A corrosion status detection and evaluation system for floating offshore wind power mooring systems

Abstract

The invention discloses a corrosion state detection and evaluation system for a floating offshore wind power mooring system. The mooring system is divided into a plurality of areas, and a detection device is arranged in each area. Each detection device is used to collect data of a corrosion state reference object in the corresponding area and transmit it to a field acquisition instrument, a data transfer station and a data analysis center. The corrosion state reference object includes seawater temperature, dissolved oxygen content and seawater flow rate, sacrificial anode protection potential and protection current density. The corrosion state evaluation system performs data preprocessing on the data detected by the detection device. The artificial neural network model corresponding to each area evaluates the corrosion degree of the current area based on the preprocessed data. The influence of the partial corrosion degree on the overall corrosion state is determined by the grey correlation degree, that is, the correlation degree between the sub-area and the entire mooring system to determine the proportion of the evaluation result of the sub-area in the corrosion state evaluation of the mooring system. Thus, the corrosion state detection and evaluation of the floating offshore wind power mooring system is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore wind power generation, and mainly relates to a detection and evaluation system for the corrosion state of a mooring system of a floating offshore wind power generation foundation platform. Background Art

[0002] Offshore wind power resources are abundant, the number of hours of power generation is high, it does not occupy land and is suitable for large-scale development. The offshore wind energy resources in the deep sea are more abundant than those in the near sea, and the development prospects are broader. For deep sea offshore wind power, fixed foundation structures are no longer applicable, and floating foundation structures are generally used. Floating offshore wind power foundation platforms can be mainly divided into three types: single column (Spar), semi-submersible (Semi-submersible) and tension leg platform (TLP).

[0003] The mooring system of a single-column foundation platform usually adopts a catenary type, and the mooring cable is in a semi-tensioned and semi-relaxed state; Figure 1 The mooring system of the semi-submersible foundation platform shown is of catenary type or taut type; while the mooring system of the tension leg foundation platform usually adopts the tension tendon type (Tether), which is always in a fully tensioned state. The function of the mooring system is to provide tension to prevent the floating foundation platform from drifting out of the range of activities when it moves due to the interference of the external environment. Since the catenary type provides restoring force through the vertical gravity of the mooring cable, the material used for its anchor chain needs to be of a certain weight. The tension type and the tension tendon type rely on the elasticity of the mooring cable to provide tension, and polyester cables can also be used instead of traditional steel cables as the intermediate cables of the mooring system.

[0004] The mooring system of the floating offshore wind power foundation platform has been in service in the harsh marine environment for a long time, and is constantly subjected to complex environmental loads and various types of damage. The corrosion of the mooring system in the marine environment will cause damage to the internal and external structures. For the corrosion status detection of the mooring system, if manual detection is used, since the detection environment is located below the ocean surface, including the bottom anchor installed on the seabed, there are problems such as practical operation difficulties, low efficiency and insufficient detection times.

[0005] Aiming at the shortcomings of the corrosion status detection technology of floating offshore wind power mooring system, the present invention proposes a corrosion status detection and evaluation device for the mooring system of floating offshore wind power foundation platform for the catenary mooring system commonly used in semi-submersible foundation platforms. Summary of the invention

[0006] The current corrosion status detection technology of the mooring system of the floating offshore wind power foundation platform requires manual detection, and manual detection requires a lot of manpower and material resources, and there are problems such as the complex process and uncertainty of modeling the corrosion status mechanism of the mooring system; therefore, the present invention proposes a corrosion status detection and evaluation system for a floating offshore wind power mooring system. On the basis of regional detection and evaluation, a machine learning method is used to perform corrosion status detection and evaluation of the mooring system of the floating offshore wind power foundation platform. The device can accurately predict the corrosion status and changes of each area of ​​the mooring system of the floating offshore wind power foundation platform through the combination of a hardware detection system and a software evaluation system.

[0007] The technical solution adopted by the present invention is as follows:

[0008] A corrosion state detection and evaluation system for a floating offshore wind power mooring system, comprising a corrosion state detection system and a corrosion state evaluation system;

[0009] The corrosion state detection system includes a plurality of detection devices; the mooring system is divided into a plurality of areas, and a detection device is arranged in each area; each detection device is sequentially connected to a field acquisition instrument, a data transfer station and a data analysis center through signals; each of the detection devices is used to collect data of a corrosion state reference object in a corresponding area; the corrosion state reference object includes seawater temperature, dissolved oxygen content and seawater flow rate, sacrificial anode protection potential and protection current density;

[0010] The corrosion state evaluation system corresponds to the detection device of each area, and performs data preprocessing on the data detected by the detection device; based on the preprocessed data, each area extracts characteristic data through the self-learning characteristics of the artificial neural network model, and evaluates the corresponding corrosion state of each area through nonlinear mapping; the artificial neural network model of each area includes an input layer, a hidden layer and an output layer; the data of the corrosion state reference object of the current area is used as the input factor of the input layer; the output layer outputs the corrosion degree of the current area, and the corrosion degrees are good, slight corrosion and severe corrosion respectively;

[0011] The corrosion status of the entire mooring system is comprehensively evaluated by setting the weight of each area's influence on the mooring system. The evaluation index of each area is quantified according to the corrosion degree: good 0, slight corrosion 50%, severe corrosion 100%; and the influence of the partial corrosion degree on the overall corrosion status is determined by the gray correlation degree, that is, the correlation degree between the sub-area and the entire mooring system to determine the proportion of the evaluation results of the sub-area in the corrosion status evaluation of the mooring system ω c , the corrosion status assessment results of the entire mooring system are obtained through weighted processing:

[0012]

[0013] Where P is the quantitative assessment result of the corrosion status of the entire mooring system, p k is the quantitative assessment result of the corrosion status of each area, k is taken as 1 to 4, corresponding to the four divided areas, and the proportion weight of the four areas in the entire mooring system is ω ck The sum is 1. The final quantitative evaluation value of the entire mooring system is 0-30% as good, 31-60% as slight corrosion, and 61-100% as severe corrosion.

[0014] Furthermore, the mooring system is a catenary mooring system, which is divided into: area 1 where the fairlead hole and the mooring point are connected, area 2 where the top anchor chain and steel cable are located, area 3 where the bottom anchor chain is located, and area 4 where the anchor is located.

[0015] Furthermore, the detection device is composed of an information acquisition unit, a signal amplification circuit and an A / D conversion module. The information acquisition unit is fixed on the mooring system; the signal amplification circuit is used to amplify the signal collected by the sensor, and the A / D conversion module converts the analog signal into a digital signal and transmits it to the field acquisition instrument on the basic platform; in each detection device, multiple information acquisition units of the same type are arranged.

[0016] Furthermore, the information collection unit in the detection device includes: using a temperature sensor to measure the seawater temperature; using a galvanic cell type dissolved oxygen sensor to measure the dissolved oxygen content; using a seawater flow rate and flow direction monitoring device to measure the seawater flow rate; using a vibrating capacitive potential probe to measure the protection potential; using a current loop method to measure the protection current density;

[0017] Furthermore, the data transfer station collects all data information of each wind turbine mooring system in the wind farm. The corrosion status data of multiple wind turbine mooring systems are gathered together at the booster station through the optical fiber laid in the submarine cable, and stored and further transmitted by the data transfer station of the booster station.

[0018] Furthermore, data preprocessing includes checking the validity of data, handling missing data, batch estimation data fusion, and data normalization.

[0019] Furthermore, the method for checking the effectiveness is as follows: the value range of the corrosion state reference object is set respectively, and the data beyond the value range is discarded. The value range of the seawater temperature is -2 to 30°C, the value range of the dissolved oxygen content is 4 to 15 mg / L, the value range of the seawater flow rate is 0.1 to 3 m / s, the value range of the sacrificial anode protection potential is -2.4 to +1.7 V; the value range of the protection current density is 0.01 to 0.25 A / m 2 .

[0020] Furthermore, for missing values ​​in the measurement data, the average value of the measurement data is used to interpolate the missing values.

[0021] Furthermore, the data fusion is estimated in batches for data from different regions. The process is as follows:

[0022] ① The information collection units in the same area are divided into front and back groups according to their spatial positions. Taking the temperature sensor as an example, the measurement data is divided into two groups: T 11 , T 12 ,…,T 1m and T 21 , T 22 ,…,T 2j , calculate the average value of the two groups respectively and standard deviation σ 1 , σ 2 :

[0023]

[0024]

[0025] Among them, T 1i is the i-th sensor in the first group in a certain area, m is the number of sensors in the first group in a certain area; T 2i is the i-th sensor in the second group in a certain area, j is the number of sensors in the second group in a certain area;

[0026] ② Estimate data fusion results and variance in batches:

[0027]

[0028]

[0029] Among them, T' is the batch estimation data fusion result, σ' is the batch estimation data fusion variance;

[0030] The above processing is performed on the information collection units of dissolved oxygen content, seawater flow rate, protection potential and protection current density respectively, and the fusion processing of these four types of data is performed;

[0031] After data fusion processing, several groups of data measured by multiple sensors of the same type in the divided area are fused into one group.

[0032] Furthermore, the data of the corrosion state reference object are normalized respectively.

[0033]

[0034] Among them, x represents the maximum value of the data to be normalized, x min and x max They represent the minimum and maximum values ​​of the data before normalization, and y represents the data after normalization.

[0035] Beneficial effects of the present invention:

[0036] 1. The floating offshore wind power foundation platform mooring system corrosion state detection and evaluation system involved in this application is based on the selected catenary mooring system foundation platform structure and corrosion mechanism, selects the corrosion-prone area, and installs the detection device. It can accurately detect the value of the corrosion state reference object in real time, and provide a numerical basis for the mooring system corrosion state evaluation system.

[0037] 2. The corrosion status detection and evaluation system of the mooring system of the floating offshore wind power foundation platform involved in this application includes a data remote transmission system, which uses a remote communication network to ensure the transmission of the corrosion status data of the mooring system of a single wind turbine from the offshore wind turbine to the onshore centralized control center. This allows the data analysis center to accurately collect the corrosion status data of the detection area, timely determine the degree of corrosion of the mooring system, and reduce the losses caused by the corrosion of the mooring system.

[0038] 3. The corrosion state detection and evaluation device for the mooring system of the floating offshore wind power foundation platform involved in this application is based on a data-driven model to fully explore the application value of big data. Checking the validity of the data, processing missing data, batch estimation data fusion and data normalization preprocessing can improve the accuracy of data mining, shorten the extraction time of the driving model data features, and provide technical guarantee for the accuracy and rapidity of the corrosion state evaluation of the floating offshore wind power mooring system.

[0039] 4. The corrosion status detection and evaluation device for the mooring system of the floating offshore wind power foundation platform involved in this application adopts an artificial neural network algorithm to measure the corrosion status reference object, detect and evaluate the corrosion degree of the mooring system in different regions, and judge whether the mooring system is in a fault. The solution is simple and reliable, and the effect is accurate and direct.

[0040] 5. The floating offshore wind power foundation platform mooring system corrosion status detection and evaluation system involved in this application avoids manual detection, reduces the offshore wind power operation and maintenance costs, and evaluates the corrosion health status of the mooring system, which also helps floating offshore wind power to reduce maintenance time and avoid losses caused by shutdowns. It can be widely promoted in the field of floating offshore wind power corrosion detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a schematic diagram of the structure of a floating offshore wind power semi-submersible foundation platform;

[0042] Figure 2 This is a schematic diagram of the mooring system of a semi-submersible foundation platform;

[0043] Figure 3 Schematic diagram of sensor arrangement for mooring system;

[0044] Figure 4This is the data collection and transmission flow chart of the mooring system corrosion status detection device;

[0045] Figure 5 It is a schematic diagram of the artificial neural network model;

[0046] Figure 6 Schematic diagram of the mooring system corrosion status detection and assessment process. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] The present invention proposes a corrosion state detection and evaluation system for a mooring system of a floating offshore wind power foundation platform, including a corrosion state detection system and a corrosion state evaluation system. The corrosion state detection system and the corrosion state evaluation system are introduced below respectively:

[0049] 1. Corrosion status detection system

[0050] The corrosion status detection system is composed of multiple detection devices, which are arranged at key parts of the mooring system. Due to the long length of the mooring system, the data information of a "point" cannot explain the corrosion status of the entire "surface", so it is necessary to consider regional detection. According to the structure of the mooring system, it is divided into four areas, and the detection data information of these areas is obtained by the detection device. The detection data information includes external environmental factors and internal operating factors, and the detection data information measured by the detection device is transmitted to the onshore centralized control center through optical fiber for unified processing.

[0051] 1.1. Introduction to the structure of catenary mooring system

[0052] Taking the catenary mooring system of a semi-submersible foundation platform as an example, the catenary mooring system is composed of a cable guide hole, a mooring cable and a seabed anchor foundation. Figure 2 , 3 As shown. The fairlead is the key structure for connecting the mooring system to the platform. The pile foundation or suction pile anchor fixed on the seabed is the key connection between the mooring system and the seabed. The floating offshore wind power foundation platform is generally fixed by three mooring cables, each of which is 120° apart. The entire mooring cable is composed of a top anchor chain, a middle steel cable, and a bottom anchor chain connected in sequence.

[0053] The fairlead is usually welded at the bottom of the water pressure plate of the foundation platform, and the contact surface with the anchor chain is arc-shaped to avoid cutting the anchor chain and reduce the relative wear between the platform and the steel chain when the platform sways with the waves. After the foundation platform moves, the tension generated by the stretching of the mooring cable provides the platform with restoring force. Therefore, under the action of the tension load, there is a large pressure on the contact surface of the fairlead with the top anchor chain.

[0054] The traditional steel mooring cable combination (anchor chain-steel cable-anchor chain) that is widely used at present is usually suitable for floating foundation platforms with a water depth of less than 1,000 meters. The top and bottom anchor chains of the mooring system are both made of steel chains, and the middle steel cable is made of a certain number of steel wire strands wrapped around an axis in a certain direction. Inside the steel cable, the gaps between the steel wires are filled with zinc to prevent seawater from entering and causing corrosion.

[0055] The seabed anchor foundation of a semi-submersible foundation platform is usually a suction pile anchor or a suction embedded plate anchor. Suction pile anchors usually use a longer steel column structure with an open lower part and a closed upper part. They are shallower than ordinary pile anchors, so there is no need to drill deep holes in the seabed. Suction embedded plate anchors have a large vertical support capacity. The anchor claws of the plate anchor are buried vertically. In the process of pulling the mooring cable for positioning, the anchor claws are fully rotated to reach the final anchoring position.

[0056] 1.2. Selection of corrosion state parameters:

[0057] According to the structure of the catenary mooring system of the semi-submersible foundation platform, the corrosion mechanism of the mooring system is considered according to the external environment and internal operation of the mooring system. Based on the two major factors affecting the corrosion state of the mooring system, namely the external environmental factors and the internal operating factors, the corrosion state of the catenary mooring system of the semi-submersible foundation platform is the result of the superposition of the external environmental factors and the internal operating factors. The external environmental factors include seawater temperature, dissolved oxygen, seawater flow rate, salinity and various biological factors; while the internal operating factors include protection potential and protection current density. In this embodiment, seawater temperature, dissolved oxygen and seawater flow rate are selected as reference objects for evaluating the corrosion state of the catenary mooring system of the semi-submersible foundation platform in the external environmental factors; the sacrificial anode protection potential and protection current density are selected as reference objects for the internal operating factors, but are not limited to the above five reference objects.

[0058] In considering the influence of external environmental factors on the corrosion rate of the mooring system, according to the corrosion mechanism of the mooring system, the uncertainty method was used to determine that seawater temperature, dissolved oxygen content and seawater flow rate are the most important factors affecting the corrosion rate of the mooring system.

[0059] Mooring systems can provide corrosion protection through cathodic protection of sacrificial anodes. To evaluate the corrosion protection effect of semi-submersible foundation platforms made of metal, the protection potential of the metal can be measured by a reference electrode. In order to achieve the required protection potential, it is generally achieved by changing the protection current density. Therefore, the protection potential and protection current density are the two internal operating parameters that affect the corrosion rate.

[0060] Therefore, among the above parameters, the parameters that have a greater impact on the corrosion degree of the catenary mooring system are seawater temperature T (℃), dissolved oxygen DO (mg / L), seawater flow rate V (m / s), protection potential E (V) and protection current density J (A / m 2 ).

[0061] 1.3. Selection of detection device

[0062] Taking into account the detection environment and the subsequent maintenance costs of the sensors, as well as the external environmental factors and internal operating factors selected in this application, the sensors and measurement methods used to collect the above parameter information are as follows:

[0063] (1) Measuring seawater temperature: Use a thermistor. The thermistor measures temperature based on the thermal effect of resistance, that is, the resistance of the thermistor changes with temperature. As long as the resistance of the thermistor is measured, the temperature can be measured.

[0064] (2) Measuring dissolved oxygen: Use a galvanic cell type dissolved oxygen sensor. Oxygen molecules pass through the film and enter the cathode to generate a reaction, which will gain electrons from the external circuit; the lead anode will output electrons to the external circuit. A signal current will flow through the circuit, and the dissolved oxygen content is proportional to the current value.

[0065] (3) Measurement of seawater flow rate: The seawater flow rate and flow direction monitoring equipment is selected, which is composed of a seawater flow rate sensor and an integrator. The ultrasonic Doppler principle is used to measure the seawater flow rate, which is compact, rugged and low-cost.

[0066] (4) Measuring protection potential: The vibrating capacitive potential probe technology is used. Mechanical vibration is used to cause the capacitance between the sensing electrode and the measured surface to change, generating an induced current. The protection potential of the measured metal surface is obtained by calculating the induced current.

[0067] (5) Measuring protection current density: The current loop method is used to measure the current in the tube using electromagnetic mutual inductance and the Hall effect principle. When current flows through the metal being measured, there will be alternating magnetic flux in the external magnetic field of the current loop installed on the mooring cable, generating an induced electromotive force. The internal measurement circuit will then generate an induced current that is proportional to the current in the metal being measured.

[0068] 1.4. Layout of detection device:

[0069] Since the catenary mooring system has a long longitudinal length and a large measurement range, the characteristic data of one point is not enough to reflect the corrosion status of the entire mooring system. In addition, different parts of the entire mooring system are located in different external environments, and the factors affecting the external environment have regional differences. Therefore, the corrosion status detection of the mooring system should be accurate to the local location. In order to quickly and accurately determine the local corrosion state of the mooring system, the layout of the detection device should consider dividing the area. Each area contains sensors for detecting all parameter types. At the same time, multiple sensors of a single type can be arranged in one area to prevent damage to some sensors. For area selection, important and vulnerable key parts are preferred; in this embodiment, the catenary mooring system is divided into the following four areas:

[0070] (1) Connection between the fairlead and the mooring point ( Figure 3 Area 1): The connection between the fairlead and the mooring point of the top anchor chain is affected by various external forces and its corrosion degree is the most serious area in the entire mooring system. Sensors should be installed at the fairlead and the top steel cable in contact with the fairlead.

[0071] (2) Top anchor chain and cable Figure 3 Area 2): The top anchor chain is easily broken due to the long-term effects of wind, waves and currents, as well as the reciprocating motion of the platform. The rust on the outside of the steel cable will also cause the steel cable to break, seriously endangering the service life of the entire mooring system. Therefore, sensors need to be installed at both locations, which can be distributed at intervals of ten meters according to the seawater depth.

[0072] (3) Bottom anchor chain Figure 3 Area 3): When the platform moves, the bottom anchor chain is dragged on the seabed, causing some wear. When the platform is in a stable state, sand and dust particles on the seabed will also be carried into the bottom anchor chain by the ocean current, causing wear on the local structure of the steel chain. The surface is hit by mud and rocks on the seabed, leaving tiny holes, which are easy to cause corrosion.

[0073] (4) Anchorage Figure 3 Area 4): The anchor foundation is a key part for fixing the semi-submersible foundation platform. It is immersed in sea mud all year round. After the platform moves, the tension provided by the pile anchor controls the platform within a certain range. The contact part with the bottom anchor chain is subject to a large load force. Detecting the corrosion parameters at this location is of great significance for corrosion status assessment. Since the anchor foundation water level is deep and the seawater pressure is high, the sensors arranged at this location need to have high pressure resistance.

[0074] 1.5. Data collection and transmission of detection devices

[0075] The detection devices (sensors) arranged on the mooring system are used to measure the parameter information in each area. The sensors transmit the measured data information to the field data collector on the basic platform through the communication bus. The data information of each wind turbine mooring system is then gathered to the data transfer station of the offshore booster station of the wind farm through optical fiber, and finally transmitted to the data analysis center of the onshore centralized control center through the optical fiber in the submarine cable, completing the data collection, transmission and monitoring process. The parameter data collection and transmission process measured by the sensor can be referred to Figure 4 .

[0076] (1) Data acquisition system:

[0077] The corrosion status data acquisition system of the floating offshore wind power foundation platform mooring system consists of sensors, signal amplification circuits and A / D conversion modules. The sensor is fixed to the mooring cable with bolts. Since the amplitude of the signal collected by the sensor is small and cannot match the subsequent circuit, the signal needs to be amplified by the signal amplification circuit, and then converted into a digital signal through the A / D conversion module and transmitted to the field acquisition instrument on the foundation platform for storage. right Collection of corrosion status data of wind turbine mooring systems.

[0078] (2) Remote transmission system:

[0079] It includes on-site collectors, data transfer stations and data analysis centers.

[0080] ① On-site collection instrument:

[0081] The field data collector is installed on the floating wind power foundation platform, and mainly plays the role of storing and transmitting the data measured by the data acquisition system. The measurement sensors of the required parameters in the present invention are arranged below the water surface. The signals collected by the sensors undergo a series of conversions and are connected to the field data collector through a communication bus, so as to realize the transmission of the corrosion status data of the mooring system of a single wind turbine from the mooring system to the foundation platform on the offshore wind turbine.

[0082] ②Data transfer station:

[0083] All data information of the mooring systems of wind turbines in a wind farm should be collected at the offshore booster station like electricity. The corrosion status data of the mooring systems of multiple wind turbines are collected at the booster station through the optical fiber laid in the submarine cable, and stored and further transmitted by the data transfer station of the booster station.

[0084] ③Data Analysis Center:

[0085] Since the offshore booster station is unmanned, the information stored in the booster station must be transmitted through the optical fiber in the submarine cable to the data analysis center of the onshore centralized control center for unified processing. The data analysis center of the onshore centralized control center can send monitoring instructions, receive telemetry data and perform data analysis. The staff can monitor the corrosion status parameters of each unit's mooring system in real time through the host computer and perform data analysis and diagnosis.

[0086] 2. Corrosion status assessment system

[0087] Based on the aforementioned mooring system corrosion state detection system, the present invention further proposes a corrosion state evaluation system. Corresponding to the regional corrosion state detection of the mooring system, the corrosion state evaluation system is also based on the detection of different regions. The evaluation system mines the detection data information characteristics of the above-mentioned regions, and uses this to establish an accurate and convenient prediction model. According to the data transmitted to the onshore centralized control center, a corrosion database for each region of the mooring system of the floating offshore wind power foundation platform is established, and the data is checked for validity, data missing is processed, batch estimation data is fused, and data is normalized. The trained artificial neural network model is used, combined with the characteristics of deep learning of artificial neural networks, to accurately evaluate the corrosion state of each region of the mooring system of the floating offshore wind power foundation platform. Finally, based on the idea of ​​gray correlation, the weight influence of each region on the whole is calculated, and the corrosion state of the mooring system as a whole is quantitatively evaluated according to the corrosion state of some regions. The specific implementation steps are as follows:

[0088] Machine learning theory in the field of artificial intelligence is applicable to application scenarios that require finding general patterns from observed data and predicting unknown data based on such patterns.

[0089] Among various machine learning algorithms, artificial neural networks are widely used. Among them, multi-layer feedforward neural networks, namely BP neural networks (Back Propagation Neural Networks), have powerful nonlinear mapping capabilities and flexible network structures, and are easy to implement from input mapping to output, without considering the complex physical model correspondence between various parameters and corrosion states, and can provide accurate prediction models. Therefore, the present invention uses BP neural networks to construct corrosion state assessment models for various regions of the mooring system of floating offshore wind power foundation platforms.

[0090] 2.1. BP neural network algorithm principle:

[0091] In the neural network, the activation function of the neuron adopts the Sigmod function

[0092]

[0093] Among them, z is the output of the previous layer network and also the input of this layer network.

[0094] The propagation between neurons is connected through weights and biases to obtain the activation value of the next layer of the network:

[0095]

[0096] in, is the weight from the i-th neuron in the l-1th layer to the j-th neuron in the lth layer, is the bias of the jth neuron in the lth layer, is the activation value of the jth neuron in the lth layer, and n is the number of nodes in the previous layer of the neural network acting on this node.

[0097] The measured regional sample data is input into the corresponding neural network for training, and the gradient descent method is used to adjust and converge the weights and biases layer by layer to ensure that the expected output of the trained neural network model is within the set error range. Here, the training error of the neural network is set to 1e-5. The error of the prediction result is expressed by the least squares method as:

[0098]

[0099] Among them, E l is the output error, is the training output of the network, is the data in the training sample, that is, the expected value of the neural network, and k is the number of nodes in the network output layer. The essence of this algorithm is to adjust the parameter weight and bias value according to the output error, and gradually reduce E l , so that the predicted value is close to the true value and a prediction model with better accuracy is obtained.

[0100] In the corrosion process of the mooring system of the floating offshore wind power foundation platform, the fluctuations of external environmental factors and the changes in internal operating parameters have a traceable influence on the degree of corrosion. Therefore, the self-learning characteristics of artificial neural networks can be used to extract feature data, and through nonlinear mapping, that is, finding the mapping relationship f from the input space to the output space, the prediction results are output by solving the transfer function f(x), and finally the corrosion state of the mooring system is evaluated.

[0101] f(x)=ωx+b (4)

[0102] Among them, ω is the weight, x is the neural node input, and b is the bias.

[0103] Given a data sample set S:

[0104] S={(x i ,y i )|x i ∈R l ,yi ∈R}, i=1,2...,n (5)

[0105] Among them, x i is the i-th input sample, y i For x i The corresponding output result, R is the real number space, R l is an l-dimensional real number space, and n is the number of samples.

[0106] 2.2 Data preprocessing:

[0107] (1) Validity check

[0108] During model training, the data needs to be checked for validity and outliers need to be removed. The present invention uses a data range to verify data validity. Taking seawater temperature as an example, seawater temperature data is generally in the range of -2 to 30°C. If the measured seawater temperature reaches an extremely high level of more than 30°C, it is inconsistent with the actual situation and the data should be avoided from being entered. Similarly, the value ranges of several other parameters are as follows: dissolved oxygen content is 4 to 15 mg / L; seawater flow rate is 0.1 to 3 m / s; protection potential is -2.4 to +1.7 V; protection current density is 0.01 to 0.25 A / m 2 ; Use the range of the above parameters to remove outliers. The selected data range should not be too small, otherwise a large amount of normal data will be eliminated, which will lead to more limitations of the trained model and greatly reduce the recognition ability.

[0109] (2) Data missing processing

[0110] When establishing a database, data loss is likely to occur due to failure to record data in time or data corruption. The corrosion state prediction model is trained based on the missing data, and the prediction accuracy of the obtained model is low. If these unknown or missing data are not classified as faults but as normal, the neural network will try to adapt to these data, which will increase the generalization error of the network. The missing values ​​in the measurement data are divided into numerical and non-numerical types. In the present invention, the missing data in the database is numerical, and the missing values ​​are interpolated according to the average value of the parameter record value.

[0111] (3) Batch Estimation Data Fusion

[0112] Since the sensors used in the present invention to detect the corrosion status of the mooring system are divided into regions for information detection, there are multiple adjacent sensors of the same type in different regions, and the measured data is numerous and not suitable for direct input into the neural network model. The aforementioned data preprocessing step takes a single sensor as the research object. If one of the sensors is damaged, the obtained data will have an adverse effect on the final data analysis regardless of whether the error is eliminated. If the model is to have good anti-interference performance, it is necessary to perform data fusion on the data measured by multiple single-type sensors in the same area under the premise of batch estimation:

[0113] ① Group the sensors in the same area by space. Taking the temperature sensor as an example, the measurement data is divided into two groups: T 11 , T 12 ,…,T 1m and T 21 , T 22 ,…,T 2j , calculate the average value of the two groups respectively and standard deviation σ 1 , σ 2 :

[0114]

[0115]

[0116] Among them, T 1i is the i-th sensor in the first group in a certain area, m is the number of sensors in the first group in a certain area; T 2i is the i-th sensor in the second group in a certain area, and j is the number of sensors in the second group in a certain area.

[0117] ② Estimate data fusion results and variance in batches:

[0118]

[0119]

[0120] Among them, T' is the batch estimation data fusion result, and σ' is the batch estimation data fusion variance.

[0121] After data fusion processing, several groups of data measured by multiple sensors of the same type in the divided area are fused into one group, so as to obtain a consistent interpretation and description of the parameters, so that the output data of the mooring system corrosion status detection system, that is, the input data of the neural network model, has better robustness.

[0122] (4) Data normalization

[0123] Since the units of the selected parameters (such as seawater temperature: ℃; seawater flow rate: m / s) and the generally applicable range of the measured values ​​(seawater temperature: -2 to 30; seawater flow rate: 0.1 to 3) are different, there is no unified evaluation standard. In order to make the network converge quickly and improve its generalization ability, it is necessary to dimensionalize it and normalize all the data:

[0124]

[0125] Formula (10) is used to normalize the data of the five types of parameters to the interval [-1, 1], so that the sensitivity of the BP neural network algorithm to the size change of the selected parameter data is close. In the formula, x represents the maximum value of the data to be normalized, x min and x max They represent the minimum and maximum values ​​of the data before normalization, and y represents the data after normalization.

[0126] 2.3. Construct BP neural network prediction model:

[0127] The present invention evaluates the corrosion status of the mooring system of the basic platform. It is necessary to establish a neural network model, which is applied to the four areas where the sensors are arranged in the above steps. The BP neural network algorithm is used. Each model contains 3 layers of neural networks. The specific neural network model can be referred to Figure 5 The parameter data of seawater temperature, dissolved oxygen content, seawater flow rate, protection potential and protection current density in different regions measured by the mooring system corrosion status detection system in the previous steps are used as the five input factors of the neural network input layer in each region, corresponding to Figure 5 Medium / 1 k 、I2 k 、I3 k 、I4 k 、I5 k As the input node, the subscript k is used to distinguish four regions, k∈{1, 2, 3, 4}. k To represent the set of input parameters, S k T k ,DO k 、V k 、E k , J k is the characteristic parameter of each region, corresponding to x in equation (5) i As input, the dataset can be represented as:

[0128] S k ={T k , DO k , V k , E k , J k} T (11)

[0129] Although increasing the number of hidden layers can further reduce the output error and improve accuracy, it will complicate the network, greatly increase the number of iterations and training time. Therefore, this model uses a simple 3-layer neural network model, and the number of hidden layers is set to 1. The number of nodes in the hidden layer depends on many factors, and its value is often determined by multiple experiments and experience. There are also relevant formulas for reference:

[0130]

[0131] n is the optimal number of hidden layer units, i is the number of input layer nodes, m is the number of output layer nodes, and a is generally a constant with a value range of 1 to 10. Regarding the selection of the number of hidden layers, the number of nodes can also be adjusted according to the error of the final output result. Here, the number of hidden layer nodes is selected as 6, H1 k ~H6 k As the hidden layer node of the model.

[0132] Y1 k 、Y2 k 、Y3 k As the output layer nodes of the model, they correspond to the three corrosion states of the output. The final output is three corrosion states (good, slightly corroded, and severely corroded). In order to facilitate the diagnosis of the final output results, these three states are independently encoded. The good state is set to (100 k ), slight corrosion is set to (010 k ), severe corrosion is set to (001 k ), these codes are consistent with y in formula (5) i The corresponding output result of the model. The data input into the database should include training data and test data. The data format can be:

[0133] 100 k :[T 1k , DO 1k , V 1k , E 1k , J 1k ] T

[0134] …

[0135] [T nk , DO nk , V nk , E nk , J nk ] T ;

[0136] 010 k :[T 1k , DO 1k, V 1k , E 1k , J 1k ] T

[0137] …

[0138] [T nk , DO nk , V nk , E nk , J nk ] T ;

[0139] 001 k :[T1 k , DO 1k , V 1k , E 1k , J 1k ] T

[0140] …

[0141] [T nk , DO nk , V nk , E nk , J nk ] T ; (13)

[0142] The number of rows represents the number of data sets, each of which has n sets put into the neural network model for training. The number of columns is seawater temperature, dissolved oxygen content, seawater flow rate, protection potential and protection current density.

[0143] After systematic learning, the neural network obtains the final output value, i.e., the degree of corrosion, and an output threshold can be set. If one of the output results reaches a certain threshold, it will be used as the final output result to evaluate the health status of each area of ​​the mooring system after being corroded by seawater. Subsequently, the degree of failure will be used to determine whether a certain part of the mooring system still meets the standard for normal operation and whether it will affect the operation of the entire mooring system.

[0144] 2.4. Mooring system corrosion status assessment device:

[0145] The mooring system corrosion status assessment device proposed in this invention is designed based on the BP neural network algorithm on the basis of the corrosion status detection of the mooring system of the floating offshore wind power foundation platform. The specific design process can be referred to Figure 6 . It realizes functions such as data processing, error analysis, model structure adjustment and fault prediction.

[0146] According to the relevant data measured by the above-mentioned sensors at the locations of different areas, a neural network corrosion status assessment system is constructed for each area, so as to achieve the purpose of corrosion status assessment of key areas of the mooring system. The neural network model structure established in the four areas is the same, but the weights and thresholds of the network after training with the data collected in each area, that is, the mapping relationship from input to output is different. The corrosion status assessment results of the four areas are given by the neural network model trained with the historical data of each area. The significance of dividing the area is to be able to more accurately diagnose the degree of corrosion failure at a specific location.

[0147] The corrosion degree of a single area can indirectly indicate the overall corrosion state. The corrosion state of the entire mooring system is comprehensively evaluated by setting the influence weights of the four areas on the mooring system. The evaluation index of each area is quantified according to the corrosion degree level: good - 0, slight corrosion - 50%, severe corrosion - 100%. The influence of the partial corrosion degree on the overall corrosion state can be determined by the gray correlation degree, that is, the correlation degree between the sub-area and the entire mooring system to determine the proportion of the evaluation results of the sub-area in the corrosion state evaluation of the mooring system ω c , the corrosion status assessment results of the entire mooring system are obtained through weighted processing:

[0148]

[0149] Where P is the quantitative assessment result of the corrosion status of the entire mooring system, p k is the quantitative assessment result of the corrosion status of each area, k is taken as 1 to 4, corresponding to the four divided areas, and the proportion weight of the four areas in the entire mooring system is ω ck The sum is 1. The final quantitative evaluation value of the entire mooring system can be defined as 0-30% as good, 31-60% as slight corrosion, and 61-100% as severe corrosion.

[0150] The above embodiments are only used to illustrate the design ideas and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design ideas disclosed by the present invention are within the protection scope of the present invention.

Claims

1. A corrosion status detection and evaluation system for a floating offshore wind power mooring system, characterized in that: Including corrosion state detection system and corrosion state assessment system; The corrosion state detection system includes a plurality of detection devices; the mooring system is divided into a plurality of areas, and a detection device is arranged in each area; each detection device is sequentially connected to a field acquisition instrument, a data transfer station and a data analysis center through signals; each of the detection devices is used to collect data of a corrosion state reference object in a corresponding area; the corrosion state reference object includes seawater temperature, dissolved oxygen content and seawater flow rate, sacrificial anode protection potential and protection current density; The corrosion state evaluation system corresponds to the detection device of each area, and performs data preprocessing on the data detected by the detection device; based on the preprocessed data, each area extracts characteristic data through the self-learning characteristics of the artificial neural network model, and evaluates the corresponding corrosion state of each area through nonlinear mapping; the artificial neural network model of each area includes an input layer, a hidden layer and an output layer; the data of the corrosion state reference object of the current area is used as the input factor of the input layer; the output layer outputs the corrosion degree of the current area, and the corrosion degrees are good, slight corrosion and severe corrosion respectively; The corrosion status of the entire mooring system is comprehensively evaluated by setting the weight of each area's influence on the mooring system. The evaluation index of each area is quantified according to the corrosion degree: good 0, slight corrosion 50%, severe corrosion 100%; and the influence of the partial corrosion degree on the overall corrosion status is determined by the gray correlation degree, that is, the correlation degree between the sub-area and the entire mooring system to determine the proportion of the evaluation results of the sub-area in the corrosion status evaluation of the mooring system ω c , the corrosion status assessment results of the entire mooring system are obtained through weighted processing: Where P is the quantitative assessment result of the corrosion status of the entire mooring system, p k is the quantitative assessment result of the corrosion status of each area, k is taken as 1 to 4, corresponding to the four divided areas, and the proportion weight of the four areas in the entire mooring system is ω ck The sum of the values ​​is 1; the final quantitative evaluation value of the entire mooring system is 0-30% as good, 31-60% as slight corrosion, and 61-100% as severe corrosion; The data preprocessing includes checking the validity of the data, processing missing data, batch estimation data fusion and data normalization; The validity check method is as follows: the value range of the corrosion state reference object is set respectively, and the data beyond the value range is discarded. The value range of the seawater temperature is -2 to 30°C, the value range of the dissolved oxygen content is 4 to 15 mg / L, the value range of the seawater flow rate is 0.1 to 3 m / s, the value range of the sacrificial anode protection potential is -2.4 to +1.7 V; the value range of the protection current density is 0.01 to 0.25 A / m 2 ; For missing values ​​in the measurement data, the mean value of the measurement data is used to interpolate the missing values; For data from different regions, data fusion is estimated in batches. The process is as follows: ① The information collection units in the same area are divided into front and back groups according to their spatial positions. Taking the temperature sensor as an example, the measurement data is divided into two groups: T 11 , T 12 ,…,T 1m and T 21 , T 22 ,…,T 2j , calculate the average value of the two groups respectively And standard deviation σ1, σ2: Among them, T 1i is the i-th sensor in the first group in a certain area, m is the number of sensors in the first group in a certain area; T 2i is the i-th sensor in the second group in a certain area, j is the number of sensors in the second group in a certain area; ② Estimate data fusion results and variance in batches: Among them, T' is the batch estimation data fusion result, σ' is the batch estimation data fusion variance; The above processing is performed on the information collection units of dissolved oxygen content, seawater flow rate, protection potential and protection current density respectively, and the fusion processing of these four types of data is performed; After data fusion processing, several groups of data measured by multiple sensors of the same type in the divided area are fused into one group; The data of the corrosion state reference object are normalized respectively. Among them, x represents the maximum value of the data to be normalized, x min and x max They represent the minimum and maximum values ​​of the data before normalization, and y represents the data after normalization.

2. A floating offshore wind power mooring system corrosion status detection and evaluation system according to claim 1, characterized in that: The mooring system is a catenary mooring system, which is divided into: area 1 where the fairlead hole and the mooring point are connected, area 2 where the top anchor chain and steel cable are located, area 3 where the bottom anchor chain is located, and area 4 where the anchor is located.

3. A corrosion status detection and evaluation system for a floating offshore wind power mooring system according to claim 1, characterized in that: The detection device is composed of an information acquisition unit, a signal amplification circuit and an A / D conversion module. The information acquisition unit is fixed on the mooring system; the signal amplification circuit is used to amplify the signal collected by the sensor, and the A / D conversion module converts the analog signal into a digital signal and transmits it to the field acquisition instrument on the basic platform; in each detection device, multiple information acquisition units of the same type are arranged.

4. A floating offshore wind power mooring system corrosion status detection and evaluation system according to claim 1 or 3, characterized in that: The information acquisition unit in the detection device includes: using a temperature sensor to measure the seawater temperature; using a galvanic cell type dissolved oxygen sensor to measure the dissolved oxygen content; using a seawater flow rate and flow direction monitoring device to measure the seawater flow rate; using a vibrating capacitive potential probe to measure the protection potential; and using the current loop method to measure the protection current density.

5. A corrosion status detection and evaluation system for a floating offshore wind power mooring system according to claim 1, characterized in that: The data transfer station collects all data information of each wind turbine mooring system in the wind farm. The corrosion status data of multiple wind turbine mooring systems are gathered together at the booster station through the optical fiber laid in the submarine cable, and stored and further transmitted by the data transfer station of the booster station.

Images