A temperature trend warning method and system for offshore wind power cables

Abstract

The present invention discloses a temperature trend warning method and system for offshore wind power cables. The method comprises: collecting historical temperature data and corresponding operating condition data of offshore wind farm cables under operating conditions, and preprocessing the historical temperature data and the operating condition data to obtain preprocessed data; establishing a cable temperature multi-step prediction model based on an extreme learning machine, and using the preprocessed data to train and verify the cable temperature multi-step prediction model; collecting real-time temperature data and corresponding operating condition data of offshore wind farm cables, and inputting them into the cable temperature multi-step prediction model to predict the real-time state trend of the cable temperature, thereby obtaining the real-time state trend data of the cable temperature; comparing the real-time state trend data of the cable temperature with a dynamic safety trend threshold interval of the cable temperature under the current operating conditions; if the real-time state trend data of the cable temperature is within the dynamic safety trend threshold interval of the cable temperature, executing the previous step; otherwise, issuing an alarm.

Description

Technical Field

[0001] The present invention relates to the field of offshore wind power cable temperature trend early warning, and in particular to an offshore wind power cable temperature trend early warning method and system. Background Art

[0002] With the vigorous development of offshore wind power generation, the application of submarine high-voltage cables will become more and more extensive. However, due to the uncertainty of operation and submarine environment, submarine cables are prone to various faults such as insulation damage and anchor damage. Once a submarine high-voltage cable fails and stops operating, it will cause very large economic losses. Therefore, in order to ensure the safe operation of submarine high-voltage cables, real-time monitoring of the operating status of submarine high-voltage cables is very necessary.

[0003] Currently, offshore wind farms are still in the stage of monitoring real-time values for cable temperature detection. Workers rely on experience to determine whether there is a problem with the wind turbine, which places high demands on the workers. Faults may not be discovered in a timely manner, leading to increased losses. Summary of the Invention

[0004] The technical problem to be solved by the present invention is that the current detection of cable temperature in offshore wind farms is still in the stage of monitoring real-time values. Workers rely on experience to judge whether there is a problem with the wind turbine, which places high demands on the workers, or they cannot discover the fault in time, thereby expanding the losses. The present invention provides an offshore wind power cable temperature trend warning method. The present invention also provides an offshore wind power cable temperature trend warning system, which can realize wind farm cable temperature status trend warning based on the multi-step prediction trend of the extreme learning machine and the slope trend constructed according to the predicted value fitting, combined with the historical normal trend of the parameters and the safety slope trend threshold. The system has a simple structure and is easy to use, and is used to solve the defects caused by the existing technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, a temperature trend warning method for offshore wind power cables is provided, comprising the following steps:

[0007] Step 1: Collect historical temperature data of offshore wind farm cables under working conditions and corresponding working condition data, and preprocess the historical temperature data and the working condition data to obtain preprocessed data;

[0008] Step 2: Establishing a cable temperature multi-step prediction model based on an extreme learning machine, and using the preprocessed data to train and verify the cable temperature multi-step prediction model;

[0009] Step 3: Collecting real-time temperature data and corresponding operating condition data of the offshore wind farm cables, and inputting them into the cable temperature multi-step prediction model to perform real-time state trend prediction of the cable temperature, thereby obtaining real-time state trend data of the cable temperature;

[0010] Step 4: Compare the cable temperature real-time status trend data with the cable temperature dynamic safety trend threshold range under the current working conditions;

[0011] If the cable temperature real-time status trend data is within the cable temperature dynamic safety trend threshold range, executing step 3;

[0012] If the cable temperature real-time status trend data is not within the cable temperature dynamic safety trend threshold range, an alarm is issued.

[0013] In the above-mentioned offshore wind power cable temperature trend warning method, the process of preprocessing the operating condition data and the historical temperature data in step 1 to obtain the preprocessed data is as follows:

[0014] Perform data cleaning, labeling, and storage.

[0015] In the above-mentioned offshore wind power cable temperature trend warning method, the method for establishing a multi-step prediction model for cable real-time temperature based on an extreme learning machine in step 2 is as follows:

[0016] Step 21: Use the state multi-step prediction structure of the extreme learning machine to establish a multi-step prediction model;

[0017] Step 22: Input the historical temperature data into the input layer of the extreme learning machine, determine the weight matrix of the hidden layer and the bias allocation random value to calculate the output hidden layer output matrix, and obtain a unique least squares minimum norm solution by solving the non-consistent equation, thereby training to obtain m prediction results for the historical temperature data;

[0018] Step 23: Based on the m prediction results, a linear fit is performed between the current prediction and the previous prediction using the least square method, using a monotonic function straight line. Get real-time status trend change data of cable temperature;

[0019] in, is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend;

[0020] Step 24: Obtain historical operating data of the cable to verify the accuracy of the multi-step prediction model, thereby obtaining the cable temperature multi-step prediction model.

[0021] In the above-mentioned offshore wind power cable temperature trend early warning method, in step 22, the historical temperature data is input into the input layer of the extreme learning machine, and the weight matrix and bias allocation random value of the hidden layer are determined to calculate and output the hidden layer output matrix. The specific method of training and obtaining m prediction results of the historical temperature data by solving the non-consistent equation to obtain a unique least squares minimum norm solution is as follows:

[0022] Assume that there are N temperature data samples recorded as ( , ),in, For historical sequence, , is a multi-step prediction sequence, , , the forecast period is , the input layer of each temperature data sample has n dimensions, the hidden layer has L dimensions, and the output layer has m dimensions. As shown in Equation 1-1, the output of a single hidden layer neural network is expressed as:

[0023] (1-1);

[0024] Where y0 is, is the activation function, is the input weight, is the output weight, The bias of the hth hidden layer unit;

[0025] After formula (1-1) is written into matrix form, it is shown as (1-2);

[0026] (1-2);

[0027] Where H is the output matrix of the hidden layer:

[0028] (1-3);

[0029] Output matrix:

[0030] (1-4);

[0031] Output weight matrix:

[0032] (1-5);

[0033] When the input weight and hidden layer bias Randomly determined as and , then the output matrix H of the hidden layer is also determined as ;

[0034] When H and Already determined, the output weight Then we can get from formula (1-6):

[0035] (1-6);

[0036] in is the Moore-Penrose generalized inverse matrix of the matrix H;

[0037] because , and the number of samples N>>L, so It can be obtained from formula (1-7):

[0038] (1-7);

[0039] in, and are mutually orthogonal matrices, The matrix is a diagonal matrix with singular values on the main diagonal, and is a diagonal matrix Take the reciprocal of the non-zero elements and then transpose;

[0040] in , , It can be obtained by decomposing the H matrix into singular values according to formula (1-8): (1-8);

[0041] in By the left singular vector composition, The right singular vector composition, The main diagonal is a singular value The diagonal matrix composed of

[0042] in , , Calculated by formula (1-9), formula (1-10), and formula (1-11) respectively;

[0043] (1-9);

[0044] (1-10);

[0045] (1-11);

[0046] Since the number of training samples N>>L, the number of rows in the H matrix is much larger than the number of columns, so is a non-uniform equation whose solution It is the least squares minimum norm solution and is unique, thereby obtaining m temperature prediction points.

[0047] In the above-mentioned offshore wind power cable temperature trend early warning method, the specific method of performing linear fitting between the current prediction and the previous prediction results using the least squares method according to the m predicted values in step 23 is as follows:

[0048] The least squares method is used to perform linear fitting on the prediction results of the current prediction and the previous prediction to obtain the linear fitting function ,in , is the prediction period, k is the slope, a is the constant term of the fitting function, which can be derived from multiple (x, y) values, b is the constant term of the fitting function, which can be derived from multiple (x, y) values, and y is the vertical coordinate value in the one-dimensional coordinate;

[0049] According to the loss function calculation formula The error E between the true value and the fitted value is calculated, where is the true value, is the corresponding fitted value;

[0050] Treat the loss function as a multivariate function to minimize the loss function and write the loss function as:

[0051] (2-1);

[0052] According to formula (2-1), and Find the partial derivative and solve it according to formula (2-2) and Make the partial derivative equal to 0;

[0053] (2-2);

[0054] Transform formula (2-2) into formula (2-3);

[0055] (2-3);

[0056] The two multivariate equations (2-2) and (2-2) can be solved to obtain and The value of is shown in formula (2-4);

[0057] (2-4);

[0058] According to formula (2-4), the linear fitting state trend function is obtained, and the monotonic function straight line is used. ,in is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend.

[0059] In the above-mentioned offshore wind power cable temperature trend warning method, the calculation method of the cable temperature dynamic safety trend threshold interval is as follows:

[0060] Gaussian mixture clustering is used to analyze a set of one-dimensional time series slope trend data samples. , the Gaussian mixture model assumes that all data samples are composed of It is a mixture of Gaussian distribution models, as shown in formula (3-1);

[0061] (3-1);

[0062] in, Represents the weights between Gaussian distribution models and satisfies , in formula (3-1) is a Gaussian distribution model in the Gaussian mixture model. As shown in formula (3-2), the Gaussian mixture model uses the maximum likelihood estimation method to estimate the value of the parameter;

[0063] (3-2);

[0064] As shown in formula (3-3), since there are too many parameters in the Gaussian mixture model, the result cannot be obtained directly by derivation, so the EM algorithm is used to iteratively solve the model;

[0065] (3-3);

[0066] The EM algorithm solution process is as follows:

[0067] First initialize the parameters ;

[0068] E-step: Calculate each data according to the current parameters From submodel possibility;

[0069] ;

[0070] M-step: Calculate the model parameters for the next iteration;

[0071] ;

[0072] ;

[0073] ;

[0074] Repeat the calculation of E-step and M-step until the algorithm converges, and solve the mean vector of the mixed Gaussian model through the EM algorithm , variance direction , and the weight coefficient vector , then the mean of the entire sample can be derived from formula (3-4);

[0075] (3-4);

[0076] According to the interval distribution estimation of the Gaussian probability distribution function, taking 95% confidence level, the upper limit vector of the sample data is obtained through formula (3-5) and formula (3-6): With the lower limit vector ;

[0077] (3-5);

[0078] (3-6);

[0079] The upper limit of the entire sample can be obtained by using formula (3-7) and formula (3-8) With lower limit ;

[0080] (3-7);

[0081] (3-8);

[0082] Get the dynamic installation trend threshold interval [ , ].

[0083] In the second aspect, an offshore wind power cable temperature trend warning system includes a historical data processing module, a prediction module establishment module, a real-time data acquisition module, a comparison module, and an alarm module;

[0084] The historical data processing module is used to collect historical temperature data of offshore wind farm cables under working conditions and corresponding working condition data, and preprocess the historical temperature data and the working condition data to obtain preprocessed data;

[0085] The prediction module establishment module is used to establish a cable temperature multi-step prediction model based on an extreme learning machine, and use the preprocessed data to train and verify the cable temperature multi-step prediction model;

[0086] The real-time data acquisition module is used to collect real-time temperature data and corresponding operating condition data of offshore wind farm cables, and transmit them to the cable temperature multi-step prediction model to perform real-time cable temperature state trend prediction to obtain real-time cable temperature state trend data;

[0087] The comparison module is used to compare the cable temperature real-time status trend data with the cable temperature dynamic safety trend threshold interval under the current working conditions; if the cable temperature real-time status trend data is within the cable temperature dynamic safety trend threshold interval, feedback data is generated and transmitted to the real-time data acquisition module;

[0088] If the cable temperature real-time status trend data is not within the cable temperature dynamic safety trend threshold interval, generating alarm data transmitted to the alarm module;

[0089] The alarm module is used for generating an alarm.

[0090] In a third aspect, a chip comprises: a processor for calling and running a computer program from a memory, so that a device equipped with the chip executes: a method as described in any one of the first aspects.

[0091] The technical solution provided by the above-mentioned offshore wind power cable temperature trend warning method and system of the present invention has the following technical effects:

[0092] The present invention mainly proposes an early warning method based on dual trend analysis combined with machine learning, namely, based on the multi-step prediction trend of the extreme learning machine and the slope trend constructed according to the predicted value fitting, combined with the historical normal trend of the parameters and the safety slope trend threshold to realize the temperature status trend warning of the wind farm cable. The method has a simple structure and is easy to use. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 This is a flow chart of a temperature trend warning method for offshore wind power cables according to the present invention;

[0094] Figure 2 This is a structural diagram of an offshore wind power cable temperature trend early warning system of the present invention.

[0095] The accompanying drawings are numerals as follows:

[0096] Historical data processing module 100, prediction module establishment module 200, real-time data acquisition module 300, comparison module 400, alarm module 500. DETAILED DESCRIPTION

[0097] In order to make the technical means, creative features, objectives and effects of the invention easy to understand, the technical solutions in the embodiments of the present invention are clearly and completely described below in combination with specific illustrations. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0098] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0099] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings in this specification are only used to match the contents disclosed in the specification so that people familiar with this technology can understand and read them. They are not used to limit the conditions under which the present invention can be implemented. Therefore, they have no substantive technical significance. Any modification of the structure, change in the proportion relationship or adjustment of the size should still fall within the scope of the technical content disclosed in the present invention without affecting the efficacy and purpose that can be achieved by the present invention.

[0100] At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in this specification are only for the convenience of description and are not used to limit the scope of implementation of the present invention. Changes or adjustments to their relative relationships should be regarded as the scope of implementation of the present invention without substantially changing the technical content.

[0101] The first embodiment of the present invention provides an offshore wind power cable temperature trend warning method, which aims to realize wind farm cable temperature status trend warning based on the multi-step prediction trend of the extreme learning machine and the slope trend constructed according to the predicted value fitting, combined with the historical normal trend of the parameters and the safety slope trend threshold. The method has a simple structure and is easy to use.

[0102] like Figure 1 As shown, in the first aspect, the first embodiment, a temperature trend warning method for offshore wind power cables, which includes the following steps:

[0103] Step 1: Collect historical temperature data and corresponding operating condition data of the offshore wind farm cables under working conditions, and preprocess the historical temperature data and operating condition data to obtain preprocessed data;

[0104] Step 2: Establish a multi-step cable temperature prediction model based on the extreme learning machine, and use the preprocessed data to train and verify the cable temperature multi-step prediction model;

[0105] Step 3: Collect the real-time temperature data and corresponding operating condition data of the offshore wind farm cables, and input them into the cable temperature multi-step prediction model to predict the real-time state trend of the cable temperature, and obtain the real-time state trend data of the cable temperature;

[0106] Step 4: Compare the real-time cable temperature status trend data with the cable temperature dynamic safety trend threshold range under the current working conditions;

[0107] If the cable temperature real-time status trend data is within the cable temperature dynamic safety trend threshold range, proceed to step 3;

[0108] If the real-time status trend data of the cable temperature is not within the cable temperature dynamic safety trend threshold range, an alarm will be issued.

[0109] In the above-mentioned offshore wind power cable temperature trend warning method, the process of preprocessing the operating condition data and historical temperature data to obtain preprocessed data in step 1 is as follows:

[0110] Perform data cleaning, labeling, and storage.

[0111] In the above-mentioned offshore wind power cable temperature trend warning method, the method for establishing a multi-step prediction model for cable real-time temperature based on an extreme learning machine in step 2 is as follows:

[0112] Step 21: Use the state multi-step prediction structure of the extreme learning machine to establish a multi-step prediction model;

[0113] Step 22: Input the historical temperature data into the input layer of the extreme learning machine, determine the weight matrix of the hidden layer and the bias assignment random value to calculate the output hidden layer output matrix, and obtain a unique least squares minimum norm solution by solving the non-consistent equation, thereby training to obtain m prediction results of the historical temperature data;

[0114] Step 23: Perform linear fitting between the current prediction and the previous prediction using the least squares method based on the m prediction values, using a monotonic function straight line. Get real-time status trend change data of cable temperature;

[0115] in, is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend;

[0116] Step 24: Obtain historical operating data of the cable to verify the accuracy of the multi-step prediction model, thereby obtaining a multi-step prediction model for cable temperature.

[0117] In the above-mentioned offshore wind power cable temperature trend early warning method, in step 22, historical temperature data is input into the input layer of the extreme learning machine, and the weight matrix and bias distribution random value of the hidden layer are determined to calculate and output the hidden layer output matrix. By solving the non-consistent equation to obtain a unique least squares minimum norm solution, the specific method for training to obtain m prediction results of historical temperature data is as follows:

[0118] Assume that there are N temperature data samples recorded as ( , ),in, For historical sequence, , is a multi-step prediction sequence, , , the forecast period is , the input layer of each temperature data sample has n dimensions, the hidden layer has L dimensions, and the output layer has m dimensions. As shown in Equation 1-1, the output of a single hidden layer neural network is expressed as:

[0119] (1-1);

[0120] Where y0 is, is the activation function, is the input weight, is the output weight, The bias of the hth hidden layer unit;

[0121] After formula (1-1) is written into matrix form, it is shown as (1-2);

[0122] (1-2);

[0123] Where H is the output matrix of the hidden layer:

[0124] (1-3);

[0125] Output matrix:

[0126] (1-4);

[0127] Output weight matrix:

[0128] (1-5);

[0129] When the input weight and hidden layer bias Randomly determined as and , then the output matrix H of the hidden layer is also determined as ;

[0130] When H and Already determined, the output weight Then we can get from formula (1-6):

[0131] (1-6);

[0132] in is the Moore-Penrose generalized inverse matrix of the matrix H;

[0133] because , and the number of samples N>>L, so It can be obtained from formula (1-7):

[0134] (1-7);

[0135] in, and are mutually orthogonal matrices, The matrix is a diagonal matrix with singular values on the main diagonal, and is a diagonal matrix Take the reciprocal of the non-zero elements and then transpose;

[0136] in , , It can be obtained by decomposing the H matrix into singular values according to formula (1-8): (1-8);

[0137] in By the left singular vector composition, The right singular vector composition, The main diagonal is a singular value The diagonal matrix composed of

[0138] in , , Calculated by formula (1-9), formula (1-10), and formula (1-11) respectively;

[0139] (1-9);

[0140] (1-10);

[0141] (1-11);

[0142] Since the number of training samples N>>L, the number of rows in the H matrix is much larger than the number of columns, so is a non-uniform equation whose solution It is the least squares minimum norm solution and is unique, thus obtaining m temperature prediction points.

[0143] In the above-mentioned offshore wind power cable temperature trend early warning method, the specific method of linearly fitting the prediction results of the current prediction and the previous prediction using the least squares method according to the m prediction values in step 23 is as follows:

[0144] The least squares method is used to linearly fit the current prediction and the previous prediction results to obtain the linear fitting function ,in , is the prediction period, k is the slope, a is the constant term of the fitting function, which can be derived from multiple (x, y) values, b is the constant term of the fitting function, which can be derived from multiple (x, y) values, and y is the vertical coordinate value in the one-dimensional coordinate;

[0145] According to the loss function calculation formula The error E between the true value and the fitted value is calculated, where is the true value, is the corresponding fitted value;

[0146] Treat the loss function as a multivariate function to minimize the loss function and write the loss function as:

[0147] (2-1);

[0148] According to formula (2-1), and Find the partial derivative and solve it according to formula (2-2) and Make the partial derivative equal to 0;

[0149] (2-2);

[0150] Transform formula (2-2) into formula (2-3);

[0151] (2-3);

[0152] The two multivariate equations (2-2) and (2-2) can be solved to obtain and The value of is shown in formula (2-4);

[0153] (2-4);

[0154] According to formula (2-4), the linear fitting state trend function is obtained, and the monotonic function straight line is used. ,in is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend.

[0155] In the above-mentioned offshore wind power cable temperature trend warning method, the calculation method of the cable temperature dynamic safety trend threshold range is as follows:

[0156] Gaussian mixture clustering is used to analyze a set of one-dimensional time series slope trend data samples. , the Gaussian mixture model assumes that all data samples are composed of It is a mixture of Gaussian distribution models, as shown in formula (3-1);

[0157] (3-1);

[0158] in, Represents the weights between Gaussian distribution models and satisfies , in formula (3-1) is a Gaussian distribution model in the Gaussian mixture model. As shown in formula (3-2), the Gaussian mixture model uses the maximum likelihood estimation method to estimate the value of the parameter;

[0159] (3-2);

[0160] As shown in formula (3-3), since there are too many parameters in the Gaussian mixture model, the result cannot be obtained directly by derivation, so the EM algorithm is used to iteratively solve the model;

[0161] (3-3);

[0162] The EM algorithm solution process is as follows:

[0163] First initialize the parameters ;

[0164] E-step: Calculate each data according to the current parameters From submodel possibility;

[0165] ;

[0166] M-step: Calculate the model parameters for the next iteration;

[0167] ;

[0168] ;

[0169] ;

[0170] Repeat the calculation of E-step and M-step until the algorithm converges, and solve the mean vector of the mixed Gaussian model through the EM algorithm , variance direction , and the weight coefficient vector , then the mean of the entire sample can be derived from formula (3-4);

[0171] (3-4);

[0172] According to the interval distribution estimation of the Gaussian probability distribution function, taking 95% confidence level, the upper limit vector of the sample data is obtained through formula (3-5) and formula (3-6): With the lower limit vector ;

[0173] (3-5);

[0174] (3-6);

[0175] The upper limit of the entire sample can be obtained by using formula (3-7) and formula (3-8) With lower limit ;

[0176] (3-7);

[0177] (3-8);

[0178] Get the dynamic installation trend threshold interval [ , ].

[0179] like Figure 2 As shown, in the second aspect, the second embodiment, an offshore wind power cable temperature trend warning system, which includes a historical data processing module 100, a prediction module establishment module 200, a real-time data acquisition module 300, a comparison module 400, and an alarm module 500;

[0180] The historical data processing module 100 is used to collect historical temperature data of offshore wind farm cables under working conditions and corresponding working condition data, and preprocess the historical temperature data and working condition data to obtain preprocessed data;

[0181] The prediction module establishment module 200 is used to establish a cable temperature multi-step prediction model based on an extreme learning machine, and use preprocessed data to train and verify the cable temperature multi-step prediction model;

[0182] The real-time data acquisition module 300 is used to collect the real-time temperature data and corresponding operating condition data of the offshore wind farm cables, and transmit them to the cable temperature multi-step prediction model to perform real-time cable temperature state trend prediction to obtain the real-time cable temperature state trend data;

[0183] The comparison module 400 is used to compare the real-time status trend data of the cable temperature with the cable temperature dynamic safety trend threshold range under the current working conditions; if the real-time status trend data of the cable temperature is within the cable temperature dynamic safety trend threshold range, feedback data is generated and transmitted to the real-time data acquisition module 300;

[0184] If the cable temperature real-time status trend data is not within the cable temperature dynamic safety trend threshold range, alarm data is generated and transmitted to the alarm module 500;

[0185] The alarm module is used to issue an alarm 500.

[0186] In a third aspect, a chip comprises: a processor for calling and running a computer program from a memory, so that a device equipped with the chip executes: a method as in any one of the first aspects.

[0187] For example, the memory may include random access memory, flash memory, read-only memory, programmable read-only memory, non-volatile memory or registers;

[0188] The processor may be a central processing unit (CPU) or a graphics processing unit (GPU). The memory may store executable instructions.

[0189] The processor may execute the execution instructions stored in the memory to implement the various processes described herein.

[0190] It will be appreciated that the memory in this embodiment may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories;

[0191] Among them, the non-volatile memory can be ROM (Read-Only Memory), PROM (Programmable ROM), EPROM (Erasable PROM), EEPROM (Electrically EPROM) or flash memory.

[0192] The volatile memory may be RAM (Random Access Memory), which is used as an external cache;

[0193] By way of example and not limitation, many forms of RAM are available, such as SRAM (Static RAM), DRAM (Dynamic RAM), SDRAM (Synchronous DRAM), DDR SDRAM (Double Data Rate SDRAM), ESDRAM (Enhanced SDRAM), SLDRAM (Synchlink DRAM), and DRRAM (Direct Rambus RAM). The memories described herein are intended to include, but are not limited to, these and any other suitable types of memory.

[0194] In some embodiments, the memory stores the following elements, upgrade packages, executable units or data structures, or a subset thereof, or an extended set thereof: an operating system and applications;

[0195] The operating system includes various system programs, such as the framework layer, core library layer, and driver layer, which are used to implement various basic services and process hardware-based tasks.

[0196] The application program includes various application programs for implementing various application services. The program for implementing the method of the embodiment of the present invention can be included in the application program.

[0197] It will be apparent to those skilled in the art that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of software and electronic hardware;

[0198] Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution;

[0199] Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0200] In the embodiments of the present application, the disclosed systems, devices, and methods may be implemented in other ways;

[0201] For example, the division of units or modules is only a logical function division, and there may be other division methods in actual implementation;

[0202] For example, multiple units or modules or components may be combined or may be integrated into another system;

[0203] In addition, each functional unit or module in the embodiment of the present application may be integrated into one processing unit or module, or may exist as a separate physical entity, etc.

[0204] It should be understood that in the various embodiments of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0205] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a machine-readable storage medium;

[0206] Therefore, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a machine-readable storage medium and can include several instructions for causing an electronic device to execute all or part of the processes of the technical solution described in the embodiments of the present application;

[0207] The above-mentioned storage medium may include ROM, RAM, removable disk, hard disk, magnetic disk or optical disk, etc., which can store program codes.

[0208] In summary, the present invention provides an offshore wind power cable temperature trend warning method and system, which is based on the multi-step prediction trend of the extreme learning machine and the slope trend constructed according to the predicted value fitting, combined with the historical normal trend of the parameters and the safety slope trend threshold to achieve wind farm cable temperature status trend warning. It has a simple structure and is easy to use.

[0209] The above describes specific embodiments of the invention. It should be understood that the invention is not limited to the specific embodiments described above. Devices and structures not described in detail should be understood to be implemented in a common manner in the art. Those skilled in the art may make various modifications, variations, or simple deductions, deformations, or substitutions within the scope of the claims, without affecting the essence of the invention.

Claims

1. A temperature trend warning method for offshore wind power cables, characterized in that: The following steps are involved: Step 1: Collect historical temperature data of offshore wind farm cables under working conditions and corresponding working condition data, and preprocess the historical temperature data and the working condition data to obtain preprocessed data; Step 2: Establishing a cable temperature multi-step prediction model based on an extreme learning machine, and using the preprocessed data to train and verify the cable temperature multi-step prediction model; Step 3: Collecting real-time temperature data and corresponding operating condition data of the offshore wind farm cables, and inputting them into the cable temperature multi-step prediction model to perform real-time state trend prediction of the cable temperature, thereby obtaining real-time state trend data of the cable temperature; Step 4: Compare the cable temperature real-time status trend data with the cable temperature dynamic safety trend threshold range under the current working conditions; If the cable temperature real-time status trend data is within the cable temperature dynamic safety trend threshold range, executing step 3; If the cable temperature real-time status trend data is not within the cable temperature dynamic safety trend threshold range, an alarm is issued; The method for establishing a multi-step prediction model for cable real-time temperature based on the extreme learning machine in step 2 is as follows: Step 21: Use the state multi-step prediction structure of the extreme learning machine to establish a multi-step prediction model; Step 22: Input the historical temperature data into the input layer of the extreme learning machine, determine the weight matrix of the hidden layer and the bias allocation random value to calculate the output hidden layer output matrix, and obtain a unique least squares minimum norm solution by solving the non-consistent equation, thereby training to obtain m prediction results for the historical temperature data; Step 23: Based on the m prediction results, a linear fit is performed between the current prediction and the previous prediction using the least square method, using a monotonic function straight line. Get real-time status trend change data of cable temperature; in, is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend; Step 24: Obtain historical operating data of the cable to verify the accuracy of the multi-step prediction model, thereby obtaining the cable temperature multi-step prediction model.

2. The offshore wind power cable temperature trend early warning method according to claim 1, characterized in that: The process of preprocessing the operating condition data and the historical temperature data in step 1 to obtain the preprocessed data is as follows: Perform data cleaning, labeling, and storage.

3. The offshore wind power cable temperature trend early warning method according to claim 1, characterized in that: In step 22, the historical temperature data is input into the input layer of the extreme learning machine, and the weight matrix of the hidden layer and the bias allocation random value are determined to calculate the output hidden layer output matrix. The unique least squares minimum norm solution is obtained by solving the non-consistent equation, thereby training to obtain m prediction results for the historical temperature data. The specific method is as follows: Assume that there are N temperature data samples recorded as ( , ),in, For historical sequence, , is a multi-step prediction sequence, , , the forecast period is , the input layer of each temperature data sample has n dimensions, the hidden layer has L dimensions, and the output layer has m dimensions. As shown in Equation 1-1, the output of a single hidden layer neural network is expressed as: （1-1）； Where y0 is, is the activation function, is the input weight, is the output weight, The bias of the hth hidden layer unit; After formula (1-1) is written in matrix form, it becomes as shown in (1-2): （1-2）； Where H is the output matrix of the hidden layer: （1-3）； Output matrix: （1-4）； Output weight matrix: （1-5）； When the input weight and hidden layer bias Randomly determined as and , then the output matrix H of the hidden layer is also determined as ; When H and Already determined, the output weight Then we can get from formula (1-6): （1-6）； in is the Moore-Penrose generalized inverse matrix of the matrix H; because , and the number of samples N>>L, so It can be obtained from formula (1-7): （1-7）； in, and are mutually orthogonal matrices, The matrix is a diagonal matrix with singular values on the main diagonal, and is a diagonal matrix Take the reciprocal of the non-zero elements and then transpose; in , , It can be obtained by decomposing the H matrix into singular values according to formula (1-8): (1-8); in By the left singular vector composition, The right singular vector composition, The main diagonal is a singular value The diagonal matrix composed of in , , Calculated by formula (1-9), formula (1-10), and formula (1-11) respectively; （1-9）； （1-10）； （1-11）； Since the number of training samples N>>L, the number of rows in the H matrix is much larger than the number of columns, so is a non-uniform equation whose solution It is the least squares minimum norm solution and is unique, thereby obtaining m temperature prediction points.

4. The offshore wind power cable temperature trend early warning method according to claim 3, characterized in that: In step 23, the specific method of performing linear fitting between the current prediction and the prediction result of the previous prediction using the least square method based on the m prediction results is as follows: The least squares method is used to perform linear fitting on the prediction results of the current prediction and the previous prediction to obtain the linear fitting function ,in , is the prediction period, k is the slope, a is the constant term of the fitting function, which can be derived from multiple (x, y) values, b is the constant term of the fitting function, which can be derived from multiple (x, y) values, and y is the vertical coordinate value in the one-dimensional coordinate; According to the loss function calculation formula The error E between the true value and the fitted value is calculated, where is the true value, is the corresponding fitted value; Treat the loss function as a multivariate function to minimize the loss function and write the loss function as: （2-1）； According to formula (2-1), and Find the partial derivative and solve it according to formula (2-2) and Make the partial derivative equal to 0; （2-2）； Transform formula (2-2) into formula (2-3); （2-3）； The two multivariate equations (2-2) and (2-2) can be solved to obtain and The value of is shown in formula (2-4); （2-4）； According to formula (2-4), the linear fitting state trend function is obtained, and the monotonic function straight line is used. ,in is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend.

5. The offshore wind power cable temperature trend early warning method according to claim 4, characterized in that: The calculation method of the cable temperature dynamic safety trend threshold interval is as follows: Gaussian mixture clustering is used to analyze a set of one-dimensional time series slope trend data samples. , the Gaussian mixture model assumes that all data samples are composed of It is a mixture of Gaussian distribution models, as shown in formula (3-1); （3-1）； in, Represents the weights between Gaussian distribution models and satisfies , in formula (3-1) is a Gaussian distribution model in the Gaussian mixture model. As shown in formula (3-2), the Gaussian mixture model uses the maximum likelihood estimation method to estimate the value of the parameter; （3-2）； As shown in formula (3-3), since there are too many parameters in the Gaussian mixture model, the result cannot be obtained directly by derivation, so the EM algorithm is used to iteratively solve the model; (3-3)； The EM algorithm solution process is as follows: First initialize the parameters ; E-step: Calculate each data according to the current parameters From submodel possibility; ； M-step: Calculate the model parameters for the next iteration; ； ； ； Repeat the calculation of E-step and M-step until the algorithm converges, and solve the mean vector of the mixed Gaussian model through the EM algorithm , variance direction , and the weight coefficient vector , then the mean of the entire sample can be derived from formula (3-4); （3-4）； According to the interval distribution estimation of the Gaussian probability distribution function, taking 95% confidence level, the upper limit vector of the sample data is obtained through formula (3-5) and formula (3-6): With the lower limit vector ; （3-5）； （3-6）； The upper limit of the entire sample can be obtained by using formula (3-7) and formula (3-8) With lower limit ; （3-7）； （3-8）； Get the dynamic installation trend threshold interval [ , ].

6. An offshore wind power cable temperature trend warning system, characterized in that: Contains historical data processing module, prediction module establishment module, real-time data acquisition module, comparison module, and alarm module; The historical data processing module is used to collect historical temperature data of offshore wind farm cables under working conditions and corresponding working condition data, and preprocess the historical temperature data and the working condition data to obtain preprocessed data; The prediction module establishment module is used to establish a cable temperature multi-step prediction model based on an extreme learning machine, and use the preprocessed data to train and verify the cable temperature multi-step prediction model; The real-time data acquisition module is used to collect real-time temperature data and corresponding operating condition data of offshore wind farm cables, and transmit them to the cable temperature multi-step prediction model to perform real-time cable temperature state trend prediction to obtain real-time cable temperature state trend data; The comparison module is used to compare the cable temperature real-time status trend data with the cable temperature dynamic safety trend threshold range under the current working conditions; If the cable temperature real-time status trend data is within the cable temperature dynamic safety trend threshold range, generating feedback data that is transmitted to the real-time data acquisition module; If the cable temperature real-time status trend data is not within the cable temperature dynamic safety trend threshold interval, generating alarm data transmitted to the alarm module; The alarm module is used to generate an alarm; The method for establishing a multi-step prediction model for cable real-time temperature based on extreme learning machine is as follows: Step 21: Use the state multi-step prediction structure of the extreme learning machine to establish a multi-step prediction model; Step 22: Input the historical temperature data into the input layer of the extreme learning machine, determine the weight matrix of the hidden layer and the bias allocation random value to calculate the output hidden layer output matrix, and obtain a unique least squares minimum norm solution by solving the non-consistent equation, thereby training to obtain m prediction results for the historical temperature data; Step 23: Based on the m prediction results, a linear fit is performed between the current prediction and the previous prediction using the least square method, using a monotonic function straight line. Get real-time status trend change data of cable temperature; in, is the slope of the fitting function, which represents the trend of the data. , the data shows an upward trend. If , the data shows a downward trend; Step 24: Obtain historical operating data of the cable to verify the accuracy of the multi-step prediction model, thereby obtaining the cable temperature multi-step prediction model.

7. A chip, characterized in that: The device comprises: a processor configured to call and run a computer program from a memory, so that a device equipped with the chip executes: a method according to any one of claims 1 to 5.

Images