces are concatenated, namely:

[0114]

[0115] in, is the vector concatenation operation, W o It is the parameter matrix required for linear transformation after splicing, and h is the number of heads (number of subspaces) of the model.

[0116] The second half of the causal attention module is layer normalized and residual connected, namely:

[0117] out=LayerNorm(Add(Q+MultiHead(Q,K,V)));

[0118] Among them, layer normalization ensures the stability of data feature distribution, and residual connection prevents network degradation.

[0119] Furthermore, position encoding is introduced on top of each attention layer to preserve the temporal information of time series data by adding the position encoding vector to the input embedding vector.

[0120] Specifically, position encoding uses a combination of sine and cosine functions to generate a unique code for each position in the sequence, as follows:

[0121]

[0122] Among them, pos represents the position of the token, i represents the dimension index, and d model This encoding method makes the encoding of each position different from that of other positions, and positions with close distances have similar encodings, which helps the model capture the relative relationship between positions.

[0123] Finally, a multi-task integrated model is obtained by adding an embedding layer and repeatedly stacking causal self-attention modules.

[0124] Next, combine Figure 2 A multi-task risk assessment method for an offshore wind power grid-connected system provided in an embodiment of the present application is introduced. Figure 2 A flow chart of a multi-task risk assessment method for an offshore wind power grid-connected system is shown, specifically, the steps of processing the to-be-processed operating time sequence data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model include:

[0125] S201, obtaining a sequence of operation time series data to be processed of an offshore wind power grid-connected system, and preprocessing the sequence of operation time series data to be processed;

[0126] S202, adding a task header symbol, a start symbol and a separator symbol to the preprocessed sequence of runtime data to be processed, to obtain a sequence of data to be encoded;

[0127] S203, digitally encode the data sequence to be encoded, and output a digital sequence to be processed;

[0128] S204, inputting the digital sequence to be processed into a trained multi-task processing model;

[0129] S205, executing the task corresponding to the task header symbol through the trained multi-task processing model, generating a predicted value, adding the predicted value to the separator symbol, and outputting an updated digital sequence to be processed;

[0130] S206, determining whether the updated digital sequence to be processed meets a preset condition;

[0131] If the preset condition is met, then the step S207 is executed:

[0132] S207, performing risk assessment of the offshore wind power grid-connected system based on the updated digital sequence to be processed;

[0133] Otherwise, execute step S208:

[0134] S208, input the updated digital sequence to be processed into the trained multi-task processing model, go to the step of executing the task corresponding to the task head symbol through the trained multi-task processing model as described in S205, and repeat the steps of S205-S206 until the preset conditions are met.

[0135] It should be noted that the runtime data sequence to be processed in S201 can be a data sequence collected in real time or a data sequence collected in the past time period. The data of the future time period can be predicted based on the data of the past time period through the multi-task processing model.

[0136] Specifically, in S202, according to the type of task to be processed, the task header symbol of the corresponding task type can be added to the beginning of the data to be processed, and a start symbol can be inserted after the task header symbol, and a separator symbol can be added to the end of the data to be processed. For the specific process, please refer to the description of S102, which will not be repeated here.

[0137] Specifically, in S203, digital encoding may be performed based on the same method as in S103 to generate a digital sequence to be processed, which will not be described in detail herein.

[0138] Specifically, in S205, the multi-task processing model can identify the task header symbol in the input digital sequence to be processed, determine the type of task to be executed, and then generate a corresponding predicted value based on the task type, and fill the predicted value after the separator symbol to obtain an updated digital sequence to be processed. If the preset conditions are not met at this time, the updated digital sequence to be processed will be re-input into the multi-task processing model until the preset conditions are met.

[0139] For example, the data sequence to be encoded corresponding to the input digital sequence to be processed is <e-predict> <s><0><1><2> , after the model performs the prediction task, the predicted value after 2 is 4, and the obtained data sequence is expressed as: <e-predict> <s><0><1><2> <4> If the preset conditions are not met at this time, <e-predict> <s><0><1><2> <4> The corresponding digital sequence is re-entered into the model, and the model predicts that the value after 4 is 6, so the output data sequence is expressed as: <e-predict> <s><0><1><2> <4> <6> .

[0140] In some embodiments, determining whether the updated digital sequence to be processed meets a preset condition includes:

[0141] If, after the trained multi-task processing model outputs an updated digital sequence to be processed, an end symbol is generated at the end of the updated digital sequence to be processed, or the number of digits of the updated digital sequence to be processed exceeds a preset digit threshold, the updated digital sequence to be processed meets the preset condition, and an end symbol is added to the end of the updated digital sequence to be processed;

[0142] Otherwise, the updated digital sequence to be processed does not meet the preset condition.

[0143] For example, if the model determines that the prediction task should be terminated at this time, the end symbol <\s> can be output after the predicted value after 4 is 6. Alternatively, the prediction task is set to predict only two values. If the number of digits in the updated digital sequence to be processed exceeds the upper limit, it is judged that the preset conditions are met.

[0144] The final output data sequence <e-predict> <s><0><1><2> <4> <6> <\s>.

[0145] It should be noted that the input and output of the model are both digital sequences in the form of digital codes. After the output results, the data can be restored to the form before encoding based on the mapping relationship between the numerical values ​​and the digital codes. This embodiment of the present application is not limited to this.

[0146] The following are device embodiments of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0147] See next Figure 3 , is a structural diagram of a multi-task risk assessment device for an offshore wind power grid-connected system provided by an exemplary embodiment of the present application. The device can be implemented as all or part of a terminal through software, hardware, or a combination of both, and can also be integrated on a server as an independent module. A multi-task risk assessment device for an offshore wind power grid-connected system in an embodiment of the present application can be applied to a terminal or a cloud. The device 30 includes a data processing unit 301, a symbol processing unit 302, a sample construction unit 303, a model training unit 304, and a risk assessment unit 305, wherein:

[0148] The data processing unit 301 is used to obtain the operation time series data sequence of the offshore wind power grid-connected system and pre-process the operation time series data sequence;

[0149] The symbol processing unit 302 is used to add a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded;

[0150] The sample construction unit 303 is used to digitally encode the to-be-encoded data sequence, and construct a sample data set based on the digitally encoded digital sequence;

[0151] The model training unit 304 is used to train the multi-task processing model based on the sample data set and output the trained multi-task processing model;

[0152] The risk assessment unit 305 is used to process the pending operation time series data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model, predict the future state probability of the offshore wind power grid-connected system according to the output result of the trained multi-task processing model, and output the risk assessment result.

[0153] It should be noted that the device 30 provided in the above embodiment only uses the division of the above functional modules as an example when executing a multi-task risk assessment method for an offshore wind power grid-connected system. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device provided in the above embodiment and the embodiment of a multi-task risk assessment method for an offshore wind power grid-connected system belong to the same concept. The implementation process is detailed in the method embodiment and will not be repeated here.

[0154] An embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any of the above-mentioned method embodiments when executing the program.

[0155] See also Figure 4 , which is a structural block diagram of an electronic device provided in an embodiment of the present application.

[0156] like Figure 4 As shown, the electronic device 400 includes: a processor 401 and a memory 402 .

[0157] In the embodiment of the present application, the processor 401 is the control center of the computer system, which can be a processor of a physical machine or a processor of a virtual machine. The processor 401 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 401 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array).

[0158] The processor 401 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also called a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.

[0159] The memory 402 may include one or more computer-readable storage media, which may be non-transitory. The memory 402 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments of the present application, the non-transitory computer-readable storage medium in the memory 402 is used to store at least one instruction, which is used to be executed by the processor 401 to implement the method in the embodiment of the present application.

[0160] In some embodiments, the electronic device 400 further includes: a peripheral device interface 403 and at least one peripheral device 404. The processor 401, the memory 402 and the peripheral device interface 403 can be connected via a bus or a signal line. Each peripheral device 404 can be connected to the peripheral device interface 403 via a bus, a signal line or a circuit board. Specifically, the peripheral device 404 includes: a display screen, a camera and an audio circuit. The peripheral device interface 403 can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 401 and the memory 402.

[0161] In some embodiments of the present application, the processor 401, the memory 402, and the peripheral device interface 403 are integrated on the same chip or circuit board; in some other embodiments of the present application, any one or two of the processor 401, the memory 402, and the peripheral device interface 403 can be implemented on a separate chip or circuit board. This embodiment of the present application does not specifically limit this.

[0162] The electronic device structure block diagram shown in the embodiment of the present application does not constitute a limitation on the electronic device 400. The electronic device 400 may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.

[0163] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method of any of the above embodiments are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0164] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution can be essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application. < / s> < / e-predict> < / s> < / e-predict> < / s> < / e-predict> < / s> < / e-predict> < / s> < / e-predict> < / s> < / predict> < / none> < / error> < / s> < / predict> < / n> < / s> < / e-repair> < / n> < / e> < / n> < / s> < / e-detect> < / s> < / none> < / error> < / none> < / s> < / predict> < / s> < / predict> < / none> < / error>

Claims

1. A multi-task risk assessment method for offshore wind power grid-connected systems, characterized in that: include: Acquire an operation time series data sequence of an offshore wind power grid-connected system, and preprocess the operation time series data sequence; Adding a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded; Digitally encode the data sequence to be encoded, and construct a sample data set based on the digital sequence after the digital encoding; Training the multi-task processing model based on the sample data set, and outputting the trained multi-task processing model; The to-be-processed operating time sequence data sequence of the offshore wind power grid-connected system is processed based on the trained multi-task processing model, the future state probability of the offshore wind power grid-connected system is predicted according to the output result of the trained multi-task processing model, and a risk assessment result is output.

2. A multi-task risk assessment method for offshore wind power grid-connected system according to claim 1, characterized in that: The obtaining of the operation time series data sequence of the offshore wind power grid-connected system and preprocessing the operation time series data sequence includes: Get the operating sequence data sequence X of the offshore wind power grid-connected system t =[x1,x2,…,x t ]; Perform anomaly detection on the data in the runtime data sequence, and if there is abnormal data, use a symbol corresponding to the abnormal type to replace the original abnormal data; if there is missing data, use a symbol corresponding to the missing data to replace the original missing data; Discretize the running time series data sequence, determine the discrete series, discretize the running time series data sequence into a time series discrete data sequence, and apply the formula: Among them, x t is an element in the runtime data sequence, x d is an element in the time-series discrete data sequence, x max is the maximum value in the runtime data sequence, x min is the minimum value in the runtime data sequence, It is a downward integer operation, N is the discrete series of time series continuous data, and ∈ is a parameter.

3. The method according to claim 2, characterized in that The step of adding a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded includes: Cutting the discrete time series data sequence to obtain sub-data sequences of preset length; For each sub-data sequence, if there is abnormal data, obtain the result data sequence after the abnormal data is processed by each task type, and pair the original data sequence before the sub-data sequence is processed by the abnormality detection and the result data sequence of each task type respectively; Based on the paired original data sequence and result data sequence, add a task header symbol corresponding to the task type at the beginning of the original data sequence, add a start symbol after the task header symbol, add a separator symbol at the end of the original data sequence, add a result data sequence corresponding to the task type after the separator symbol, and add an end symbol at the end of the result data sequence; Output the sample data sequence to be encoded.

4. The method according to claim 3, characterized in that The step of digitally encoding the data sequence to be encoded and constructing a sample data set based on the digital sequence after digital encoding includes: Taking each data in the sample data sequence to be encoded as a subword, digitally encoding each data according to a preset encoding method to obtain a digital sequence after digital encoding; Based on the original data sequence and the result data sequence after the original data sequence is processed by the corresponding task type, the digital sequence is displaced to generate input sample data corresponding to the original data sequence and output sample data corresponding to the result data sequence; The sample data set is constructed based on all the input sample data and the output sample data.

5. The method according to claim 1, characterized in that The step of training the multi-task processing model based on the sample data set and outputting the trained multi-task processing model comprises: Training the multi-task processing model in batches based on the sample data set, and optimizing the network parameters of the multi-task processing model by gradient descent method; For each batch of sample data, calculate the value of the cross entropy loss function; Determining whether the multi-task processing model has converged according to the value of the cross entropy loss function of each batch; If converged, the trained multi-task processing model is output.

6. The method according to claim 1, characterized in that The processing of the to-be-processed operating time series data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model includes: Acquire a sequence of operation time series data to be processed of an offshore wind power grid-connected system, and preprocess the sequence of operation time series data to be processed; Adding a task header symbol, a start symbol and a separator symbol to the preprocessed sequence of runtime data to be processed, to obtain a sequence of data to be encoded; Digitally encoding the data sequence to be encoded, and outputting a digital sequence to be processed; Inputting the digital sequence to be processed into a trained multi-task processing model; Execute the task corresponding to the task header symbol through the trained multi-task processing model, generate a predicted value, add the predicted value after the separator symbol, and output an updated digital sequence to be processed; Determining whether the updated digital sequence to be processed meets a preset condition, and if the preset condition is met, performing a risk assessment of the offshore wind power grid-connected system based on the updated digital sequence to be processed; Otherwise, the updated digital sequence to be processed is input into the trained multi-task processing model, and the process proceeds to the step of executing the task corresponding to the task header symbol through the trained multi-task processing model.

7. The method according to claim 6, characterized in that The step of judging whether the updated digital sequence to be processed meets a preset condition includes: If, after the trained multi-task processing model outputs the updated digital sequence to be processed, an end symbol is generated at the end of the updated digital sequence to be processed, or the number of digits of the updated digital sequence to be processed exceeds a preset digit threshold, then the updated digital sequence to be processed meets the preset condition; Otherwise, the updated digital sequence to be processed does not meet the preset condition.

8. A multi-task risk assessment device for offshore wind power grid-connected systems, characterized in that: include: A data processing unit, used to obtain an operation time sequence data sequence of the offshore wind power grid-connected system and pre-process the operation time sequence data sequence; A symbol processing unit, used for adding a task header symbol, a start symbol, a separator symbol and an end symbol on the basis of the preprocessed runtime data sequence to obtain a sample data sequence to be encoded; A sample construction unit, used for digitally encoding the to-be-encoded data sequence, and constructing a sample data set based on the digitally encoded digital sequence; A model training unit, used to train the multi-task processing model based on the sample data set and output the trained multi-task processing model; The risk assessment unit is used to process the to-be-processed operating time sequence data sequence of the offshore wind power grid-connected system based on the trained multi-task processing model, predict the future state probability of the offshore wind power grid-connected system according to the output result of the trained multi-task processing model, and output the risk assessment result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A non-transitory 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 the method according to any one of claims 1 to 7 are implemented.

Images